Discovering Crowdsourcing Theory, Classification and
Transcript of Discovering Crowdsourcing Theory, Classification and
Eindhoven, Februa y 2009 r
Discovering Crowdsourcing
Theory, Classification and Directions for use
Simone A.M. Geerts
BSc Industrial E TUE 2006 ngineering and Management Science —
Student identity number 0550626
in partial fulfillment of the requirements for the degree of
Master of Science
in Innovation Management
Prof.dr.ir. M.C.D.P. Weggeman
dr. J.M.P. Gevers, TU/e, HPM
Supervisors:
, TU/e, OSM
i
ii
TUE. Department Industrial Engineering and In
eries Master Theses Innovation Management
novation Sciences.
S
ubject headings: innovation strategy, internet S
iii
ACKNOWLEDGEMENTS
This master thesis project is the final assignment in the MSc program Innovation Management at the Eindhoven University of Technology. The project has been executed in cooperation with Altuition, a consulting firm located in Den Bosch.
This report symbolizes the end of my time as a student. The choice for Industrial Engineering and Management Science has proved to be right from the start. By seizing the opportunities this study has provided me, my time in Eindhoven has been a wonderful experience. It has made me confident about myself and my capabilities.
One of these opportunities was this master thesis project. When I started this project, I had – like most people – never heard of crowdsourcing. This project has given me the opportunity to discover and study this interesting topic, which I believe has the potential to impact the world in important ways.
Although this project has been mostly an individual assignment, the completion of this project would not have been possible without the help of several people. Here, I would like to take the opportunity to thank all of them for their effort and support.
First, I would like to thank my supervisor, Mathieu Weggeman, for providing me the opportunity to study this topic, for his insightful advice and feedback, and for giving me the space to find my own path. Second, I would like to thank my second supervisor, Josette Gevers, who has helped me through her very useful comments and questions to develop this report as it is now.
Furthermore, I would like to thank everyone at Altuition for giving me the opportunity to do this project. Especially, my thanks go out to my company supervisor, Marcel Weber, for his interesting thoughts on the subject. Thank you for the support, cooperation and trust in my work.
One of the most interesting parts of this study consisted of the interviews with crowdsourcing practitioners. Therefore, I would like to thank all interviewees for contributing their time to tell me their story; each of them were an inspiration to my work.
Last, but certainly not least, I would like to thank everyone else who has helped me during my study to achieve this important step in my life. Special thanks go out to my parents, boyfriend and (study) friends, who always supported and trusted me and provided me with their wise advice.
Simone Geerts
Eindhoven, February 2009
iv
SUMMARY
The topic of this study is crowdsourcing, which is defined as the online outsourcing of a task to (a group of) private individuals in the form of an open call. Crowdsourcing can be seen as an emerging set of new business models, focusing at involving the crowd in activities such as concept development, problem solving or production. Although crowdsourcing has the potential to change industries, no research has been conducted into this topic so far.
Therefore, this study provides important findings that are relevant for both theory and practice. The ultimate objective is design‐oriented, and is focused on developing a protocol that can support practitioners in the application of crowdsourcing. To be able to provide the necessary info arm tion, four research questions are formulated:
? 1. Which types of crowdsourcing exist and how can they be distinguished
2. Which type of participants are expected? 3. Which conditions must be satisfied to generate productive outcomes? 4. Which efforts are needed from the company to facilitate the crowdsourcing process?
This study consists of several phases to find answers to the research questions and develop a protocol. First of all, a literature study is conducted into fields related to crowdsourcing, including user driven innovation, open innovation, open source software development, and online communities. Relating these concepts to each other and to crowdsourcing reveals that these concepts are important to understand the crowdsourcing trend, but since none of these concepts fully include crowdsourcing, it should be seen as a separate topic that should be studied in more detail.
The second phase consists of an exploration of practice, in which field examples are explored to find relevant concepts to be included in the study, formulated as attributes. These attributes are selected based on their perceived relevance in explaining the differences between the crowdsourcing cases. Attributes are found on possible interventions that companies can decide on, differences in contexts in which these interventions are believed to be relevant, and outcomes that can be used to measure success. In total, ten intervention attributes, two context attributes and four outcome attributes are selected. The intervention attributes can be seen in the upper row of table I. The context attributes contain the NPD phase in which the crowd is involved and the field of the activity. The outcome attributes involve the size and activity of the crowd, the composition of the participants, the number and quality of the contributions, and the effort that is needed from the company to make the crowdsourcing initiative succeed.
To answer the first research question, the intervention attributes are used to develop a typology of crowdsourcing. From the analysis, it is found that the attributes follow an internal consistency, allowing the types to emerge from clustering. In table I, the types and their characteristics are shown. The first type is crowdcasting, in which a particular challenge is broadcasted to a crowd, generally organized as a competition with a financial reward. The second type is crowdstorming, which involves an online brainstorming session, where interaction between participants is important. In many cases, this involves a company asking the crowd for new product or service ideas. The third type is crowd production, in which the crowd creates a product or database together or creates a market of individual contributions. The fourth type is crowdfunding, where instead of the spare time, abilities and knowledge of the crowd, their spare money is used. The crowd is for example used to fund artists, companies or
each other. These types all represent different business models that can be used for different purposes.
Type
Role of the company
Type of participation
Party that takes
initiative
Whose contributions
are used
Financial rew
ards
Interaction
Input control
Output control
IP
Crowdcasting mediator/
own ini tiative
upload company in itiative individual yes yes/no
contributions not public; registration
company company
Crowdstorming own ini tiative forum crowd
in itiative combination no yes registration company company/ public
Crowd production
product own initiative
weblog; wiki
crowd in itiative crowd no yes/no registration crowd public
market mediator upload; forum
crowd initiative p2p yes yes/no registration crowd crowd
Crowd funding
product mediator/
own initiative
financial company initiative crowd yes no
registration; type of
participation crowd not
relevant
market medi
v
ator financial crowd initiative p2p yes no
registration; type of
participation crowd not
relevant
TABLE I: CHARACT ISTIC OF CR
After the types of crowdsourcing have been determined, research questions two, three and four remain to be answered. This is done by combining the findings from the literature study with the findings from the case studies, and developing propositions that describe relations between concepts. The relations that were found are shown in figure I. Furthermore, the activities that have been found to require the most effort from the company are technical issues, achieving and maintaining critical mass, community management, processing results, training, and management
ER S OWDSOURCING TYPES
FIGURE I: THEORETICAL PROPOSITIONS
support.
These findings are then translated into protocols for both crowdcasting and crowdstorming. The protocols describe first how to make a choice between these two types, based on the size of the target group, the social technographics profile, the importance of marketing objectives, the type of innovation and the NPD phase in which the crowd should be involved. After the definition phase, in which this choice is made, four phases follow, slightly different for each type, as can be seen in figure II. The activities for each type in each phase iffer substantially, as can be seen in figure III. d
FIGURE II: CROWDCASTING (LEFT) AND CROWDSTORMING (RIGHT) PHASES
This study contributes to both organization theory and management theory. The contributions to organization theory consist of the typology and the theoretical propositions. The typology can be considered a theoretical framework for crowdsourcing, which can be used to structure future research. The propositions that were developed provide an important basis for future research, since their relevance has been shown, although they need further testing. The typology also contributes to management theory, as it provides relevant information for practitioners about the possibilities of crowdsourcing. The contribution to management theory furthermore consists of the protocol, since it provides support to organizations that want to apply
FIGURE III: EFFORT IN CROWDCASTING (LEFT) AND CROWDSTORMING (RIGHT)
crowdsourcing for innovation purposes.
Although much information has been collected, from which many important findings could be developed, the results of this study should not be considered definitive. Because of the emerging nature and broad range of the topic, this study forms a scientific basis for the subject to mature from further studies and experiences. Since the propositions have only been subjected to initial testing, further testing is needed to validate the findings. Furthermore, more quantitative research is needed to determine the relative importance of each of the relations. Since the
vi
protocol was developed from these propositions, the protocol needs to be validated as well.
This study has shown that crowdsourcing is more than just a hype, and can be considered a new business model that is likely to be relevant for many organizations in many fields. This study addresses a gap in theoretical knowledge and gives organizations the tools to get ahead of the game, by taking their advantage of the knowledge and guidance that this study provides.
vii
INDEX
Acknowledgements _________________________________________________________________________________________ iii
Summary _____________________________________________________________________________________________________ iv
Index __________________________________________________________________________________________________________ vii
1 Introduction ____________________________________________________________________________________________ 1
1.1 Definition of crowdsourcing _____________________________________________________________ 1
1.2 Causes of the trend ________________________________________________________________________ 2
1.3 Research objectives _______________________________________________________________________ 3
2 Research design and methodology _________________________________________________________________ 5
2.1 Research approach ________________________________________________________________________ 5
2.2 Research quality ___________________________________________________________________________ 8
2.3 Structure of report _______________________________________________________________________ 10
3 Exploration of theory _______________________________________________________________________________ 11
3.1 User driven innovation _________________________________________________________________ 11
3.2 Open innovation _________________________________________________________________________ 12
3.3 Open source software development __________________________________________________ 14
3.4 Online communities _____________________________________________________________________ 15
3.5 Crowdsourcing in perspective _________________________________________________________ 16
3.6 Conclusion ________________________________________________________________________________ 17
4 Exploration of practice _____________________________________________________________________________ 19
4.1 Intervention attributes _________________________________________________________________ 19
4.2 Context attributes ________________________________________________________________________ 23
4.3 Outcome attributes ______________________________________________________________________ 24
4.4 Conclusion ________________________________________________________________________________ 26
5 Crowdsourcing types and their characteristics _______________________________________________ 27
5.1 Typology of Howe ________________________________________________________________________ 27
5.2 Analysis ____________________________________________________________________________________ 29
5.3 Crowdsourcing types ____________________________________________________________________ 31
viii
5.4 Conclusion ________________________________________________________________________________ 35
6 Building propositions for crowdsourcing ______________________________________________________ 37
6.1 Types of participants in crowdsourcing ______________________________________________ 37
6.2 Conditions for crowdsourcing request _______________________________________________ 44
6.3 Facilitating the crowdsourcing process ______________________________________________ 53
6.4 Conclusion ________________________________________________________________________________ 55
7 Towards a protocol__________________________________________________________________________________ 57
7.1 Definition phase __________________________________________________________________________ 58
7.2 Crowdcasting _____________________________________________________________________________ 60
7.3 Crowdstorming ___________________________________________________________________________ 63
7.4 Conclusion ________________________________________________________________________________ 66
8 Conclusion _____________________________________________________________________________________________ 69
References ___________________________________________________________________________________________________ 73
Appendix A __________________________________________________________________________________________________ 81
A.1 Crowdcasting intervention and context attributes _________________________________ 82
A.2 Crowdstorming intervention and context attributes ______________________________ 84
A.3 Crowd production intervention and context attributes ___________________________ 85
A.4 Crowdfunding intervention and context attributes ________________________________ 87
A.5 Hybrids intervention and context attributes ________________________________________ 88
A.6 Crowdcasting outcome attributes _____________________________________________________ 89
A.7 Crowdstorming outcome attributes __________________________________________________ 91
A.8 Crowd production outcome attributes _______________________________________________ 92
A.9 Crowdfunding outcome attributes ____________________________________________________ 94
A.10 Hybrids outcome attributes ____________________________________________________________ 95
Appendix B __________________________________________________________________________________________________ 97
B.1 Goldcorp challenge ______________________________________________________________________ 97
B.2 InnoCentive _______________________________________________________________________________ 97
B.3 Dell’s IdeaStorm __________________________________________________________________________ 98
B.4 iStockphoto _______________________________________________________________________________ 99
B.5 Nabuur ____________________________________________________________________________________ 100
ix
B.6 NS: Mijnproefstation.nl_________________________________________________________________ 104
B.7 Battle of Concepts _______________________________________________________________________ 108
B.8 Wikipedia ________________________________________________________________________________ 111
B.9 RedesignMe ______________________________________________________________________________ 113
B.10 Assignment Zero ________________________________________________________________________ 116
B.11 Threadless _______________________________________________________________________________ 117
2
1
1 INTRODUCTION
The main topic for this master thesis is crowdsourcing, which stands for outsourcing to the crowd. It can be seen as part of a larger trend towards more diversity and a more active role for consumers in the innovation process. This trend towards diversity can be found in the literature about R&D generations (Nobelius 2004; Rothwell 1994; Liyanage et al. 1999; Berkhout et al. 2006). The four generations describe a development from having large internal R&D departments that push the technology‐driven innovations into the market, to searching for innovation in other departments within the company, to consulting other organizations in the supply chain, including customers, and even competitors. Crowdsourcing is the next phase in this development, as it uses even more external sources of innovation. The increasing involvement of customers in the innovation process is also related to crowdsourcing, as in many cases the crowd consists of customers. According to Prahalad and Ramaswamy (2004), the meaning of value and the process of value creation are rapidly shifting from a product‐ and firm‐centric view to personalized consumer experiences. Informed, networked, empowered, and active consumers are increasingly co‐creating value with the firm. Consumers are increasingly organized into online communities where they not only interact with firms, but also with each other. Increasingly more companies are using these online communities to communicate with their customers (Sawhney et al. 2003). Crowdsourcing allows consumers to participate in a much more active way. Although there are other possible purposes for crowdsourcing, the main focus of the study consists of applications for innovation.
1.1 DEFINITION OF CROWDSOURCING
Crowdsourcing is an upcoming trend where tasks that are traditionally performed by employees of companies are now increasingly performed by people who use their spare time for these tasks. The name ‘crowdsourcing’ first appeared in Wired Magazine in June 2006 in an article by Jeff Howe who defines it as ‘the act of taking a job traditionally performed by a designated agent (usually an employee) and outsourcing it to an undefined, generally large group of people in the form of an open call’1. So crowdsourcing stands literally for outsourcing to a crowd.
Three aspects of this definition should be highlighted. First of all, for an activity to be crowdsourced, it should be traditionally performed by a designated agent, which represents the outsourcing part of the definition. This does however not mean that the activity should be initiated by the organization that would traditionally perform the activity. The opportunity for the crowd to get involved in the activity can also be provided by an entrepreneur providing a platform and building a business model around it, or it can be initiated by the crowd itself. The main point is that it concerns a task that is traditionally performed by employees of an organization in response to a request from their boss, while this task is now performed by someone who chooses to do this in his or her spare time. Second, the crowd should be undefined, which illustrates the difference between outsourcing to a known party such as a company or to a much wider audience of private individuals. The issue of who actually forms this crowd will be discussed extensively in this report. The final important aspect of the definition is the open call, which emphasizes the more active role of the crowd, which is
2
traditionally seen as passive and merely consuming, in that the individual members can decide for themselves if, how and when they want to participate.
In this study, most aspects of the original definition are used, but it is changed slightly because of two reasons. First of all, limiting the study to the cases where the crowd is undefined, as is stated in the original definition, would exclude some very important cases. Therefore the crowd is seen as a group of private individuals, performing the particular task in their spare time. Second, there is another characteristic that is not mentioned by Howe, namely the role of the Internet. In some ways, crowdsourcing can be considered a new form of existing concepts, instead of a new phenomenon in itself. As Howe (2008) argues, the trend is appearing now because of the use of the Internet, not because the Internet made crowdsourcing possible, but because it made it more effective. Through the Internet, much more people can be reached, resulting in the reinvention of existing tools and concepts, such as creating a virtual version of the traditional idea box, but also in new opportunities like online collaboration. Especially the possibilities of the latest Internet generation, Web 2.0, have enormous effects on the opportunities of organizations to interact with other parties because it facilitates online collaboration and sharing among users (Albors et al. 2008). Because the objective of this study is to research the crowdsourcing trend, which is enabled by the Internet, only cases that outsource online work are considered. Therefore, the definition of crowdsourcing that is used in this study is:
Crowdsourcing is the online outsourcing of a task to (a group of) private individuals in the form of an open call.
1.2 CAUSES OF THE TREND
Howe (2008) argues that four fundamental developments have caused crowdsourcing to emerge. The first one is that private individuals are given increasingly more opportunities to work on tasks that contribute to economic production, but are performed in their spare time and are not considered part of their jobs. While people have the potential to excel at multiple fields and tasks, the industrial revolution has caused people to perform ever more specialized jobs. Firms have taken up all the tasks that were once performed by individuals, families and communities. As futurist Alvin Toffler already predicted in 1980, consumers do not want to remain passive, they will become ‘prosumers’ (Toffler 1980). This is also argued by Prahalad and Ramaswamy, who argue that the role of the consumer has changed ‘from isolated to connected, from unaware to informed, from passive to active’ (2004, p. 4). The second development is the open source software movement. This trend in the software industry inspired many entrepreneurs to apply its principles to fields other than software. The third development consists of the increasing accessibility of information and decreasing cost of production tools, like digital cameras and editing software. This is most clearly seen in the creative industry, including film making, photography, music, and fashion. The fourth and last development is the emergence of online communities, in which the online population is organized. Consumers are increasingly using online tools to share ideas and to interact (Casalo et al. 2008). Although the definition of community in this context is ambiguous, the fact that an online platform is available for people to contribute to various activities, is an important development in this context.
Although not much is written in literature about crowdsourcing specifically, the developments that have led to the emergence of crowdsourcing suggest some other concepts that are relevant in this context. First of all, open source software literature has provided many concepts and
3
propositions that also seem to be relevant for studying crowdsourcing. Furthermore, much is written about involvement of users in product development (Von Hippel 1988, 2005; Desouza et al. 2008; Barnard and Wallace 1994; Griffin and Hauser 1993). Previous research also exists on the fourth development, the emergence of online communities. Finally, because open innovation is believed to be the precursor of crowdsourcing, the findings in this field are also likely to be relevant in crowdsourcing. These fields in literature are described and discussed in this report.
1.3 RESEARCH OBJECTIVES
The objective of this study is twofold. First of all, there is a knowledge problem, because there is no theoretical knowledge about crowdsourcing (Denyer et al. 2008). Therefore, a theoretical framework will be developed that describes and structures crowdsourcing in its totality. This theoretical framework defines the scope of the topic and describes the different types that exist with their associated characteristics. Furthermore, propositions will be developed from the findings that will address the knowledge problem. Secondly, there is a construction problem, as there is no practical knowledge about how to apply crowdsourcing. Therefore, a protocol will be developed, which will support organizations that want to apply crowdsourcing (Van Aken 2007a; Denyer et al. 2008).
This protocol is partly focused on the needs of the organization where the study has taken place. This company is a small consulting firm, named Altuition, which was founded in 1997 and consists of 12 consultants. Altuition is active in the consultancy sector, specialized in helping companies to better listen to their customers. Currently, they are using methods that give small groups of customers the opportunity to express their opinion and needs to an organization. Because crowdsourcing could be helpful to get customers to participate and discuss their ideas and opinions, this study is relevant for this organization. Crowdsourcing can improve or supplement their methods as not only more customers can be involved in the process, but they can have a more active role, because they can decide when and how to participate.
Although the final research objective is design‐oriented, namely to develop a protocol that organizations can use when they want to introduce crowdsourcing to their organization, the total study will be mainly theory‐oriented (Dul and Hak 2008). This is because the main objective of this study is to gain a better understanding of crowdsourcing in general. And while the protocol will acknowledge the needs of Altuition, its objective is to provide guidelines for practitioners in general. The information in the protocol will be derived from answering several research questions, which will be presented in this section.
First of all, because there are no previous studies about this topic, a theoretical framework has to be developed to gain a better understanding of crowdsourcing in general and which types exist. Since a clear definition of the concept is already presented in the introduction, the next step is to investigate which variations exist within the crowdsourcing field. Therefore, the first
earch question is: res
d and1. Which types of crow sourcing exist how can they be distinguished?
The objective of this question is to identify different types of crowdsourcing and their most common characteristics. The framework will contribute to theory, as it will contribute to the understanding of this new topic and can be used to structure future research into
4
crowdsourcing. The framework will also contribute to practice, as it will be used as a basis for the protocol.
As was shown through the changes in the definition of crowdsourcing, investigating the participants that form the crowd is an important issue. While in some cases the crowd is undefined, as in the original definition, in other cases there is a more specific target group or some people are even excluded from participation. Therefore, the composition of the crowd is e of the major topics of this study. The second research question is therefore: on
2. Which types of participants are expected?
This question focuses on the characteristics of the crowd in general and for each type specifically. This information helps to gain an understanding of the possibilities of crowdsourcing.
An important issue that is derived from literature search into related topics is the motivation to participate. Because this study is focused on the organizations that are applying crowdsourcing instead of the crowd itself, these motivations are seen as conditions that should be met to achieve both a high quantity and quality. According to Boxall and Purcell (2003), performance is not only influenced by motivation, but also by ability and opportunity. Since crowdsourcing provides the opportunity to participate, conditions that increase the ability to participate have to be identified. For example, the specificity of the request can influence the ability of the crowd to contribute. Furthermore, the topic of the request can influence the motivation and ability of the crowd, since the topic can be too complex to contribute or too uninteresting. These issues are likely to influence the design choices of a crowdsourcing initiative. Therefore, the third
earch question is: res
3. Which conditions must be satisfied to generate productive outcomes?
The final research question is related to the previous one, but has a more practical focus. Apart from the conditions that are elaborated on in the previous question, to start up and maintain a crowdsourcing initiative, certain efforts are needed from the organization to make this a ccess. Therefore, the fourth research question is: su
4. Which efforts are needed from the company to facilitate the crowdsourcing process?
Issues like promotion, processing contributions and moderation are relevant here. Since it is unlikely that much quantitative information can be found on this research question, the focus will be on more qualitative results. Therefore, the answer to this question will mainly consist of activities that need resources from the company.
This introduction is the starting point for this study, providing a definition and context to the subject. The research questions are believed to be covering the total field of crowdsourcing, and will be used to guide data collection and analysis. In the next chapter, the research methodology will be explained.
5
2 RESEARCH DESIGN AND METHODOLOGY
In this chapter, the methodology that is used in this study is explained. This methodology is mainly based on the works of Dul and Hak (2008) and Yin (2003), who both focus on case study research, which is the main research strategy in this study. The work of Van Aken (2004; 2007a; 2007b) and Denyer et al. (2008) is used to describe the methodology that is used for the protocol, since they focus on design methods.
2.1 RESEARCH APPROACH
As mentioned earlier, this study falls under the category of theory‐oriented research, as explained by Dul and Hak (2008), because the main goal of the study is to contribute to theory development and to the knowledge of practitioners in general, but not to one specified practitioner. The reason for this is that little knowledge is available about the topic, which makes it difficult to conduct a more focused and detailed practice‐oriented study, which would imply hypothesis testing. The lack of previous studies is also the reason that the research objective is theory‐building, as opposed to theory‐testing. This means the objective of the study is to develop theoretical propositions regarding the topic, in this case crowdsourcing. According to Dul and Hak (2008), propositions consist of concepts and specifications of relations between concepts, which together form a theory. Because the ultimate objective of this study is the development of a protocol, which is design‐oriented, the propositions will be formulated as design propositions instead of theoretical propositions (Van Aken 2007a; Denyer et al. 2008). However, before the theory‐building phase can start, exploration of both theory and practice is needed. In the following sections, the phases of this study will be explained further.
Exploration of theory
According to Dul and Hak (2008), the aim of exploration of theory in theory‐oriented research is to find candidate propositions for testing. Because no literature is available directly about crowdsourcing, related literature was explored. These related concepts are the ones already mentioned in the introduction, namely user driven innovation, open source software development, online communities, and open innovation. A description of these concepts and their relation to each other and to crowdsourcing is given in the next chapter.
From studying this literature, some relevant concepts were found that are likely to be useful in the context of crowdsourcing. In some cases, some propositions could be found that seemed relevant to the research questions. However, many of these propositions need even initial testing in the context that they were developed in, so their relevance cannot be assumed without questioning. Furthermore, it is likely that other concepts and propositions are relevant in the context of this study, which could not be found in literature. Therefore, an exploration of practice was needed both for the purpose of confirming the relevance of the concepts and propositions that were found and for finding new concepts.
Exploration of practice
The sources for the exploration of practice are the many cases of crowdsourcing that can be found in documentation on the Internet and in management literature. Information on
6
are described separately.
The purpose of the first research question is the development of a theoretical framework of crowdsourcing. The answer to this question will be mainly descriptive. This question is answered by conducting a multiple case study, using the documentation as main source of information. This process occurred iteratively with the exploration of practice, since the same
crowdsourcing examples was found in press articles, weblogs that collect examples (e.g. Springwise and Crowdsourcingdirectory), company websites and management literature. The reason that this source of information was chosen is simply that this is the only information on crowdsourcing that is currently available. To ensure the validity of the information, for most cases more than one source was used, as can be seen in appendix A. Management literature provided additional information on examples, but also more general clues that were useful in formulating relevant propositions. Through exploration, the concepts to be included in the study were developed more clearly, operational definitions could be developed, and the research questions and research design were refined (Cooper and Schindler 2003).
To determine which cases and information to include or exclude in the study, it was important to determine the unit of analysis. As will be explained in the next chapter, because the topic is so broad and the cases have different characteristics, this is certainly not a trivial issue. Since the protocol is focused on organizations that want to apply crowdsourcing, these organizations are important to determine the unit of analysis. Although some organizations are based on crowdsourcing in their totality, this is not a suitable unit of analysis, as in most cases only part of the organization is affected by crowdsourcing. Therefore, only those parts of the organization that are directly affected by crowdsourcing are included in the unit of analysis. This is however not sufficient, as crowdsourcing by definition considers people outside the organization. Therefore the participants are included in the unit of analysis.
This part of the study can be seen as an exploratory multiple case study as explained in Yin (2003). According to Yin (2003, p. 1), ‘case studies are the preferred strategy when ‘how’ and ‘why’ questions are being posed, when the investigator has little control over events, and when the focus is on a contemporary phenomenon within some real‐life context’. As the main research question is a ‘how’ question, the researcher does not have the possibility to control events, and crowdsourcing is certainly a contemporary phenomenon where there is no clear boundary between case and context, case study research would be the right choice in this study. This part of the study both revealed new concepts and confirmed the relevance of some of the concepts that were found in the literature study. Comparison of the cases resulted in a list of attributes on which cases could be categorized. Because the objective of the exploratory phase was to identify relevant attributes, all cases that matched the definition were included to make sure the attributes were as exhaustive as possible. The attributes will be defined and explained in chapter four.
Although the exploration phase addresses several concepts that are likely to be relevant when studying crowdsourcing, the propositions that were found are limited, in that they are not able to answer the research questions sufficiently. This is the reason that the study continues with theory‐building research, as opposed to theory‐testing research.
Theorybuilding research
In this section, the main part of the study will be explained. The first research question, which requires the development of a typology for crowdsourcing, has a somewhat different character than the other three questions. While the first research question is directly focused at crowdsourcing and uses mainly the exploration of practice, the analysis of the other three questions is based on the exploration of theory. Therefore, the approaches to these questions
7
information sources were used. The iterative approach that was used is shown graphically in Yin (2003, p. 50), as it shows a feedback loop from conducting the case studies to selecting cases and the data collection protocol. In this phase of the study, many cases led to new searches for examples and new attributes to find information on. The data was collected in a table, with the attributes forming the columns and the cases forming the rows, which is shown in appendix A. This table also served as the data collection protocol, since this made sure data collection was structured and the same information was collected for each case.
Since there are no strict rules that can be applied to the number of case studies that are needed, the number of case studies depend on the perceived differences between and within cases and the perceived complexity considering external validity (Yin 2003). Because the goal was to discover different types with associated characteristics, the study should have at least two individual cases within each of the types, so that the theoretical replications across types are complemented by literal replications within each type. As explained earlier, the sources of information in this phase of the study are the same as used for the exploratory case studies. All the cases for which sufficient information was available were included in this phase of the study. This resulted in 115 cases that were included in this multiple case study, varying from 15 to 40 cases for each of the four types. This variation is due to the fact that the cases were selected according to the definition of crowdsourcing in general, since the classification was not known yet.
The analysis that was conducted on this table resembled the way the structure of organizations was analyzed by Mintzberg (1979). According to his configuration theory, effective structuring requires an internal consistency among the design parameters. The objective of the analysis on the crowdsourcing cases was to find an internal consistency that could lead to the development of a typology. This typology was found through clustering (Yin 2003) where the objective was to find clusters of cases with similar internal consistencies. Consistent with Mintzberg’s results, not all cases match one configuration perfectly, there are hybrids, but in all cases they resemble one configuration better than the others. These configurations form the typology that answers the first research question.
While the result of the first research question is mainly descriptive, the results of the remaining three research questions will be more explanatory. The basis for studying these research questions, which discuss the participants, the necessary conditions, and the organization’s efforts, was the exploration of theory. The analytic strategy that is used in this phase of the study is relying on theoretical propositions. This means that while openness towards alternative propositions was required because of the exploratory nature of the study, the propositions that were developed from literature guided data collection and analysis (Yin 2003). Because the theoretical propositions will eventually be used for the protocol, which needs a holistic design, not one theoretical perspective will dominate, but all relevant information will be discussed (Van Aken 2007b).
To study these three research questions, and develop propositions that are relevant for answering these questions, two sources of information were used, namely documentation and interviews. Documentation included management literature and again the information found on the Internet, as additional information was available besides the attributes that were used to develop the typology. Additionally, interviews were held with key informants in organizations that use crowdsourcing, to gain a deeper understanding of the choices and challenges that organizations face when they are using crowdsourcing. In total, four interviews were held. For practical reasons, these interviews were only held with Dutch organizations. It is likely that each case has some unique characteristics that make generalizability of findings problematic. Therefore, the interviewees were not only approached as information sources on their own
8
case, but also seen as experts on the topic. Therefore, respondents were asked to assess their case’s generalizability, as described in Weiss (1994).
Although the goal of this study is theory‐building, of which the main result consists of theoretical propositions that need further testing, the propositions are shown to be true in the cases that were studied, which can be considered initial testing (Dul and Hak 2008). Both descriptive and explanatory propositions were developed. Descriptive propositions typically state the existence, size, form, or distributions of some variable, while explanatory propositions describe a relation between concepts (Cooper and Schindler 2003). However, the explanatory propositions will be formulated as design propositions (Van Aken 2007a; Denyer et al. 2008). This choice was made because the explanatory propositions are not only relevant for theory, but also for practice in the sense that they will be used for the protocol. Furthermore, the difference between the two possible formulations is not significant in this study. The theoretical propositions can be easily derived from the design propositions, in the case that they are used for further testing.
Developing the protocol for crowdsourcing
The final phase of the study consists of processing the results of the previous phases into a protocol. The typology that was developed to answer the first research question forms the basic structure of the protocol. The propositions that were developed in relation to the other three research questions will provide content for the protocol. These propositions are formulated as design propositions according to CIMO‐logic: in this class of problematic Contexts you may use this Intervention type, which will produce through this generative Mechanism these Outcomes. The outcomes are generally direct outcomes, not bottom line outcomes like success or profit (Van Aken 2007a). An example of such a proposition is: ‘If you have a project assignment for a geographically distributed team (class of contexts), use a face‐to‐face kick‐off meeting (intervention type) to create an effective team (intended outcome) through the creation of collective task insight and collective commitment (generative mechanisms)’. This formulation facilitates the translation into a protocol and the application of the research in practice. The interventions, contexts and outcomes in this study mainly consist of the attributes that were developed in the exploration of practice. The mechanisms are derived from the explanatory part of the study, and are mainly based on literature confirmed by empirical findings.
The protocol consists of a decision diagram, showing the possible options for the types and related design choices together with their advantages and disadvantages. When a type has been chosen and associated design choices have been made, the protocol continues for each type with four additional phases. The conditions and efforts that are needed in each phase will be explained according to the findings in terms of the design propositions. Because they are formulated according to CIMO‐logic, they can be directly translated into the protocol and provides practitioners with important information on which interventions will be important in which context. As explained by Van Aken (2004; 2007a; 2007b), the protocol cannot be applied directly, but is meant to support the design of solutions by professionals.
2.2 RESEARCH QUALITY
In this paragraph, the quality of the research is assessed in terms of construct validity, external validity and reliability, which are the main quality criteria for case study research (Yin 2003). A fourth quality measure is internal validity, but according to Yin (2003), this is only relevant in the case of explanatory research. Although the propositions include causal relations, the
9
Reliability
Reliability is achieved by demonstrating that the operations of a study can be repeated, with the same results. The goal of reliability is to minimize the errors and biases in the study (Yin 2003). Reliability is a necessary, but no sufficient condition for validity (Cooper and Schindler 2003). Especially with a multiple case study, using a protocol to guide data collection is essential for increasing the reliability of the study (Yin 2003). In this study, to make sure the data collection was structured, table shells were used (Yin 2003), with the cases forming the rows and the attributes forming the columns. The largest threat to the reliability of this study is however in forming the table shells, in developing the concepts and their measures. As argued by Dul and Hak (2008), in exploration and theory‐building research, there is no specific method available for finding candidate concepts; they just emerge from the data. This emergence is based on the
causality are only based on theory, not on empirical data. Since the propositions are the result of theory‐building research, the internal validity can only be assessed after further testing.
Construct validity
Construct validity is the degree to which the operational measures reflect the theoretical concepts being studied (Yin 2003). Construct validity is usually questioned when rather abstract concepts have to be measured (Cooper and Schindler 2003). In this study, however, the concepts that are measured are already practical, because these concepts were developed from practice, instead of theory. However, a disadvantage of this approach would be that there is no way to evaluate construct validity with previous research (Cooper and Schindler 2003). To assure that construct validity can be assessed, the concepts that are used for this study are extensively explained in chapter four. Multiple sources of evidence are used, as both documentation and interviews were used in this study to improve the validity of this study. Considering the documentation, for each case also multiple sources were used. Additionally, the results of the interviews as interpreted for the study were reviewed by the interviewees to make sure the interpretations are valid (Yin 2003).
External validity
External validity determines the extent to which the study’s findings can be generalized (Yin 2003; Cooper and Schindler 2003). Because for answering the first research question more than one hundred cases are used, the external validity of these results is rather high. Specifically, cross case synthesis is used as an analytic technique to increase the validity of the study (Yin 2003). This technique is used by organizing the data into a table, which makes the comparison of cases easier. The validity of the results of the remaining research questions was increased because more than two cases were studied for each type (Yin 2003). However, they are based on much less cases than the findings on the first research question, which would make the external validity of the study questionable. There are two reasons to argue against this. First of all, Yin (2003) argues that the generalization issue in case study research differs from other methods. Survey research, for example, relies on statistical generalization, where the results are argued to be generalizable to samples and universes because a representative sample from a population is studied. Case study research, however, resembles experimental research in that it relies on analytical generalization, which is generalization to theoretical propositions. This is the reason that propositions from related literature are studied in the context of crowdsourcing, although it is acknowledged that not all results of this study can be traced back to previous literature and theoretical propositions. Second, because only part of the study is based on theoretical propositions that were found in previous studies, the main focus of the study is theory‐building. This means the results are only subjected to initial theory‐testing, and it is acknowledged that further research is needed to increase the validity of this study.
observation of differences and similarities between cases, which indicate that a particular concept might be relevant in explaining the variances between cases. As explained in Miles and Huberman (1994), the practical significance of the differences found in the comparisons was assessed. Although a critical view was held during this process, it remains to be subjective.
2.3 STRUCTURE OF REPORT
The remainder of this report will be organized as shown in figure 1. The size of the arrows show the importance of the influence of one phase on the other. In chapter three, the results of the exploration of theory will be given. This consists of a description of the related literature, together with their relation to each other and to crowdsourcing. The results of the exploration of practice are presented in chapter four, consisting of the attributes that were found relevant for the analysis of the crowdsourcing examples and are thus included in the study. These attributes are used in chapter five, where the first research question will be discussed, forming a typology for crowdsourcing. In chapter six, the remaining research questions will be discussed, analyzing the results of the theory‐building research. The results of the four research questions will be processed into a protocol, which will be presented in chapter seven. The protocol consists of the relevant activities and design choices to make in each phase of the process, along with the related advantages and disadvantages. Finally, chapter eight will present a general conclusion, summarizing the conclusions for the research questions and the implications for both theory and practice.
10
Exploration of theory
Chapter 3
Exploration of practice
Chapter 4
RQ1: crowdsourcing
types
Chapter 5
RQ2:participants
Chapter 6.1
RQ3:conditions
Chapter 6.2
RQ4:efforts
Chapter 6.3
Protocol
Chapter 7
Theory-building
FIGURE 1: STRUCTURE OF REPORT
11
3 EXPLORATION OF THEORY
In this chapter, the theoretical concepts that are related to crowdsourcing is described. The related literature consists of user driven innovation, open innovation, open source software development and online communities. Additionally, their relation to each other and to crowdsourcing will be discussed. Several propositions that were found in this literature will be used for the theory‐building phase, which is described in chapter six.
3.1 USER DRIVEN INNOVATION
Users can be a valuable source of information in the innovation process (Von Hippel 1988; Desouza et al. 2008; Griffin and Hauser 1993; Magnusson et al. 2003), as insufficient knowledge of real customer needs is seen as one of the main reasons for market failure of new products (Barnard and Wallace 1994). Because customers are an important target group in crowdsourcing, the literature about involvement of customers in innovation in general is discussed in this paragraph.
The most common traditional methods to involve consumers in the product development process are focus groups, surveys, and questionnaires. However, often limited information can be identified with these methods, as they are focused on obtaining information about current products and tend to control the process too much (Barnard and Wallace 1994; Ciccantelli and Magidson 1993). Furthermore, they tend to treat customers as objects of research, as passive participants, which prevents the company from getting the most out of this information source. Although many new methods have been developed that are improvements of traditional methods, where the process is less controlled and the focus is more on tacit knowledge, none of them have been able to take a deeper look into the tacit knowledge of a large number of consumers.
Crowdsourcing approaches users with an open call, encouraging them to participate in ways that they are interested in and communicate with each other instead of only to the company, which makes the whole process more natural. This means more tacit knowledge can be captured while more users can be involved. Crowdsourcing is significantly different from conventional (online) market research, because consumers are invited to contribute their creativity and problem solving skills by generating and evaluating new product ideas, and discussing and refining these ideas into more detailed product concepts (Füller et al. 2006). The literature on user driven innovation reveals important information about when these customers should be involved, who should be involved, and how organizations can help customers to express their needs.
Many researchers argue that users should be involved in the process as early as possible (Ciccantelli and Magidson 1993; Füller and Matzler 2007). According to Nambisan (2002) it has been proven that customers’ input in new product development (NPD) is mostly relevant in continuous innovation, while they are less relevant when it concerns emerging technologies and markets.
Von Hippel’s concept of lead users has found wide acceptance (Brockhoff 2003). According to Von Hippel, lead users have the ability to identify product characteristics that become valuable for most customers in the future. Furthermore, lead users are likely to initiate innovation
12
activities as they personally can benefit from them (Von Hippel 1989). Particularly radical innovations often trace back to the ideas of lead users (Janzik and Herstatt 2008). This specific target group is therefore likely to be relevant for crowdsourcing. Hung et al. (2008) found that lead users are more likely to share their inventions and to respond to problems posted by others.
The possible motivations of users are discussed by Brockoff (2003). He mentions financial benefits, extra services during product use, and reputation as possible rewards. Concerning financial rewards, he argues that having no rewards might lead to decreased knowledge sharing, while having high rewards might lead to a decrease in quality as the customers are only motivated by the rewards.
One of the reasons that organizations fail to adapt products to the needs of their customers is that customers find it difficult to make their needs explicit. A helpful theory to illustrate this is the Kano model (Kano 1984), which distinguishes between three types of customer needs, namely basic, performance and excitement needs. Basic needs are product features that customers expect to find in a product, while excitement factors are features that delight the customers because they do not expect to find this, which represents an important part of the reasons why a product is bought. However, only performance factors, which have a linear relation to customer satisfaction, are articulated by customers, while basic and excitement factors are not mentioned. This is also the reason that many studies have found that using customers as a source of innovation only results in incremental innovations (Füller and Matzler 2007). Several researchers argue that visual representation of products, for example in the form of virtual prototypes, can help customers articulate their basic and excitement needs (Füller and Matzler 2007; Mascitelli 2002), because an experience is created which helps to create the need (Lauglaug 1993). The possibilities of doing this are increased by the Internet, which also makes this issue relevant for crowdsourcing.
In literature, the question has arisen whether the quality of input that comes from customers is good enough compared to professionals. Kristensson et al. (2004) have found that ordinary users create significantly more original and valuable ideas than professional developers and advanced users, while these create more easily realizable ideas.
The literature on user driven innovation shows that users are an important source of information in innovation, not only because the company depends on fulfilling their needs, but also because they are able and willing to innovate. One of the most difficult challenges is to capture the tacit needs of a large group of customers. Because crowdsourcing offers the opportunity to engage a large group of customers in a natural conversation, in which the use of virtual representations of the product is possible, crowdsourcing can offer important improvements to traditional methods.
3.2 OPEN INNOVATION
Open innovation is related to crowdsourcing in the sense that they both focus on using external sources for innovation, and combining internal R&D with the knowledge of external sources. Increasingly more companies feel the need to apply open innovation principles, as their R&D
ebudgets are increasing at a faster rate than sales (Dodgson t al. 2006; Howe 2006).
The open innovation paradigm tells companies to look outside their own organization for innovation (Chesbrough 2003). The closed innovation model is based on the assumption that
13
successful innovation requires control and should happen within the own organization, and that useful knowledge is scarce, hard to find, and often unreliable. The open innovation model assumes that knowledge is widely distributed, and generally of high quality, which makes it attractive to search for this knowledge (Chesbrough 2003; 2006).
Because of a changing business environment, companies realized they needed to work together with other parties, such as customers, rivals, academics, and firms in unrelated industries, to improve their innovativeness (Chesbrough 2003; 2006). ‘The role of R&D needs to extend far beyond the boundaries of the firm. Companies must integrate their ideas, expertise and skills with those of others outside the organization to deliver the result to the marketplace, using the most effective means possible’ (Chesbrough 2003, p. 41).
One of Chesbrough’s main arguments is that the technologies that are developed are in itself not very useful. For organizations to create a competitive advantage, the business model is of much greater importance, because this creates and captures the value from the technologies. Utilizing external technology can help leverage a firm’s business model, for example by speeding up and improving the internal R&D, but also by creating complementary products and services that stimulate the acceptance of the internally developed technologies (Chesbrough 2006). The own technology competences appear to decrease in importance, while accessing technologies, markets and partners, and expertise in fields different from the companies’ core technologies become increasingly important. Enabling creativity, opportunity recognition, and connectivity into new domains are important core competencies in open innovation. New roles are created, such as idea hunters and gatherers, to identify potentially useful external technologies (Chesbrough 2006).
Open innovation appears to be a good model for pursuing more radical innovations. Compared to incremental innovations, more risk is involved in radical innovations. Because many traditional methods for selecting ideas are based on short term metrics, the more radical innovations are often abandoned. When working together with other parties, and with the opportunity to license a technology to other parties, the risk of developing radical innovations can be decreased. Furthermore, radical innovations require expanding a companies’ domain, which makes it more likely that a company needs partners from other fields to speed up and improve the development process (Chesbrough 2006). Another important source of radical innovation are start‐ups, which have an important role in the open innovation model, because they give companies the possibility to invest in them to gain access to new technologies (Chesbrough 2006).
Open innovation assumes that innovation markets are an effective way to manage innovation. They are necessary to make sure no good ideas go to waste, because companies value opportunities differently, and with innovation markets ideas can flow to the right places where they are most useful. Intellectual property (IP) has an important role in these innovation markets. While in the closed innovation model, IP was used to protect technologies from being used by other parties, IP is now actively used to exchange knowledge (Chesbrough 2003; 2006).
Crowdsourcing can be seen as an extreme form of open innovation. Where open innovation implies markets between organizations, crowdsourcing implies markets between organizations and the crowd, or even a market between members of a crowd. The benefits that are found in open innovation, such as risk reduction, increasing development speed and improved innovativeness are likely to be similar in crowdsourcing. As in open innovation, new roles are created, like community managers. Additionally, a similar shift in the way of thinking is needed to apply crowdsourcing, because organizations need to open up to the crowd and show their commitment and trust. This requires important challenges for many organizations, which are
14
used to having control of their processes and the communication towards the world outside their company.
3.3 OPEN SOURCE SOFTWARE DEVELOPMENT
As already mentioned in the introduction, the open source software (OSS) movement is one of the causes of the crowdsourcing trend. Many crowdsourcing cases are inspired by OSS (see for example appendix B.1 on the Goldcorp Challenge and appendix B.8 on Wikipedia). Crowdsourcing is sometimes even defined as applying OSS principles to fields other than software.1
OSS is software of which the source code is available to the public and therefore can be used, changed, and redistributed according to specific licensing rules to make sure no one can appropriate the code. The success of OSS like Linux is widely acknowledged. Although many people did not believe in the concept at the start (because why would people do this work voluntarily, working with people they have never met before?), there were many people who were willing to do the job, and they did it successfully. Several open source products have demonstrated that these projects can indeed lead to software systems with high functionality
A nd S . and quality ( mant a till 2007)
While open source was at first seen as competition to proprietary software developers, companies like IBM are now increasingly investing in open source (Chesbrough 2006; Amant and Still 2007; Van Wendel de Joode 2005). They found ways to use OSS not only for value creation, but also for value capturing. The difficulty with software is that development and production are very close, which makes it more vulnerable to share with competitors than for example scientific knowledge (Grand et al. 2004).
One of the most researched topics of OSS is the motivation of people to participate. Because the participants are typically not paid for their effort, only intrinsic motivation must generate the willingness to participate. Hertel et al. (2003) have found three main motivators, namely the identification with the community, personal benefits from the code and participants’ tolerance of time investments. These motivators were also found by other researchers (Bloem and Van Doorn 2006; Amant and Still 2007). An interesting finding is that many participants start participating with direct utility as motivator, because they need a solution to a personal problem, but continue for other reasons, which are more intrinsic and community‐related (Shah 2003).
An OSS community consists of several layers. There is often one project leader, several (core) developers, active users, and passive users. A general rule of thumb in OSS is that about one percent of participants are core developers, ten percent are active participants, and the remaining part consists of passive users (Amant and Still 2007). Specifically in OSS, it is estimated that the number of active users, who report bugs but do not program themselves, is about three million, while there are about ten to thirty thousand core developers (Bloem and Van Doorn 2006).
Open source communities are often described as anarchies (Van Wendel de Joode 2005), but they seem to be more of similar to meritocracies (Chesbrough 2006; Bloem and Van Doorn 2006). A meritocracy is about empowering the experts, who are chosen and respected by the members of the community, because they have shown their expertise, not because of their creditentials.
15
From this description of OSS, some important issues emerge that are likely to be relevant in crowdsourcing as well. First of all, the findings on the motivation of participants provide important suggestions for crowdsourcing. Intrinsic motivations are found to be sufficient in OSS, and although financial rewards are possible in crowdsourcing, in many cases intrinsic motivations also seem to be important. Additionally, there are several characteristics of OSS that are also likely to be found in crowdsourcing, such as the meritocratic organization and the division of participants’ activity. The mechanisms that were found in OSS literature can help explain some of the crowdsourcing cases and can be used to inspire even more applications of crowdsourcing.
3.4 ONLINE COMMUNITIES
Much literature is available on online communities, sometimes called virtual communities. Online communities are groups of people who have a common interest and interact with each other online (Dholakia et al. 2004; Farquhar and Rowley 2006; Sicilia and Palazon 2008). In 2007, approximately 230 million people worldwide were active in online communities (Janzik and Herstatt 2008). As the definition of crowdsourcing that is used in this report already states that a platform is provided online for which users provide content, the most important characteristic of an online community for this study is that there is interaction between the members. As will become clear later in this report, not all crowdsourcing cases include interaction between participants, which means that not all crowds can be called communities.
The literature mainly focuses on customer communities, which are generally analyzed from a marketing perspective. One of the challenges from a marketing perspective is to recognize that the traditional focus on the company‐consumer relationship now has to be complemented with a consumer‐consumer relationship (Farquhar and Rowley 2006). However, online communities can be used for other purposes than marketing alone. Online communities can be used to engage customers in product development or customer service activities, and have become an important source for identifying the needs and problems of users (Janzik and Herstatt 2008). Although it is acknowledged that online communities are very useful for marketing purposes, this perspective is not discussed in this study.
One important study into online communities is the one of Nambisan (2002), who integrates different theoretical perspectives to develop a framework for online communities. First of all, he argues that the roles of customers vary in different phases of the NPD process, namely concept, development, testing and customer service. It is argued that the activities in the first three phases are mainly task oriented, because the interaction is mainly between the company and the customer, while in the customer service phase, there is more interaction between the customers, making the activity more socially oriented. Related to this issue is the way an identity is formed, as a task orientation results in an individual identity, while a social orientation results in a social identity, where an identity is based on being a member of a community. These concepts are also related to the motivation of participants. Farquhar and Rowley (2006) argue that motivations are linked to the value a community creates. Several studies have found both individual and social benefits to be relevant (Nambisan 2002; Bogazzi and Dholakia 2002; Farquhar and Rowley 2006). Personal benefits include direct benefits like knowledge and financial rewards, while social benefits consist of the benefits participants receive from the interaction with other participants, such as a social identity and reputation within the community. An overview of possible motivations for participation in online
16
communities is given by Antikainen and Väätäjä (2008) and Janzik and Herstatt (2008). Most of these motivations can be categorized as social or community‐related benefits.
The literature on online communities is very useful in the context of crowdsourcing. Especially the discussion about customers’ motivations and the involvement in different phases of the NPD process shows many similarities with crowdsourcing. However, not all crowdsourcing examples use online communities, because interaction between participants is not always possible. Furthermore, the focus on customers shows only one side of the story, as customers are not the only relevant target group in crowdsourcing.
3.5 CROWDSOURCING IN PERSPECTIVE
In this paragraph, crowdsourcing will be placed in perspective with the other four concepts. First it is important to acknowledge that these four concepts are also interrelated. As both open innovation and OSS rely on external sources for innovation, they are closely related. However, because open innovation focuses on capturing value from innovation, while OSS focuses on value creation, there are important differences. Therefore, open innovation and open source software are only related when a business model is used to capture the value from OSS, which is increasingly done by companies that use complementary products or services to capture the value of OSS. User driven innovation is related to both open innovation and OSS, as users are an external source of information and many participants in open source software communities are also users. The relation is however only partly overlapping, because customers are not the only target group of open innovation as well as for OSS. Online community literature is mainly related to user driven innovation, as the literature on online communities mainly discusses customer communities. Therefore, much of the literature on user driven innovation has been used for developing theories on online communities. Online communities are also closely related to OSS, since open source software is developed in online communities.
This is where crowdsourcing comes in. First of all, crowdsourcing can be seen as an extreme form of open innovation. It tends to go a little further than the general principles of open innovation, as open innovation usually still implies collaboration between companies and institutions, while crowdsourcing also involves private individuals, who perform the task in their spare time. An important similarity between open innovation and crowdsourcing is the use of a business model for value capturing. This is important for the relation with open source software. Open source software can be seen as a form of crowdsourcing, as a community of individuals creates something that was traditionally done by companies. However, in contrast to open source, the value of the contributions of the crowd in crowdsourcing is in many cases captured by companies, which make large profits of the crowd’s contributions (Brabham 2008). Crowdsourcing is related to user driven innovation, but gives the users a much more active role than traditional methods to capture customer needs.
As can be seen from the analysis, crowdsourcing is related to many concepts, but it often represents an extreme form or an improvement to the more traditional methods. Compared to more traditional user driven innovation, crowdsourcing enables companies to capture the more tacit needs of more customers, while at the same time improving the relationship with these customers. Compared to open innovation as described by Chesbrough (2003, 2006), crowdsourcing makes even more use of external sources of innovation. Compared to the applications of online communities that are discussed in literature, crowdsourcing involves even more purposes for online communities, such as using other target groups than customers,
17
and for other tasks. Therefore, crowdsourcing can indeed be seen as a new business model, which offers companies increasingly more possibilities for multiple objectives.
3.6 CONCLUSION
The chapter consisted of the literature research that was done for this study. The concepts that are related to crowdsourcing were described and placed in perspective, relating them to each other and to crowdsourcing. This analysis showed that crowdsourcing is a variant of different concepts and that it can be seen from different perspectives. Crowdsourcing could be considered as an improvement to more traditional methods of user driven innovation, as it can capture the more implicit needs of a large number of customers. It can also be seen as an extreme form of open innovation, as crowdsourcing uses even more external knowledge sources as is advocated in open innovation literature. Open source software development can be seen as a well‐developed example of crowdsourcing. Finally, online communities can be important for crowdsourcing, as these are an important part of many crowdsourcing examples.
The findings in research in these related fields will be used to guide data analysis. First of all, the literature study has helped to guide the identification of relevant concepts to include in the study, which will be presented in the next chapter. Second, the results of the literature study will be used in chapter six, where the theory‐building phase of the study will be presented and propositions for crowdsourcing will be developed.
18
19
4 EXPLORATION OF PRACTICE
This chapter describes the results of the exploration of practice, in terms of the attributes that were found useful in describing the crowdsourcing cases and their differences. These attributes represent the possible characteristics of the crowdsourcing cases. The information found on these attributes forms the most important source for answering the research questions, together with the findings from the literature study. Although some of the attributes were based on the concepts found in the literature study, as described in the previous chapter, most of the attributes emerged from exploring crowdsourcing examples that were found in documentation. Although there is no specific method that was used for finding these attributes, the selection is based on the perceived relevance in explaining the differences between cases. Therefore, the paragraphs will not only contain a description of the attribute itself, but will also discuss for which purpose it is included.
The attributes are grouped into different categories. The categories represent the different parts of design propositions according to CIMO‐logic (Van Aken 2007a; Denyer et al. 2008), namely intervention attributes, context attributes, and outcome attributes. Mechanism attributes will not be included, as the theory will present the mechanisms that explain why the intervention would lead to the desired outcome. The interventions, which will be presented in 4.1, represent design choices and will be used to identify the typology that will be described in the next chapter. Each of the ten intervention attributes consists of a limited number of options. In 4.2, two context attributes will be presented that represent possible contexts in which the interventions are relevant. In 4.3, the outcome attributes are described, which represent the outcomes of the crowdsourcing initiative that are used as a measure for success. In contrast to the intervention attributes, there is no limited amount of possibilities, but the content is more open. Information on the outcome attributes is not available for all cases, but as much information as possible is collected. The information that is collected on these attributes can be found in appendix A.
4.1 INTERVENTION ATTRIBUTES
The first ten attributes represent design choices that organizations in the cases have made, consciously or unconsciously. These attributes are the main source of information to answer the first research question and to develop a typology for crowdsourcing. However, they are used to answer the remaining three research questions as well. For each of the attributes, a limited number of possible choices is available, presented between brackets.
Role of the company (mediator / own initiative)
One of the most important differences between the cases is that some organizations act as a mediator between the crowd and other parties, while in other cases crowdsourcing is used for the company’s own purposes. In the case of a mediator, the case represents an organization that connects the crowd to another organization. A mediator gives organizations the opportunity to engage the crowd without actually organizing everything themselves, implying that the crowdsourcing initiative is essentially outsourced to another organization. In the case of an own initiative, the case represents an organization that has created an opportunity for the crowd to participate through their own website.
20
This issue caused difficulties in defining the unit of analysis, as described in chapter two. In the case of a mediator, the unit of analysis includes the whole organization and the two other parties that the company mediates between. In the case of an organization’s own initiative, the unit of analysis includes only the part of the company that was directly influenced by crowdsourcing, together with the crowd itself.
Type of participation (voting; rating; commenting; forum; weblog; social network; wiki; upload; financial)
This attribute shows in what way the crowd can participate and gives an idea of how open the request is. To explain the options of this attribute, it is necessary to explain something about the technologies that make crowdsourcing possible and more efficient. Most of these technologies fall under the umbrella of Web 2.0, which is the latest generation of Internet technologies. Web 2.0 differs from Web 1.0 in that it is much more dynamic because it facilitates collaboration and sharing (Albors et al. 2008). According to Hung et al., it ‘emphasizes the construction of common, open, and friendly spaces for collective intelligence’ (2008, p. 345). The easiest types of participation are voting and rating, which have become increasingly commonplace online. They are used to get people’s opinion about certain topics and to structure content according to these opinions. Another easy way to participate is commenting on existing content, either from other participants or from the organization. Voting, rating and commenting are often used in combination with more complex forms of participation, like weblogs, forums and social networks. Weblogs (or just blogs) are user‐generated web journals, that people can fill with text and images to share their opinions and knowledge (Albors et al. 2008; Dearstyne 2007). Forums are another often used technology, where people can post questions or comments and can respond to each other’s posts, which are often categorized for easy searching. The difference between merely commenting and a forum is that participants are allowed to start their own discussions on forums, which is not the case when only commenting. Social networks offer people the opportunity to make online connections with people they already know from the offline world or with people they have never met before. People can become a member of a social network and maintain profiles, connect with each other, and interact. Often it also provides a platform for mini‐applications ‘from managing event invitations to playing Scrabble’ (Li and Bernoff 2008, p. 22). The final technology that will be described here is the wiki, which is a website that provides the opportunity for collaborative production, as it enables users to provide and edit content (Li and Bernoff 2008; Albors et al. 2008). Besides these typical Web 2.0 technologies, there are two more possibilities for the crowd to participate. Some organizations ask the crowd to upload their contributions, which is often combined with commenting. Finally, there is one type of contribution that is not related to the others, which is a financial contribution. In these cases, instead of contributing time or knowledge, the crowd is asked to contribute their money.
Party that takes initiative (crowd initiative / company initiative)
Although there is always an organization involved that must provide at least a platform for the crowd to work on, the extent to which the crowd can take initiative to start participating differs among cases. Therefore, this attribute shows the distinction between a company initiative and a crowd initiative. In a company initiative, the company must provide a specific request or challenge for the crowd to work on. The crowd is allowed to work on this task until a specific deadline, and then has to wait for the company to provide another task. This results in the crowd working in batches. In a crowd initiative, the activity is much more continuous. For example, the crowd is allowed to create their own topics on a forum or wiki, and start their own
21
tasks and discussions, giving the crowd much more control over the process. This distinction between batch and continuous work is also made by Füller and Matzler (2007) in the context of virtual customer involvement.
Whose contributions are used (individual / crowd / combination / p2p)
This attribute shows whether the contributions of the whole crowd are aggregated to be used as a whole or that one or a few contributions are picked from the crowd. Surowiecki’s (2004) work The Wisdom of Crowds only discusses the first option. In fact, one of the main arguments in his book was that it is better to use the contributions of a crowd than to try and find an expert, because a crowd of random people is smarter than the smartest people within it. In these cases, the contributions of the crowd are for example averaged, or their votes are used for decision making. Howe (2008, p. 16) however argues that ‘crowdsourcing is an immense talent‐finding mechanism’. By this he means that although in many cases the contributions of the whole crowd are aggregated, in many other cases, the initiative is set up as a competition, and one ‘winner’ is picked from the crowd. Besides these two extremes (which are labeled ‘crowd’ and ‘individual’), there are two other options. In many cases, for example when a forum or wiki is involved, the contributions of only part of the crowd are used, and multiple contributions are combined, which is labeled ‘combination’. In yet other cases, there is a market with peer to peer exchanges, which is characterized as ‘p2p’.
Financial rewards (yes / no)
This attribute shows whether there are financial rewards involved for the participants or not. This attribute has to do with the motivation of participants, an issue that is much discussed in related literature, mainly about online communities and open source software development. In these cases, participants generally receive no financial reward (Nambisan 2002; Füller et al. 2006; Amant and Still 2007), which has challenged researchers to find other sources of motivation. Although Howe (2008) concludes that crowdsourcing participants are not primarily motivated by money, many crowdsourcing cases use financial rewards. This financial reward can be a prize in a competition, a reward for being active, or a profit share. Although there are differences in the way the financial rewards are formed, this attribute only shows whether there is a possible financial incentive for participants.
Interaction (yes / no)
This attribute shows whether it is possible for the participants to interact with each other, and therefore shows whether the crowd can be seen as a community. This attribute is closely related to the previous one, as it is also related to the motivation of participants. Research into motivations other than financial was started in open source software literature, as many people did not understand and had not expected people to do all this programming work for free. Research into both open source software communities and other online communities found that one of the most important sources of motivation consists of community‐related benefits, oalso called social benefits (Bloem and Van Doorn 2007; Amant and Still 2007; Nambisan 2002; Bagozzi and Dholakia 2002). Chiu et al. (2006) found that community‐related outcome expectations are positively related to both the quantity and quality of contributions. Motivation can for example be related to reputation systems (Antikainen and Väätäjä 2008). By providing meaningful contributions to the community, the participants can enhance their reputation, which is often presented on the website in the form of a ranking. Although reputation systems are sometimes also used in cases without interaction, the effect is likely to be larger when participants feel that they are part of a community.
22
Output control (company / crowd)
This attribute represents the extent to which the organization actually gives the crowd the possibility to influence the organization. In the simple case of voting or rating, the organization can choose to follow the crowd’s opinion or decide to disagree with it. This attribute can also be important for the motivation of the crowd, as (the feeling of) control, or at least influence, can increase the motivation of participants, as this gives meaning to their actions (Howe, 2008). An organization can show its trust and commitment towards the crowd by following its opinion.
Howe (2008) argues that communities are the basic organizing force behind crowdsourcing and provide a context for the tasks, and thus meaning, which is considered an important motivator. However, the exploratory case studies show that not all cases really use communities, as it is not always possible for the crowd to interact with each other. Therefore, it is important to know in which cases there is really an online community or just a crowd.
There is another reason why it is important to know whether there is interaction between the participants. According to Surowiecki (2004), one of the most important reasons why a crowd can be wise, is diversity. However, independence and decentralization is needed to maintain diversity in a crowd. However, interaction decreases the independence of the participants, which implies that the diversity also decreases (Howe 2008). Since this could be a reason for companies to decide not to let the crowdsourcing participants interact, this attribute will
n r tprovide important information. I chapte six, his issue will be discussed in more detail.
What is important to mention here is that only financial benefits and community related benefits are included in the study as sources of motivation. Motivational issues are much researched in literature, and more sources of motivation have been found. However, during this study, some other studies were being conducted about motivational issues in crowdsourcing. Therefore, the choice has been made not to focus on motivation in this study. The reason that financial and community‐related benefits are included is that while some motivators can be seen as outcomes of the process, these two issues represent important design choices that cannot be excluded from the protocol. Furthermore, because participants are not included as direct source of information, these attributes only show that participants could be motivated by financial rewards or the opportunity for social benefits, not whether they really are.
Input control (contributions not public; type of participation; screening; registration)
This attribute shows the control that the organization has over the input of the contributions of the crowd. This control can be applied in different ways. First of all, the contributions can be shown on the website, where everyone can see them, or the company can choose to let the crowd submit their contributions in a more private way, so that only the company has access to it. Secondly, the type of participation also influences the need for control, as voting, rating or financial participation needs less control than more open contributions. Thirdly, the organization can choose to screen the contributions before posting them on the site. This shows important information about the extent to which the crowdsourcing organization is willing to give up some of its control. It must be said here that many cases do not perform screening before posting, but do allow moderation by removing posts that contain inappropriate content. This moderation will not be considered control in this study, as the control issue in this study mainly concerns interventions like removing content that speaks negatively about the organization or is off‐topic. A final type of input control is the need to register before participating, as it enables the company to ban participants when needed. This attribute is included in the study, because organizations are often worried about giving up part of the control over their organization (Hielkema et al. 2008). Therefore, the influence of input control on the quality of contributions is important information for the protocol.
23
Therefore, this attribute shows whether the organization commits itself to always following the crowd or whether the company still has the control to decide for itself how the contributions will be used.
Intellectual Property (company / crowd / public / not relevant)
This last design choice shows the distribution of intellectual property rights, whether the rights belong to the participants, the company (or to a third party in case of a mediator), or the public. The issue of intellectual property is much discussed in open source software and open innovation literature (Amant and Still 2007; Chesbrough 2003, 2007). Although crowdsourcing is often compared to open source software cases, there are some important differences, including possible financial rewards and the IP issue. Although some crowdsourcing cases show many similarities with open source software, in many cases the participants give up the IP rights to the organization, which makes money from it (Brabham 2008). In some way, this resembles the open innovation model, as it proposes selling licenses that are not used to other parties that have a better use for it.
The exploration of both theory and practice has led to the identification of these ten attributes on which a distinction of crowdsourcing types is likely to be based. However, before organizations can make a choice between the several options for each design choice, they need to know what options are relevant in their context. Therefore, the next paragraph will present two context attributes that represent the starting point for organizations to base their choices on.
4.2 CONTEXT ATTRIBUTES
The typology will be based on the clustering of the intervention attributes that were described in the previous paragraph. These types form the basis for the protocol. However, for an organization to be able to make the choice for a type, they need to know the relation with some of the characteristics of their organization and the objective they are trying to achieve with crowdsourcing. The two attributes presented in this paragraph form the context of the crowdsourcing initiative. They represent the context in which the relation between the intervention and outcome attributes are relevant. They will be discussed in chapter six, to determine which type and interventions are most relevant in which context. The first attribute is a distinction between the phases in the innovation process that crowdsourcing can be applied to; the second one is related to the type of task that is crowdsourced.
Phase (concept; development/design; testing; customer service; production; decision making; support)
According to Nambisan (2002), the design principles in an online community depend on the role of the participants, which can be resource, co‐creator, tester or in product support. Therefore, it is important to analyze these roles in the examples. The roles as described by Nambisan (2002) correspond to the phases in the innovation process. These phases are also defined by Brockhoff (2003), who uses idea generation, concept development, development and engineering, and prototype testing. However, these studies are only based on the involvement of customers. In crowdsourcing, more roles are possible, for example in production or for more specific one‐time projects or tasks for support functions. The phases that are defined in literature, together with the other relevant options, are translated into concept, development/design, production, testing, customer service, decision making and support.
24
Field (e.g. science, design, consumer goods, si , journal m
The second context attribute is the field that the task or company belongs to. Since crowdsourcing was first applied in software development, many critics argued that this would only work with software, because of its unique characteristics of an intangible product that is easy to share. But since crowdsourcing became more popular, it was applied to increasingly more fields, which raises the important question whether there are any limits to the fields to which crowdsourcing can be applied. Therefore it is important to investigate whether there are any cases in which it would not work, for example because organizations do not want to get the crowd involved in everything, or the crowd is just not capable of performing all possible tasks. Although it is not possible to be exhaustive in this attribute, the main field the crowdsourcing
mu c is )
activity is related to will be shown.
These two attributes provide the context in which an organization operates. Including these attributes therefore provides important information on whether certain interventions are relevant in the context of a particular organization. They are likely to influence the success of crowdsourcing, in terms of the outcome attributes that are presented in the next paragraph.
4.3 OUTCOME ATTRIBUTES
The following attributes represent the outcomes of the crowdsourcing initiative. They are measures for the success of crowdsourcing. The first two are related to the second research question, and shows information about the expected participants in crowdsourcing. The third outcome attribute is a more general measure of success. The fourth outcome attribute represents the fourth research question, namely the organization´s efforts. Although this can be seen as an input for crowdsourcing, it is also an output of certain design choices. Certain efforts are needed for certain interventions. Therefore, this attribute will be considered an outcome for the purposes of this study.
As already mentioned, these attributes have a more open character than the intervention attributes. There are two reasons for this. The first one is that not all documentation provided sufficient information to allow for a more structured approach. Secondly, because the objective of this study is to find relevant propositions, all information that might be relevant is collected, not posing too much limits on the data collection procedure.
Size and activity of crowd
An important outcome attribute is the size of the group of participants (whether a community or a crowd) and how the activities are distributed among the participants. This information is important because without sufficient (active) participants the crowdsourcing effort will not succeed. Studies of open source software communities have shown that most participants are only reading and not contributing themselves. As a general rule, only ten percent of participants respond to other people’s contributions, and only one percent actually contributes new content. Although this rule of thumb from open source software is not relevant for all crowdsourcing cases, it is likely that the distribution of activity is similar in some crowdsourcing examples. This attribute therefore shows the total size of the crowd, and for which cases relevant and available, the distribution of activity of these participants.
25
Composition p rticipant
This attribute considers several characteristics of the crowd, such as their professional background, which country they are from, and their ages. It is also shown whether there is a specific group of participants that responds to the call. As was stated in the definition, crowdsourcing makes an open call. The fact that everyone is invited does however not mean that there can be no specific group responding to the call. It also does not mean that the organization is not allowed to promote the initiative to a specific target group. This attribute is directly related to the second research question, which investigates the expected participants in crowdsourcing.
a s
Number and quality of contributions
This attribute shows the extent to which the crowdsourcing initiative is considered a success. This is mainly measured in the number and quality of contributions, as a sufficient amount of contributions is considered a necessary condition for success, but also quality is important. Quality is as much as possible measured quantitatively, for example by the number of contributions that have resulted in concrete projects. Because this information is not available for most cases, more qualitative information on the success of the initiative as mentioned in the documentation is also included in this attribute. The reason that not much information on the success of the cases is available is mainly because many cases are still in the start‐up phase and are mostly experimenting with the crowdsourcing model. This does not only mean that not much information on outcomes is available in general, but also many initiatives do not have concrete expectations of the results., which makes it difficult to measure success.
Company effort
This attribute is directly related to the fourth research question and collects all information on activities that need considerable resources from the organization. This attribute focuses both on the activities that need resources and the amount of resources. The reason for this is that limited information is available on the amount of resources, and it is likely that this will differ across organizations. Information on the activities that need resources allows organizations to decide for themselves how much resources are needed for this activity in the context of their own organization.
According to Howe (2008), the hardest part of crowdsourcing is to achieve critical mass and sustain the volume and activity of the crowd. Therefore, it is interesting to investigate what organizations have done to achieve this critical mass and to keep the crowd active. Obviously, many cases involve much advertising and promotion activities, but there are also other issues that are important. It is for example possible to organize some specific event as introduction to a more continuous crowdsourcing initiative.
Besides initiating contributions, also retention of participants needs effort. According to Nambisan (2002), providing explicit knowledge to the community about for example relevant products and technologies can help the community in its contributions. Active participation also seems to be an important success factor in many examples. This also includes giving feedback about what is done with contributions, again because this can give meaning to the activity of the crowd.
The four outcome attributes provide important information for chapter six, where the research questions about the expected participants, the conditions for crowdsourcing to work and the efforts that are needed from the company are discussed.
26
4.4 CONCLUSION
In this chapter, the attributes that will be used for the analysis of the research questions were described. The attributes are organized into different categories, including intervention, context and outcome attributes. This makes sure the attributes can be used to develop relevant design propositions according to CIMO‐logic that can be applied in the protocol. The context and outcome attributes will be used in chapter six to develop the design propositions, and in chapter seven, where these will be applied in the form of a protocol. But first, the intervention attributes will be used in the next chapter, where a typology for crowdsourcing is developed.
27
5 CROWDSOURCING TYPES AND THEIR CHARACTERISTICS
In this chapter, the first research question Which types of crowdsourcing exist and how can they be distinguished? is discussed. This question is answered by matching the patterns in the intervention attributes to the only typology of crowdsourcing that is available at the time of writing. Howe (2008) has distinguished six types of crowdsourcing, which will be described in the following paragraph. However, these types are only described briefly in his work and it is not clear how these types should be distinguished, because they are mostly explained through illustrating examples. Actually, Howe himself acknowledges that they are ‘neither compre‐hensive nor fireproof’ (p. 279) and that there is overlap between these types. Therefore it is useful to examine these types in more detail and investigate whether they can be explained by the intervention attributes presented in the previous chapter. The data on these attributes can be found in appendix A. For the examples that are mentioned in the text, also a reference to their website is added. This chapter will start with a brief explanation of the types as Howe describes them. After that, it will be explained how the data is analyzed. Finally, a typology is presented based on this analysis, which can be considered an improved version of the typology of Howe.
5.1 TYPOLOGY OF HOWE
In this paragraph, the types of crowdsourcing as defined by Jeff Howe (2008) are described. He distinguishes four main types, which are crowd wisdom, crowd creation, crowd voting and crowdfunding. Crowd wisdom is divided into three subtypes, namely prediction markets, crowdcasting and brainstorming. This means that there are six types in total that are explained in this paragraph.
Prediction markets
Prediction markets are organized like stock markets, which instead of shares uses possible outcomes to invest in, which can be for example the winner of an election or the Oscars. Many companies also use prediction markets internally, allowing their employees to trade in outcomes such as inventory, sales goals, and manufacturing capacity, thereby crowdsourcing the decision making process. The possible outcomes are presented as stocks, and participants can buy and sell stocks according to their knowledge about the outcome. The outcome with the highest value is considered the most probable outcome. Prediction markets make use of large, diverse networks of people who are not experts, but often possess unique knowledge. Generally, prediction markets do not represent a community, as interaction between the participants could bias the outcome. Because of the monetary incentive, people are motivated to release their private knowledge and to trade as soon as possible, which causes information to disperse very quickly. Unlike polls, the ‘foolish’ also have an incentive to stay away, as they are likely to lose money. The markets outperform the experts because collectively the people trading in them have access to far more data. However, all but one case of prediction markets do not have a financial incentive, because it is considered gambling and therefore illegal. Since prediction markets resemble gambling and the results of prediction markets are not used by any organization, the prediction market type will not be considered crowdsourcing and is therefore not included in the study.
28
Crowdcasting
In a crowdcasting network, someone with a problem broadcasts it to the widest possible audience in the hope that someone will solve it. The most typical example of crowdcasting is InnoCentive, a platform that opens up scientific problems to a crowd of about 160,000 people who are interested in science and compete for a prize (appendix B.2). Since an open call is made to a large and diverse crowd, the chances of finding a solution are much better than with more traditional approaches.
Idea Jam
This is essentially just a massive, online brainstorming session, which Howe calls ‘idea jam’, after the initiative of IBM (Bjelland and Wood 2008). The idea jam is close to crowdcasting, with the only major difference being that the call for submissions in an idea jam is much more open‐ended. Instead of attempting to solve a particular problem, solutions are created for problems that do not yet exist, by allowing the crowd to discuss whatever topics they are interested in. Generally a forum is provided where people can discuss current products and provide ideas for future products.
Crowd creation
Howe (2008) defines crowd creation as crowdsourcing the activities that use the creative energy of participants. Crowd creation is in many ways similar to user generated content, although crowdsourcing generally involves building a business around it. As opposed to crowdcasting and prediction markets, interaction between participants seems crucial, as crowdsourcing creative work generally involves a tight community with a deep commitment to the task and each other. This is also needed because financial incentives are generally absent. The community aspect also increases the quality of the work, as participants are striving to enhance their reputation.
Crowd voting
In crowd voting, the crowd is given an opportunity to express their opinion through voting or rating. This important information can be used by companies for decision making. This does not only make sure the opinions of important stakeholders are included in the decision, but also shows them the company’s commitment and trust. Furthermore, involving the crowd often implies going through all the contributions they provide. Crowd voting can be used to manage this time consuming task, as the crowd can be used to structure all this information. Therefore, crowd voting is often combined with other types of crowdsourcing.
Crowdfunding
The final type is crowdfunding, which instead of using the crowd’s spare time or talents, uses their spare money. Three typical examples of crowdfunding will explain the concept. First, the peer to peer micro lending site Kiva2, where people can directly lend money to small businesses in the Third World. The second example is Sellaband3, where the crowd can provide funding for upcoming artists, who can make an album when enough money is collected. Third, the crowd can fund a whole company, as is the case with MyFootballClub4, where the crowd gets a vote in ll important decisions in exchange for their money. a
29
5.2 ANALYSIS
In this paragraph, the typology of Howe is compared to the information of the case studies. First of all, the types of Howe are briefly analyzed, followed by a more extensive analysis of this study’s data. From this analysis a new classification of crowdsourcing is developed.
Before starting the analysis of the data in this study, a short analysis of the typology of Howe is appropriate. First of all, the approach of Howe is not scientific, as he mostly uses examples to illustrate the various applications of crowdsourcing, presenting the types only briefly in his conclusion with little argumentation for his choice. Secondly, the types are distinguished according to the types of activity of the crowd, namely using its wisdom, creativity, opinion, or money. As Howe acknowledges himself, this causes overlap and ambiguity between the types.
Furthermore, the distinction between subtypes is not really clear. For example, the only difference between crowdcasting and the idea jam that is mentioned is the more open request of an idea jam. However, there seem to be other, more implicit, differences, like the use of financial rewards and the possibility of interaction between participants. While these
mcharacteristics seem to differ in the examples that are given, they are not entioned explicitly.
These characteristics are important for organizations that want to apply crowdsourcing. Because the ultimate objective of this study is the development of a protocol, the implications for organizations are the main focus in this study. Therefore, the new typology will be based on more structural characteristics instead of the content of the activity. This will provide a clearer distinction between the types that is more relevant for the protocol.
The typology of Howe will be projected on the data that was collected of 115 crowdsourcing cases. This data can be found in appendix A. The new typology should not only provide more insight into crowdsourcing in general, but the types should be facilitating the development of the protocol. Therefore, the types should represent different business models that are possible in crowdsourcing.
The first step is to translate the descriptions of Howe into the intervention attributes as defined in the previous chapter. Although it is possible to start this analysis at any type, crowdcasting and the idea jam are the first types that will be discussed.
According to Howe, the main difference is the more open request of an idea jam compared to crowdcasting. This characteristic can be translated into the extent to which participants can take initiative. Because crowdcasting implies that a specific request is broadcasted to the crowd, it is considered a company initiative, because the participants depend on the company, which determines how and when the crowd is allowed to participate. Because an idea jam allows the crowd to suggest new ideas without a company specifically asking them, the idea jam is
considered a crowd initiative.
However, there seem to be more implicit distinctions between the two. For example, in crowdcasting only one solution to a problem must be found. In contrast, in an idea jam the request is more general, allowing for multiple contributions to be used. Often, the results arise from the discussions between participants, which means a combination of the crowd’s contributions are used. When translated into the intervention attributes, in crowdcasting, the contribution of an individual participant is used, while in the idea jam, a combination of contributions is used.
30
form.
After these four types, namely crowdcasting, crowdstorming, crowd production and crowdfunding, were identified, 14 cases remain unclassified. The classification can be seen in table 1, where the types are shown according to their characteristics as explained in this paragraph. The number of cases for each type are shown between brackets. The remaining cases represent hybrids between types, as they have characteristics of more than one type. There are two types of hybrids, as can be seen from table 1, representing combinations of crowdcasting and crowdstorming. Hybrid 1 resembles crowdcasting in that the contributions of individuals are used, but they are different because they do not include a specific request or a deadline. An example is the company Fluevog
This distinction, based on the party that takes initiative and whose contributions are used, results in 31 crowdcasting cases and 16 idea jam cases. However, since the term idea jam is directly derived from the IBM case, which in the new typology is not a typical case of this type, there is a need for a change in terms. The cases where the crowd can take initiative and a combination of contributions is used are therefore named crowdstorming, formed by combining crowdsourcing and brainstorming. It is important to realize that the definition of these first two types already implies a deviation from Howe’s typology, since creative tasks are included in both crowdcasting and crowdstorming. Therefore, crowd creation as defined by Howe cannot be used to determine the new typology. Since crowd creation is not directly translatable to the attributes that are used in this study, this analysis will continue with crowd voting.
The characteristics of crowd voting can also be translated into the intervention attributes, because the contributions of the whole crowd are used, and the company should provide something to vote on, implying a company initiative. However, looking at the cases that conform to this description, all of these cases involve crowdfunding. While some cases combine crowdfunding with crowd voting, others are pure crowdfunding. This is not surprising, as funding also requires a company to provide something to fund and the money of the whole crowd is used. Crowd voting will therefore not be considered a separate type, but can be used in combination with the other types. Whether this is the case can be seen in the attribute ‘type of participation’. The cases that are considered a company initiative where the contributions of the whole crowd are used are classified as crowdfunding. However, there are more crowdfunding cases, where instead of combining the funding of the whole crowd, the contributions are used in a p2p form. These will also be included in this category, resulting in a total of 15 cases. The two
nsubtypes will be distinguished as ‘product’ a d ‘market’.
Until now, the categories are determined by two attributes, namely which party takes the initiative, and whose contributions are used. Looking at the categorization according to these attributes, most of the remaining cases (33) involve a crowd initiative that uses the contributions of the whole crowd. In all these cases, the crowd provides content for a website or some database. Since this is not related to innovation, in contrast with crowdstorming and crowdcasting, this type will be named crowd production. Similar to crowdfunding, there are also cases that have similar characteristics, but act as a market instead of using the whole crowd for one product. Most of these cases represent microstock photo sites like iStockphoto (appendix B.4). Similar to crowdfunding, these cases are considered a subtype of crowd production, distinguishing between ‘product’ and ‘market’ cases. Its characteristics are that the crowd can take the initiative to start participating, while the contributions are used in a p2p
5, which allows the crowd to contribute shoe designs on a continuous basis, and just occasionally picks some interesting ones that go into production. Hybrid 2 resembles crowdcasting in that specific requests are frequently provided, but that more than one contribution is used or that the result of a discussion is used. Examples are Nabuur (appendix B.5) and MijnProefstation (appendix B.6). These hybrids are not considered separate types, as they are too similar to crowdcasting and crowdstorming.
However, including them in one of these types would make the types more inconsistent, resulting in difficulties in developing the protocol. Therefore, the hybrids are not included in the remainder of this study.
Attributes Whose contributions are used
Individual Combination Cro pwd/p2
Party that takes
initiative
Crowd initiative
Hybrid 1
(5)
Crowdstorming
(16)
Crowd production
(product: 33; market: 7)
Company initiative
Crow ting dcas
(31)
Hybrid 2
(9)
C rowdfunding
(product: 10; market: 5)
31
5.3 CROWDSOURCING TYPES
TABLE 1: CATEGORIZATION OF TYPES
In this paragraph, the new typology will be further explained. The types are categorized according to only two of the ten attributes, namely whose contributions are used and the party that takes initiative. According to these attributes, four types were identified. In this paragraph, the types will be analyzed further, describing other characteristics that are common for each type. The results are summarized in table 2. From the table can be seen that although the classification is based on only two attributes, this has implications for the other attributes. This means an internal consistency exists within each of the types. This internal consistency confirms that this typology is useful for the protocol, as not many choices need to be made within a type.
Crowdcasting
The sample of crowdcasting examples includes 31 cases. Generally, crowdcasting involves competitions where individuals can upload their submissions and compete for a financial reward (26 cases). The winner is chosen by the organization that broadcasted the challenge, who also receives the rights of the winning contribution. In many cases (15) the contributions are not published on the website. Although this is true in only about half of the cases, this characteristic is not present in any of the other types, which makes this an important characteristic of crowdcasting. In these cases, there automatically is also no interaction between participants, since there is nothing to interact about. For this reason, a simple upload possibility is sufficient in crowdcasting. Although a forum or commenting on these submissions is a possibility in 14 of the cases, the focus is on the individual submissions, which implies the community aspect of these crowdsourcing cases is different than for example in crowdstorming cases, where the comments are much more important for the outcome. The community aspect is related to the competitive form of crowdcasting. Because individual contributions are rewarded,
32
Type
Role of the company
Type of participation
Party that takes
initiative
Whose contributions
are used
Financial rew
ards
Interaction
Input control
Output control
IP
Crowdcasting mediator/
own ini tiative
upload company in itiative individual yes yes/no
contributions not public; registration
company company
Crowdstorming own ini tiative forum crowd
in itiative combination no yes registration company company/ public
Crowd production
product own initiative
weblog; wiki
crowd in itiative crowd no yes/no registration crowd public
market mediator upload; forum
crowd initiative p2p yes yes/no registration crowd crowd
Crowd funding
product mediator/
own initiative
financial company initiative crowd yes no
registration; type of
participation crowd not
relevant
market mediator financial crowd initiative p2p yes no
registration; type of
participation crowd not
relevant
TABLE 2: CHARACTERISTICS OF TYPES
it could be considered unfair to allow interaction between the participants, because one could benefit from other contributions. Because the participants do not need to interact and the contributions are not published on the website, the organization does not have to worry so much about controlling the activities of the crowd. Although registration is generally needed, this is only to be able to make the necessary payment, and is not used as a control mechanism.
Crowdcasting can be organized by a mediator (14 cases) or the organization itself. The reason to choose a mediator is that when the company would organize their own initiative, they would need to develop new challenges on a regular basis. It is likely that many companies have some challenges that are interesting to broadcast to the crowd, but cannot provide enough of them to keep participants coming back. This is a good reason to use a mediator, since they broadcast challenges of several companies, and can therefore provide sufficient activities for a larger crowd. The cases where the company organizes the initiative themselves all represent one‐time or periodic competitions.
Generally, the number of contributions per contest is manageable. For example, the average number of contributions in Battle of Concepts is 47 (appendix B.7), in CrowdSpring6 it is 69, and in 99designs7 72 (appendix A). Therefore, only five of the cases include crowd voting, as there is generally no need to structure the contributions. Crowd voting can also be used as a reputation system, as the number of votes can be seen as an indication of the quality of the contributions. Furthermore, crowd voting could be used to give the participants a sense of control over the results. However, in all cases, the company still has the final vote, and the participant’s votes are used to determine a short list, as is the case with Google 10100 (8), where more than 100,000 contributions were submitted. Typical examples of crowdcasting are InnoCentive (appendix B.2) and Battle of Concepts (appendix B.7).
Looking back to the literature study, crowdcasting mostly resembles open innovation, since it is specifically designed to outsource part of the R&D function. Like in open innovation, the role of the internal R&D department changes and new roles need to be created. Crowdcasting is an extreme form of open innovation, since the search for external innovation is much broader than is described in literature.
33
Crowd production
In total, 40 cases of crowd production were included in the study. As explained before, the ‘product’ cases represent initiatives in which the crowd produces something together, for example a database that is used for research, or content for a website. An important category in this type is citizen journalism, which allows amateur journalists to create and structure articles
Crowdstorming
16 cases of crowdstorming were identified. As opposed to crowdcasting, in crowdstorming forums are the main method for participation. This allows the participants to take much more initiative, because they can start discussions and propose new topics, not being dependent on the organization. Because the discussion often refines the original ideas that are posted, a combination of the contributions is used by the crowdsourcing company.
None of the cases include financial rewards, as the discussion between participants is often more important than the individual ideas. Because of this interaction, the participants form a community, which creates social benefits for the participants. Because crowdstorming allows participants to contribute anything they want, organizations often worry about losing control. However, many companies realize that forums are only useful and credible when there is transparency, and when participants are allowed to communicate their negative feelings about the company or its products. The only control that remains is the ability to ban participants, as registration is needed. There is only one case that does not need registration, but in this case all contributions are screened before they are posted on the website. Although most cases do not require screening beforehand, most cases include some way to notify the company when contributions are not appropriate, which enables the company to remove posts timely when needed.
There are two cases that act as mediators, namely IdeaScale9 and GetSatisfaction10. All other cases have integrated the activities on their own website. As will be discussed in the next chapter, the participants in crowdstorming are generally customers, which are encouraged to discuss about current or future products. Therefore, crowdstorming enables customers to have a more active role in product development compared more traditional methods, where customers rarely offer new product ideas without being prompted by a company (Nambisan 2002). Although sometimes the distinction is difficult to make, the initiative can focus on new product ideas or on support for current products. This has implications for the output control and IP, as for future products, the company decides whose contributions to use and how and generally claim IP rights on the ideas, while with a customer support forum, the customers themselves can decide how to use the information and IP is not an issue.
When the community that revolves around these forums is growing, it becomes harder to identify the most relevant ideas and discussions. Therefore, crowd voting is often used in crowdstorming, letting the crowd show which contributions they think the company should pursue. Although the company makes the final decisions, it shows commitment to the community when they implement the ideas that receive the most votes, as is the case in Dell’s IdeaStorm (appendix B.3). This case is one of the typical examples of crowdstorming. Because customers are generally the main target group for crowdstorming, most companies integrate
ithe forum on their own website, as it is often also considered an important market ng tool.
Crowdstorming mostly resembles the findings in the literature about online communities, although also user driven innovation literature provides important information that is relevant to this crowdsourcing type. As in literature, most crowdstorming initiatives are customer communities, designed to provide customer service or to generate new product or service ideas.
34
that are used to fill a website, or even a newspaper or a TV station. While in crowdcasting and crowdstorming the crowdsourcing companies still have the major voice in how the contributions are used, crowd production generally allows the crowd to have much more control. Although the cases where a database is created the company still controls the results, where the crowd decides a website’s content, the crowd is usually in control of the output (25 cases). The only input control that is used is registration.
The most well‐known example in this type is Wikipedia (appendix B.8), the user generated encyclopedia. Wikis and weblogs are considered common types of participation in crowd production, as they are not used in other types of crowdsourcing, but also forums and uploads are frequently used. Crowd voting is often used to structure the content, as the volume of contributions can be substantial. None of these cases include financial rewards, which is not surprising since the contributions of all participants are used equally. However, what is remarkable is that eleven of these cases also miss the possibility of social benefits, as there is no interaction. This implies that other motivations are important besides social and financial benefits. This will be further discussed in the next chapter.
In the ‘market’ cases, the crowd provides something that can be traded among the crowd, like photos. While all ‘product’ cases are a company’s own initiative, the ‘market’ cases mediate among participants. In contrast with the ‘product’ cases, there are financial benefits involved. Because the financial benefits are not dependent on a competition, the participants form a community where important knowledge is shared, which is for example the case in iStockphoto (appendix B.4).
Open source software development can be seen as the first application of crowd production, sometimes also called peer production in this context. Therefore, literature provides important information that is relevant to crowd production, as can be seen in the next chapter. As in open source software, the focus is on value creation as opposed to value capturing. Only recently, business models started to emerge to capture the value of open source software. It is expected that this will also be the case in crowd production cases. For example, iStockphoto also started out merely creating value for the community, while only later on an approach was found to capture this value (appendix B.4). Similarly, Wikipedia is about creating value, but since it is so successful and so many people make use of this online encyclopedia (appendix B.8), it is likely to someone is going to find a way to create a business model around it.
Crowdfunding
This type includes 15 cases in total, of which 5 represent a market between borrowers and loaners, and 10 represent possibilities to fund a project or company. The ‘product’ category actually represents two different types of funding, to be distinguished by mediator versus own initiative. In the case of a mediator, the crowd can choose to fund for example an artist (Sellaband3), a journalist (Spot.us11), or an entrepreneur in a Third World country (Kiva2). In the case of an own initiative, the crowd is asked to invest in a new company, generally in exchange for a vote in important decisions the company has to make, as is the case in for example MyFootballClub4.
Crowd voting can be combined with crowdfunding in several ways. First of all, participants can receive a vote in important decisions in exchange for their funding. Second, voting can be used as a requirement before starting fundraising, like with Bandstocks12. Although voting in combination with the other types is used to give the participants a sense of control, in crowdfunding they actually have control, as the votes are generally followed by the company.
35
Crowdfunding is not directly related to any of the concepts that were described in the literature study. It emerged mainly as a way to get people involved in decisions that were used to be made by large corporations only. Many examples have shown that that the crowd is perfectly able to decide who or what deserves funding and even to manage an organization by itself. The limits of this business model are not in sight yet.
5.4 CONCLUSION
In this chapter, a new and improved typology for crowdsourcing has been developed. The initial typology of Howe has been useful as a starting point for the analysis. However, the distinction between the types based on the type of activity resulted in overlap and ambiguity.
The improved typology is based on differences in two attributes, namely the party that takes the initiative and whose contributions are used. This has resulted in four types, namely crowdcasting, crowdstorming, crowd production and crowdfunding. It was found that although the distinction is made based on only two types, there is an internal consistency within the cases of a particular type. This is a confirmation that this typology is useful for the protocol, since this significantly limits the amount of choices that organizations have to make when they want to apply crowdsourcing.
The first type is crowdcasting, in which a particular challenge is broadcasted to a crowd, generally organized as a competition with a financial reward. An example of this type is InnoCentive (appendix B.2)
The second type is crowdstorming, which involves an online brainstorming session, where interaction between participants is important. In many cases, this involves a company asking the crowd for ideas. An example of this type is Dell’s IdeaStorm (appendix B.3)
The third type is crowd production, in which the crowd creates a product or database together or creates a market of individual contributions. Typical examples of this type are Wikipedia (appendix B.8) and iStockphoto (appendix B.4).
The fourth type is crowdfunding, where instead of the spare time, abilities and knowledge of the crowd, their spare money is used. The crowd is for example used to fund artists, companies or each other.
These four types do not cover all crowdsourcing cases, since a few cases are found to be hybrids. Although these hybrids are not included in the remainder of the study, they should not be forgotten. It is important to study these cases in future research, since they too can provide important information on the possibilities of crowdsourcing and how to organize a crowdsourcing initiative.
Furthermore, although crowd voting is not considered a separate type, it is often used in combination with these four types, especially in crowdstorming and crowdfunding. Whether this is the case can be seen in the attribute ‘type of participation’. The influence of crowd voting on the crowdsourcing initiative will be discussed in the next chapter.
While the typology is important for structuring the protocol, not much content can be provided at this stage of the study. Therefore, before the protocol can be developed, the remaining three research questions need to be answered. This is done by developing propositions that have proved to be relevant in both theory and practice, and are presented in the next chapter.
36
37
6 BUILDING PROPOSITIONS FOR CROWDSOURCING
In this chapter, the remaining three research questions will be discussed. The literature study from chapter three will form a basis, together with management literature and the data from the cases that were studied. This data can be found in appendix A. Since the theoretical concepts are in most cases not relevant for crowdfunding, the focus will be on the other three types.
For each research question propositions will be developed. As explained in chapter two, these propositions need further testing in the future to be considered valid answers to the research questions. However, they are initially tested through the data that was collected in this study. The propositions are descriptive or explanatory (Cooper and Schindler 2003). The descriptive propositions consider only one concept, while the explanatory propositions describe the relation between two or more concepts. Because the ultimate objective of this study is to develop a protocol, the explanatory propositions are used and formulated as design propositions (in contrast to theoretical propositions). According to Van Aken (2007a) and Denyer et al. (2008), design propositions should be formulated according to CIMO‐logic, including a context, intervention, mechanism and outcome. The components of these propositions are chosen according to the different categories of attributes shown in chapter four. The intervention attributes that were used to develop the typology in the previous chapter will also be used for the propositions. The outcomes that were measured are the number of participants, and the quantity and quality of contributions. However, because the study has not identified relations between interventions and specific measurable outcomes, the outcome that will be used in the propositions is the more general ‘probability of success’. Besides the two context attributes that are described in chapter four, the crowdsourcing types are also used as context variables. However, as can be seen later on in this chapter, it is not possible to formulate all explanatory propositions according to CIMO‐logic, because in some propositions there is no information about relevant interventions. These propositions are nevertheless relevant for this study and for the protocol. For each proposition it is shown whether it is descriptive, formulated according to CIMO, or just includes CMO.
6.1 TYPES OF PARTICIPANTS IN CROWDSOURCING
In this paragraph, the second research question Which type of participants are expected in crowdsourcing? will be discussed. The answer to this question consists of propositions, which are initially tested through the current case studies, but need further testing in future research. The issues that are discussed in this paragraph are the diversity of the crowd, the abilities of the crowd and the types of activity that can be expected from the participants.
Diversity
Because crowdsourcing only concerns online initiatives, the most general definition of the crowd is the total Internet population, which includes about one billion people (Howe 2008). In 2007, approximately 230 million people worldwide were active in online communities (Janzik and Herstatt 2008). A study from the Pew Internet and American Life Project shows that more than a third of all US Internet users have posted something on the web (Lenhart et al. 2004). But more importantly, it is shown that the online population increasingly represents the offline population (Hof et al. 1997). While a few years ago, mainly young, white males with a specific interest in technology were engaging in online activities, currently the people posting content
38
online come from every income level and ethnicity, making the total online population more diverse.
Proposition 1: The diversity between the participants will increase in the future (descriptive)
There is also diversity in geographical location. Although some crowdsourcing cases focus on a particular geographical location or have a limited audience because of language constraints, many online communities represent the global community. In crowdcasting, the diversity in geographical locations is evident. InnoCentive for example has participants from more than 175 countries, only one third coming from the US (appendix B.2). The first edition of the Peugeot design competition13 had more than 2000 submissions coming from 80 nationalities. Although not all cases include information about the representation of countries, the participants of MechanicalTurk14, Netflix15, and CrowdSpring6 all come from more than 100 countries. The crowdstorming cases unfortunately do not report about the composition of the participants, except for MyStarbucksIdea16, where 75 percent of participants are from the US. Crowd production initiatives like Marketocracy17 (130 countries), NowPublic18 (140 countries), Wikipedia (appendix B.9) (250 languages), WeBook19 (100 countries), and A Million Penguins20 (130 countries) clearly show a diverse audience. In crowdfunding cases, a variety of nationalities are represented, but generally less than with crowd production or crowdcasting. For example, Sellaband3 has both believers and artists from 60 countries, BeerBankroll21 has participants from 5 countries, and in MyFootballClub4, 75 percent of participants is from the UK.
Diversity is one of the major strengths of a crowd and thus of the application of crowdsourcing. One of the reasons is that decentralization and independence, which both influence diversity (Surowiecki 2004), are easily found in a crowd. This is possible because a crowd mainly consists of weak ties, which means there is not much contact between the members, and the amount of redundant knowledge is limited (Granovetter 1973). According to Surowiecki (2004), diversity is an important condition for the crowd to be wise. Because many of the examples from his book are closely related to crowdsourcing, the principles from the wisdom of crowds are likely to be relevant for crowdsourcing.
While nationality contributes to diversity, there are other measures that are important. Surowiecki mainly mentions diversity in the background of people and their knowledge as important measures. As Brabham (2008) argues, there are different types of diversity, such as diversity in identity or skills. Which of these is most important for crowdsourcing is not yet known and is the reason that several measures are used in this study, also depending on availability. It is believed that diversity in nationality contributes to diversity in personality, which accounts for different perspectives on the broadcasted questions and challenges. Some cases provide important information to indicate the importance of diversity of background and skills, which are discussed in the next section.
The data shows that diversity is mainly important in crowdcasting, as was also argued by Page (2007), who found that diversity is important in problem solving. The case that is most researched concerning this topic is InnoCentive (appendix B.2). Research has found that most problems were solved by people on the boundary or outside the scientific field of the problem. This is also shown by more traditional examples, such as a sculptor who invented the ballpoint pen, a journalist who invented the parking meter and the veterinary surgeon who invented the pneumatic tire22. Also in the GoldCorp challenge, diversity proved to be successful. While many of the participants were geologists, there were also students, consultants, mathematicians, and military officers who provided insightful analyses of the data because they used methods that GoldCorp had never thought of before (appendix B.1). With Battle of Concepts, the diversity issue is shown in two ways (appendix B.7). First of all, because none of the participants, which
39
Ability
Although in some cases diversity trumps ability (Surowiecki 2004; Page 2007), in many cases ability is still important. However, one of the criticisms on crowdsourcing is that the crowd is lacking ability for some tasks, because they are amateurs. However, the term amateurs is often accompanied with characteristics such as not receiving payment or not having the required knowledge and skills, which is not always the case in crowdsourcing. Therefore, crowdsourcing requires different definitions of the concepts amateur and professional. An insightful work is
are students and young professionals, belong to the field of expertise of the battle, they represent a diverse group compared to the people the organization would traditionally involve. Because they are relatively young and inexperienced, they often can provide a distinctive view on problems. Moreover, because the participants all represent different studies, which represents differences in knowledge, they can offer diversity in the concepts they submit. The importance of diversity is shown by the example of an organization that placed a battle, but had already brainstormed about the issue themselves. Nevertheless, there were some concepts that were totally new, which all were written by participants from different studies. Therefore,
diversity is likely to be important for crowdcasting.
For crowdstorming, diversity seems less of an issue, since crowdstorming occurs in communities consisting of people with common interests. This is probably also the reason that not much information is mentioned about the composition of the participants in documentation. Often a specific group is targeted for participation. For example, most of the crowdstorming cases included in this study represent customer communities. Marketing objectives are important in this case, often more than using the information the participants provide. Whether this information is provided by a diverse group is not so much an issue, as long as this is recognized by the company (Li and Bernoff 2008).
In crowdfunding cases, diversity also does not seem to be an important issue, which is not surprising since it does not matter who is providing the money.
In crowd production, diversity is mostly important for self‐organization. When the crowd is more diverse and has different interests, different types of tasks are done and more work can be performed using more knowledge. Therefore, diversity seems to be an important condition in crowd production. However, in some cases all tasks are the same, for example in GalaxyZoo23 and Clickworkers24, which makes diversity unnecessary. However, because not much information is available about the diversity of participants in crowd production, this is an important topic for further research.
Surowiecki (2004) found that independence and decentralization are important conditions to maintain diversity. Although crowdsourcing generally is decentralized, the participants are not always independent. Although the initiative starts with a crowd connected by weak ties (Granovetter 1973), when participants are able to interact, and when they form a community, the ties become stronger, and independence is decreased. Therefore, diversity is decreased, because the community becomes increasingly homogeneous (Howe 2008). Although interacting and sharing knowledge would make the individuals in the crowd smarter, according to Surowiecki (2004) it would make them collectively dumber. Because most crowdsourcing cases are still in an early phase, there is not much information available about this effect. Therefore, further research is needed to investigate whether this relation is really present and whether it affects crowdsourcing cases with a community negatively.
Proposition 2: In crowdcasting and crowd production, interaction decreases the diversity of the crowd, which has a negative influence on probability of success (CIMO)
40
Proposition 3a: Many of the expected participants are proams (descriptive)
The importance of professionalism differs between the crowdsourcing types. The types of crowdsourcing can be placed on the continuum from amateur to professional, not only considering the extent to which knowledge is important, but also the commitment that is needed to perform the task. For crowdfunding, neither of these is necessary, this can be done by amateurs. Because crowdstorming generally asks for the crowd’s opinions instead of knowledge, participants can be considered amateurs. However, because engaging in crowdstorming requires more effort and commitment than crowdfunding, crowdstorming is placed somewhat further towards professionalism. Leadbeater (2004) identifies two levels of pro‐ams, one being the more serious and committed amateurs, the second being quasi‐professionals, representing pre‐, semi‐, and post‐professionals. While there are exceptions, from the cases can be concluded that crowdcasting generally requires knowledge and skills that belong to the quasi‐professionals, while crowd production generally only needs the first level of pro‐ams. This scale can be seen in figure 2, showing the continuum as developed by Leadbeater in the lower part, and the crowdsourcing types in the upper part. Although the amount of
presented by Charles Leadbeater, called The ProAm Revolution. He recognizes like Toffler (1980) that a new era is emerging after the industrial revolution. However, instead of using the term prosumers, a combination of producers and consumers, Leadbeater uses another perspective and focuses on the characteristics of these prosumers. According to Leadbeater (2004) they are working increasingly to professional standards, and have the dedication and commitment associated with a professional, but are likely to earn little or no money from these activities and pursue them mainly for the love of it. Because this provides a new view on the meanings of amateur and professional, he uses the term pro‐am to indicate that there are several levels of amateurism. Some examples will illustrate the need for this new definition. While InnoCentive’s solvers are all amateurs in the traditional sense of the word, amateurs are generally not considered to be able to solve problems that the world’s largest R&D departments cannot. In fact, more than a third have doctorates, but they do not consider working on the problems their main job (appendix B.2). While 96 percent of contributors to iStockphoto are amateurs in the sense that they do not consider photography their job, more than half have had at least one year of formal schooling in some related field (appendix B.4). In his work, Leadbeater argues that while the twentieth century was shaped by large hierarchical organizations with professionals at the top, the next two decades will be influenced by pro‐ams. Crowdsourcing acknowledges the existence of pro‐ams, and gives them the opportunity to perform their hobby in a more professional way. Because crowdsourcing does not consider credentials, and is often organized as a meritocracy, they are given the opportunity to develop and show their skills. That there is no need to attract professionals in crowdsourcing, but that pro‐ams are the target group that makes crowdsourcing successful, is shown by both theory and practice. Kristensson et al. (2004) and Magnusson et al. (2003) found that amateurs come up with more original and valuable ideas than professionals, whose ideas are more feasible, but less original and valuable. Robertson (1984) also argues that creativity is not linked to intelligence, and while some cases have a need for intelligence, creativity is often very important in crowdsourcing. And indeed, the quality of the crowd’s contributions has often been surprising. Besides the results of iStockphoto and InnoCentive, which have been mentioned before, also Battle of Concepts showed some surprising results. One of the employees of a company that had posted a battle was so surprised about the results that he could not believe that the concepts were not made by a professional agency (appendix B.7). Furthermore, the Cornell Lab of Ornithology25 acknowledges that they do not only need the crowd for birding because they outnumber the professionals, but that in many cases amateur birders are more competent than professionals (Howe 2008). This all indicates that while the participants do not have the credentials of a professional, their contributions are of professional quality.
professionalism that is needed especially in crowd production can differ, this is the general onclusion for each of the types. c
FIGURE 2: CROWDSOURCING TYPES ON THE PRO‐AM CONTINUUM
Proposition 3b: Professionalism increases the probability of success in crowdcasting and crowd production (CMO)
It is important to note here that in crowdcasting and crowd production, both diversity and professionalism are important. Therefore, professionalism does not necessarily refer to a specific field, because the background of the participants needs to be at least somewhat different. It is not easy to acquire such a crowd that has both diversity and ability. Generally, a mediator is used in cases of a specific community. Examples are a community of students in Battle of Concepts (appendix B.7), a community of designers in cases like RedesignMe (appendix B.9) and CrowdSpring6, or a community of scientists at InnoCentive (appendix B.2). In the interview with Maxim Schram from RedesignMe, an interesting metaphor was used (appendix B.9). He argued that before there were supermarkets, people used specialist stores like bakeries and butchers. The store owners had the advantage of being able to focus on a limited expertise, while consumers knew where to go for specific products. When supermarkets emerged, people continued to get their bread from the bakery instead of the supermarket, both because they believed the quality was better at the specialist store and because there was just too much choice in the supermarket. Schram believes that focusing on one specialty is the right strategy in this emerging field. When there is more knowledge about and experience with crowdsourcing, it is possible that people can find their way in supermarkets, but for now it is better for both companies and the crowd to act as specialist stores. This metaphor is especially relevant for crowdcasting, as people ‘shop’ for challenges. This is also acknowledged by Battle of Concepts, since it was mentioned in the interview that the specific target group helps to convince companies to place a battle (appendix B.7). Using a mediator instead of starting an own initiative could help with this issue. Because a mediator collects similar challenges from different organizations, it is generally easier for a mediator to focus on a specific expertise than it is for a company creating their own initiative. Furthermore, the pro‐ams that are needed for crowdcasting are more likely to find and join the mediator community. Therefore, the
a i i char cterist c that crowdcast ng generally uses mediators is not surprising.
For crowd production this is different, since the initiative is often not part of a larger organization but stands on its own, which makes it easier to focus on a specific expertise and target group. Furthermore, as mentioned before, diversity and professionalism are not relevant in all crowd production cases.
Proposition 3c: In crowdcasting, using a mediator will increase the probability of success, because the mediator can focus on a specific expertise and target group (CIMO)
41
42
Activities o the c owd
Until now, some of the characteristics of the crowd are shown, such as diversity and professionalism. In addition, the expected activity of the participants is also important. According to Li and Bernoff (2008), the expected activity of participants depends on the technology that is used. They have provided a useful distinction between types of participants in their book Groundswell. A distinction is made between creators, critics, collectors, joiners, spectators and inactives. Creators are the most active group and contribute content themselves. Critics are active in rating and commenting on other people’s contributions. Collectors engage in activities like tagging and voting. Joiners maintain profiles on social networking sites and want to become part of a community. Spectators only consume other people’s content, while inactives perform none of these activities. Research company Forrester provides an online tool to determine the composition of a specific target group according to age, country and gender
f r
26. This composition is called the social technographics profile. This could help companies to decide which technologies would appeal to their target group. For example, when there are relatively few creators in the target group, setting up a wiki would probably attract many participants. Similarly, when the target group includes many spectators, a blog could be a good idea. Relating this to the crowdsourcing types, crowdcasting only needs creators. Crowd production would also need relatively many creators, although critics and spectators also have an important role. Because in crowd production the crowd generally produces for themselves, spectators have an important role. Although creators are needed in all types of crowdsourcing, crowdstorming allows more of a balance between creators, critics and spectators. Furthermore, joiners are important in crowdstorming and crowd production, as they are likely to join the community, especially when social networking features are included. Therefore, the social technographics profile can be useful in determining the type of crowdsourcing. It is important to note that the social technographics profile relates to the total target group and not directly to the participants. For example, when an individual in the target group is considered a creator because he regularly creates content for Wikipedia, this does not guarantee that this person will also create for another initiative. However, according to Li and Bernoff (2008) the social technographics profile provides an important indication for the expected activities, as it shows their ability and motivation to provide online content.
Proposition 4: Using the social technographics profile to determine the type of crowdsourcing will increase the probability of success, since it can predict the expected type of activity of participants (CIMO)
As explained in chapter three, a similar distinction between the activities of participants is made in open source software. However, only three levels of activity are identified, which are similar to the roles of creator, critic and spectator. This distinction is only relevant in cases with interaction, since otherwise there are no different roles. If the findings in open source software would be similar to crowdsourcing, there would be a small percentage of participants who contribute new content, and a somewhat higher percentage who change or comment on other people’s content, while the remaining participants only read and occasionally vote. While no information about the division of activity among participants is available for most cases, some cases suggest that this principle of open source software communities is also applicable to crowdsourcing. For example, one of the crowdstorming cases, Lego Mindstorms27, is discussed in Groundswell, showing these three types of participants. Similar results are found for Wikipedia (appendix B.8).
Proposition 5a: In cases with interaction, only a small portion of participants are creators, a somewhat higher part are critics, while the large majority consists of spectators (descriptive)
43
Conclusion
In this paragraph, six propositions were developed. They represent the answers to the second research question, which contains important information both for theory and practice. First of all, the expected participants in crowdsourcing are forming an increasingly diverse group. Diversity is important in both crowdcasting and crowd production. Although interaction between the participants can also be beneficial, as will be explained in the next paragraph, it can decrease the diversity because sharing knowledge makes the crowd more homogeneous. Pro‐ams are an important category of expected participants, as they already have the motivation
While above theory from open source software describes the relative activity of all participants, something can also be said of the distribution of activity among the creators. In some cases, the activity of the crowd seems to be exponentially distributed. In some cases, the top contributors are responsible for a large part of the content. For example in the case of Theyworkforyou.com, which asks the crowd to timestamp video content of the House of Commons28, the top contributor is responsible for 20 percent of the content. At Digg29, one percent of contributors is responsible for 32 percent of the page views. Even more extreme, more than half of the Wikipedia entries are contributed by only one percent of participants (appendix B.8). On the other side of the scale, a large percentage of tasks is done by one‐time contributors, as in the Clickworkers case24. From the data, this principle seems to be mostly relevant in crowd production cases.
Proposition 5b: In crowd production, the contributions of creators is exponentially distributed (descriptive)
Both descriptive propositions 5a and 5b are important for further research, to investigate what is the most dominant distribution of activity. Considering that only a small percentage of participants are creators, and many participants are one‐time contributors, organizations need a very large target group to create enough activity. According to Li and Bernoff (2008), about one to five percent of a company’s customers will participate in a forum. Several thousand participants are needed to get the forum going. Therefore, the initial target group needs to in the hundred thousands to be large enough to even have a chance of success.
Proposition 5c: In crowdstorming and crowd production, the larger the target group the higher the probability of success, because it is expected that only a small percentage of the target group will participate (CMO)
Finally, in the case of a mediator, the expected participants do not only consider the crowd but also the companies that are engaged in crowdsourcing. The interviews with Battle of Concepts and RedesignMe revealed that not all companies are ready to involve the crowd (appendix B.7 and B.9). This was also found by a large marketing survey among companies for Marketing Online (Hielkema et al. 2008). Consistent with Li and Bernoff (2008), they have experienced that companies are harder to convince of the usefulness of crowdsourcing than consumers. Many companies are concerned about sharing information and intellectual property issues. The fear that competitors will use their efforts is making many companies hesitant to apply crowdsourcing. But as increasingly more companies are confronted with the success of crowdsourcing and come to realize that the crowd talks about them anyway, crowdsourcing will become increasingly more accepted. This last proposition from this paragraph has a somewhat different perspective than the others, because this is mainly important for the mediators, instead of the companies that want to apply crowdsourcing.
Proposition 6: Companies are more difficult to engage in crowdsourcing than consumers (descriptive)
44
and ability to participate in these kinds of activities, and crowdsourcing gives them the opportunity they are searching for. Pro‐ams are both committed and knowledgeable participants, and are needed in crowd production and crowdcasting. The type of participants could also be defined according to the social technographics profile, as identified by Li and Bernoff (2008). The type of participants according to the social technographics profile show which technology is most suited for a specific target group. The expected activity of these participants was also shown in this chapter. When different roles are possible, only a small portion actually contributes new ideas or solutions, a somewhat larger portion changes or responds to these contributions, while most participants only read other people’s contributions. Even among the creators, there is a small group that is responsible for a large portion of the contributions, while many contributions are made by one‐time contributors. This means that a large number of participants is needed, which is only possible when the target group is sufficiently large.
6.2 CONDITIONS FOR CROWDSOURCING REQUEST
In this paragraph, the third research question Which conditions must be satisfied to generate productive outcomes? will be answered. Again, some of the propositions are descriptive, while others are explanatory (CMO or CIMO). As explained in 1.3, conditions to increase motivation and ability are important concepts, since they influence the success of crowdsourcing. Although the ability of the participants is already discussed in 6.1, motivation is an important issue in this paragraph. Furthermore, issues on knowledge creation and transfer, the conditions of the request, the phases of the innovation process, control, and the company’s efforts will be discussed extensively.
Motivation
First of all, there are some conditions that are needed to motivate the crowd to participate. As explained in chapter four, only financial and community‐related benefits are included in the study as they are considered most important for motivation. Although Howe (2008) argues that financial rewards are not the main motivation of participants, the data shows that this is in fact important, since about half of the cases (52 percent) offer a financial reward. The importance of financial rewards was also found by Antikainen and Väätäjä (2008). As can be seen from the previous chapter, financial rewards are relevant for crowdcasting and crowd production. The importance of financial rewards is evident in the case of Battle of Concepts (appendix B.7). While generally a reward of about €5000 is divided among the winners, there were two battles that did not offer a financial reward. Only ten concepts were submitted for these battles, compared to an average of 47. Also the UK Government case shows the importance of a financial reward, as they added a competitive element later on, because the response was not as high as they had hoped30.
Proposition 7a: Financial rewards increase the motivation of participants in crowdsourcing, which increases the probability of success (CIMO)
User driven innovation literature argues that while rewards are needed to increase the volume of the contributions, the quality could be decreased when using financial rewards. This could be the reason that financial rewards are absent in crowdstorming, as the user driven innovation literature is mostly relevant for crowdstorming, since users are the main target group. Therefore, participants need to be rewarded for the quality of their contributions, not for their participation only. This was also found by Battle of Concepts, which started rewarding study
45
Knowledge creation and transfer
Although allowing interaction between participants can be useful for increasing the motivation of participants, it can also influence the success of crowdsourcing more directly. As Nonaka (1994) and Nonaka and Toyama (2005) argue, creating and especially developing knowledge requires interaction. According to Joost Dekkers of Battle of Concepts, people can get inspired by seeing and communicating about contributions of others (appendix B.7). This was also evident from Assignment Zero, where people really started participating only after social networking features and forums were included in the website design (appendix B.10). Therefore, interaction can be an important condition to increase both the amount and quality of the contributions. The amount of contributions can be increased because people are inspired by
associations for participation, but revised their strategy by rewarding only the top ten contributions (appendix B.7). However, in crowdstorming and in the product category of crowd production it is difficult to reward individuals for their contributions, as the usefulness follows from discussions and the combination of tasks. Financial rewards are possible in the market form of crowd production, as it is possible in these cases to rewards individuals.
Because financial rewards are not possible in all cases, there must be other forms of motivation. The importance of community‐related benefits is argued by several researchers, as enabling interaction between participants is important for the value of the community and therefore the motivations of participants (Bagozzi and Dholakia 2002; Farquhar and Rowley 2006; Nambisan 2002). This is also evident from cases like Nabuur (appendix B.5), RedesignMe (appendix B.9) and Lego Mindstorms27, where there are no financial rewards, but there is a strong community. The opportunity to make new connections with people with similar interests is an important motivator in these cases.
Proposition 7b: Interaction increases the motivation of participants in crowdsourcing, which u sincreases the probability of s cces (CIMO)
However, when financial and community‐related benefits would be the only possible motivators, each case would have at least one of them, which is not the case. Still 17 cases have no financial rewards and no opportunities for interaction between participants, of which 11 in crowd production. Although this still implies that the two motivations are very important, because at least one of them is included in 85 percent of the cases, this means that other motivations are important as well. Looking at these cases, the relevant motivations differ. In some cases, it just does not take so much effort to participate, like in the cases of Google PageRank31, CitySense32 or TomTom33. In other cases, there are other personal benefits, like getting your contribution broadcasted on TV or published in a newspaper, like in the cases of Converse, CurrentTV34 and Gannett. Therefore, this topic needs further research, to find which motivators are important besides financial and community‐related benefits.
What is remarkable is that 33 cases use both financial rewards and interaction, making use of both motivators. One possible explanation comes from open source software literature. It was found that most participants start participating because of personal benefits, but decide to keep participating because of community‐related benefits (Shah 2003). This could also be the case for crowdsourcing, as financial motivations are the main reason to start participating, while participants acknowledge later on that the community‐related benefits are also an important motivator. Also Antikainen and Väätäjä (2008) have found that both types of motivations were important.
Proposition 7c: Financial benefits are an important motivator in the decision to start participating, while interaction is an important motivator to continue participating (CIMO)
46
each other’s ideas. The quality of the contributions can be increased because interaction allows further development of the ideas. The importance of being able to build on each other’s ideas is evident from an example that is not typically crowdsourcing, but shows the importance of interaction in collaborative production. Twice a year, Mathworks organizes a ten‐day web‐based Matlab programming contest. The source code is visible to all contestants, so the participants can build on each other’s work. Not only do the leading contributions include code from multiple participants, representing the combined efforts of participants, but on average, the best solution at the end of a contest period exceeds the best solution from day one by a magnitude of one thousand (Gulley 2001, 2004; Howe 2008). Such a result would never have been possible without this transparency.
However, because this interaction occurs online, the results are accessible to everyone. While in some cases this does not seem to be a problem, the fact that competitors are able to access the knowledge makes many organizations hesitant to allow interaction between participants or even to use crowdsourcing in general. This issue seems to be related to the field the activity is in. This can be seen most clearly in crowdcasting, in which about half of the cases have interaction, while the other half does not even show the contributions online. Many of the cases without interaction represent scientific research, while many of the cases with interaction represent creative activities.
A theoretical perspective that provides insights into this issue comes from knowledge management. The concepts analytical and synthetical knowledge base (Laestadius 1998) are useful to understand this issue. Analytical knowledge refers to industrial settings where scientific knowledge is highly important, and where knowledge creation is often based on cognitive and rational processes, or on formal models. Examples are genetics, biotechnology and general information technology. Knowledge is typically more often codified than in synthetical knowledge. The reason for this is that knowledge creation is often based on the application of scientific principles and methods, because knowledge processes are more formally organized and outcomes tend to be documented in reports, electronic files or patent descriptions. This increases the mobility of the knowledge. There are generally more radical innovations than with synthetical knowledge. A synthetical knowledge base refers to industrial settings, where the innovation takes place mainly through the application of existing knowledge or through new combinations of knowledge. Examples include plant engineering, specialized advanced industrial machinery, and shipbuilding. It often takes the form of applied research, but also product or process development. Knowledge is created in an inductive process of testing, experimentation, computer‐based simulation or through practical work. Although knowledge is partly codified, most knowledge is tacit, because it is based on experiences in certain situations. Compared to an analytical knowledge environment, the innovations are more incremental because it mostly involves the modification of existing products and processes (Laestadius 1998; Asheim and Gertler 2005; Asheim and Coenen 2005).
As crowdsourcing generally requires that tasks must be taken out of their context, and thus be mobile, fields with an analytical knowledge base are likely to be suited for crowdsourcing. Because of the analytical knowledge base, scientific problems are relatively easy to take out of their context. Therefore it is not surprising that crowdsourcing scientific problems like at InnoCentive was one of the first applications of crowdsourcing. The analytical knowledge base makes it easier for InnoCentive not to reveal the identity of the seekers, and it is easier to identify a solution, because the requirements can be easily specified. Therefore, crowdcasting is especially suited for fields with an analytical knowledge base. However, it is also necessary to protect the seeker companies by not revealing their identity and the application of the results, because when a task can be taken out of context, the results are likely to be useful in other
47
contexts as well, which makes it highly sensitive information. However, this is not necessary in all crowdsourcing cases, as not in all fields the results are useful to competitors.
As already mentioned, the crowdcasting examples that involve creative tasks publish the contributions on the website and allow interaction between participants. Besides the fact that it is more difficult to take the challenges out of their context because the knowledge is more synthetical, there is no need for protection. Unlike scientific knowledge, a design product is not sensitive to intellectual property issues, not because it is not relevant in another context, but because it is not beneficial to duplicate it. This is probably the reason that the crowdcasting cases where the participants form a community represent mostly creative tasks. While in most crowdcasting cases, interaction would cause difficulties because participants could use each other’s contributions while competing for a prize, this does not seem to cause problems in cases like CrowdSpring6 or RedesignMe (appendix B.9). Even a typical crowdcasting case like Battle of Concepts is considering to implement a plan with interaction, but thinks this would work best with design challenges (appendix B.7). The only exception to this theory is the Netflix prize, which represents a statistical challenge, where contributors share their solutions, even while there is a million dollar prize at stake15. Furthermore, a community is important for creative tasks as a social environment gives creative production a context in which the work has meaning (Howe 2008).
Proposition 8a: Interaction increases the probability of success in fields with a synthetical knowledge base, because participants can get inspired by each other’s ideas (CIMO)
Proposition 8b: Interaction decreases the probability of success in fields with an analytical knowledge base, because the competitive advantage of the company is risked (CIMO)
These propositions are only relevant for crowdcasting and crowd production, because within these types there is variation in the interaction attribute. Important to notice is that these crowdsourcing types also need diversity, on which interaction has a negative influence. For future research, it is important to investigate the size of these effects to be able to compare the
n o o fferent cinfluence of interactio on the success of cr wds urcing in di ontexts.
For crowdstorming and crowdfunding, the knowledge base is less relevant. Because participants are discussing about a topic that is familiar to all of them, they are able to share experiences. Because this does not require a task to be taken out of its context, fields with a
ossibsynthetical knowledge base are p le in crowdstorming and crowdfunding.
An analytical knowledge base is also relevant in crowd production because people have to be able to work on smaller parts of a product simultaneously. This is related to the requirement of modularity, which is considered an important success factor for open source software development. Modularity requires an analytical knowledge base, because the modules have to be taken out of their context.
Proposition 9a: Modularity is only possible with an analytical knowledge base (descriptive)
Many participants prefer relatively small tasks, with a clear beginning and end, and preferably a clear result. For a crowd to achieve critical mass, it is important to allow people with different amounts of time to spare to get involved. This was mostly clear from the interview with Nabuur, who realized that there weren’t enough small tasks for people to get involved more easily, for example by starting with smaller tasks to get familiar with the concept and the community, before spending a lot of time on the projects (appendix B.5). Many cases that include small tasks show that the majority of tasks were done by one‐time contributors, as in the Clickworkers
48
because it increases the participants’ creativity (CIMO)
As is mentioned in the previous section, the conditions of the request influence the type of contributions that the crowd is able to provide. According to some researchers, the request also influences the type of innovation, although the findings are not conclusive. Kristensson and Magnusson (2005) found that giving technological restrictions, which makes the request more specific, results in more incremental ideas, while giving no restrictions produces more radical ideas. However, Coyne et al. (2007) found that providing more specific requests results in more breakthrough innovations, arguing that participants need some guidance to think ‘out‐of‐the‐box’. Considering the differences between crowdstorming and crowdcasting, crowdstorming generally provides more incremental innovations, because the current products are generally the basis for the ideas and discussions. When more specific requests are provided like in crowdcasting, the participants can be guided and inspired by several questions they have to answer before coming up with an innovative idea (Coyne et al. 2007). Although this is an important issue for future research, the outcomes from the cases resemble the view of Coyne et al. (2007) more than that of Kristensson and Magnusson (2005).
case24. This shows the importance of the involvement of people who have only limited time to spare. To make crowdsourcing a success, also these people should be able to get involved.
Proposition 9b: Modularity increases the probability of success in crowd production, because it facilitates the distribution of resources (CIMO)
Conditions of request
Small tasks that represent modules are often relatively simple tasks with clear boundaries. This simplicity seems to be very important in crowdsourcing. Too many rules and constraints will demotivate people, especially the ones with little time to spare. Several practitioners consider simplicity to be one of the most important conditions for the success of crowdsourcing. RedesignMe, for example, argues that participants should understand a challenge within 30 seconds (appendix B.9).
Proposition 10: In crowdsourcing, simplicity increases the probability of success, because it increases both the opportunity and motivation of participants (CIMO)
Modularity and simplicity both make the request more specific. As mentioned before, people can get inspired by other people’s contributions. Similarly, people can already get inspired by the challenge itself, as experienced in Battle of Concepts (appendix B.7). This view is shared by several practitioners, who have argued that some requests are simply too open. This was for example experienced by Assignment Zero, a project that asked the crowd to report on crowdsourcing (appendix B.10). Participation was improved by focusing the request on interviews, instead of a more general request for content. This was also argued by NineSigma CEO Paul Stiros in Business Week, who argued that a request that is too open only provides ‘a ton of irrelevant information’35. This is also found by Magnusson et al. (2003), who argue that just asking for ideas is not enough. Also practitioner Carl Lens of CreativeCrowds36 argues that a balance between abstraction and concreteness has an important influence on the amount of creativity you can expect of a crowd. In his view, an open call for ideas as in most crowdstorming cases is simply too open. As a result, the creativity of ideas will not be deep enough37. Coyne et al. (2007) argue that most people are not very good at unstructured, abstract brainstorming, and that they need some guidance. By providing a basis for making and comparing choices the process is facilitated,sicne they can determine whether they are making progress.
Proposition 11: A more specific request increases the probability of success in crowdsourcing,
49
Proposition 12a: For breakthrough innovations, crowdcasting is most successful, because participants need some guidance to be able to change their perspective (CMO)
Proposition 12b: For incremental innovations, crowdstorming is most successful, because participants use current products as a basis for their contributions (CMO)
Phases of the innovation process
As explained in chapter four, the phase of the innovation process in which the crowd is involved can also influence the type of crowdsourcing that is best suited. As the name already shows, crowd production is generally used for production. Crowdcasting and crowdstorming are generally used for innovation, although crowdcasting is also used for support activities like logo design. According to Nambisan (2002), the roles of customers can vary according to the phase of the innovation process the customers are involved in, namely concept, development, testing, or customer support. Of the cases that use crowdsourcing in the concept phase, about half use crowdcasting (8) while the other half uses crowdstorming (10). Although further research is needed on this issue, some reasoning can provide insight. Since crowdstorming generally involves an open request for ideas, the concept phase seems the most suitable phase to use crowdstorming in. However, in crowdcasting it is also possible to ask for concepts, as is for example the case in Battle of Concepts (appendix B.7). Of the cases that use crowdsourcing in the development or design phase, 14 cases involve crowdcasting, while only 4 use crowdstorming. This implies that crowdcasting is the most suited approach for development or design. Concerning the cases that use testing, of which three cases are included in this study, they are categorized as hybrids. The reason is that while testing involves specific tasks similar to crowdcasting, a combination of contributions are used. Since these cases mainly resemble crowdcasting, it can be concluded that the crowdcasting approach is most suited for testing. Finally, customers can also have a role in the support phase, as many questions that customers have about their products can be answered by other product owners. Because interaction in a forum would be the most logical approach, crowdstorming can be used. This can also be seen from the data, since all cases which use crowdsourcing for customer service have applied a crowdstorming approach.
Proposition 13a: In the concept phase, crowdstorming is the most successful approach, because participants respond to an open request (CMO)
Proposition 13b: In the development phase, crowdcasting is the most successful approach, because a more specific request is common (CMO)
Proposition 13c: In the testing phase, crowdcasting is the most successful approach, because the request is more specific (CMO)
Proposition 13d: In the customer service phase, crowdstorming is the most successful approach, because participants can interact with each other (CMO)
According to Nambisan, interaction in the first three phases is mainly between the company and the customers, while in the support phase there is customer‐to‐customer interaction. Similarly, he argues that activities in the first three phases are mainly task oriented, while only customer service is socially oriented. As is shown in the previous chapter, this is not always the case, as crowdstorming is also used in the concept and development phases, which implies interaction between the participants. The difference between task orientation and social orientation therefore depend on the type of crowdsourcing, instead of the NPD phase.
50
Proposition 14: The difference between task orientation and social orientation is related to the type of crowdsourcing (descriptive)
It is important to recognize that Nambisan only considers involvement of customers, while the participants of crowdsourcing are not restricted to customers. Generally, customers are the main target group in crowdstorming, while other sources of information are involved in crowdcasting. Therefore, when the objective is to involve end users in the innovation process, crowdstorming is the most suitable approach. This is often related to marketing purposes, as crowdstorming can influence the image of the organization and because involving (potential) customers increases the probability that these participants will buy the new product. Although idea generation and new product development purposes are recognized, Deloitte’s 2008 Tribalization of Business Study38 has found that the top business objectives of communities are to generate more word of mouth and increase product and brand awareness. However, when it is preferable to include other sources than customers, crowdcasting is the most suitable approach.
Proposition 15: When marketing objectives are important, crowdstorming is the most successful approach, because of the target group and the possibilities for interaction (CMO)
Control
An issue that is making many companies hesitant to use crowdsourcing is control. Companies are used to having control not only within the organization but also for a large part outside the organization, concerning the communication about the company and its image (Hielkema et al. 2008). Putting some of this control in the hands of the crowd can therefore be a big step for organizations. Companies have to find a balance in the amount of control they are willing to hand over to the crowd. Having a sense of control can motivate the crowd to participate. As explained in chapter four, there is a difference between input and output control. Input control represents the amount of control of which contributions are published on the website, while output control refers to the amount of control that the crowd has in the decision which contribution to use or how. In most crowdsourcing cases the crowd has no output control, this generally stays with the company. An exception is crowdfunding, since the crowd provides financial resources for which they receive a vote in all important decisions, as is the case in for example MyFootballclub4 and Beerbankroll21. Sometimes the crowd has even more opportunities to be involved in exchange for their financial support, like in the case of Nvohk39, where the crowd can not only vote but also suggest ideas. Although voting is common in crowdstorming, and following the votes will contribute to the motivation of participants, the company generally retains its right to make the final decisions. However, compared with crowdcasting the crowd is much more in control. In crowdcasting, the company has the final vote in the decision on the winner and how to use its contribution, but since participants are paid for their effort this is unlikely to influence motivation. There is only one exception in this category, namely Threadless, which follows the votes of the crowd every week in choosing which T‐shirts to produce (appendix B.11). In the crowd production cases where the contributions are not directly used by a company, input and output control is the same. This is because the crowd decides what content is placed on the website and this content is directly used, generally as a source of information, by the crowd itself.
Input control in some cases also has an important influence on success. As Deloitte38 found, problems with online communities include companies exerting too much control over content and comments. Letting go of this control can have a positive influence, as is shown in the Chevrolet example. When Chevrolet crowdsourced the commercial video for the new Tahoe, the company was respected for the fact that they did not remove any negative contributions (Rose
51
because it increases the ability and motivation of participants (CIMO)
Secondly, providing feedback is argued to be an important condition. This gives participants a sense of control and gives meaning to their activities. Crowdsourcing practitioners consider feedback to be very important, not only to motivate people, but also because participants deserve to know what is done with their contributions. For example, Battle of Concepts encourages the companies that place a battle to give feedback via a weblog (appendix B.7). The NS has a topic about research results on their website (appendix B.6). Both Kraftfoods
2006). Although many companies are concerned about the negative content that the crowd can post on their website, it appears that this assumption is not legitimate. Li and Bernoff (2008) have found that 80 percent of the comments about products are positive. They even argue that the other 20 percent is needed to maintain credibility. However, it can still be difficult, especially in the beginning. For example, the communication at Dell’s IdeaStorm was not very positive at the start, and mainly revolved around customer problems. Only after a while the tone got more positive, and since then 160 new ideas from the community are implemented (appendix B.3). This positive tone can also be seen at MyStarbucksIdea.com, where participants who are irritated by long queues do not complain but instead offer suggestions for change40. Although increasingly more companies realize this, several companies still screen all the contributions before placing it on the website (Robeco41, UKGovernment30, Associated Content42, Lego Factory43, and all the microstock photo sites). Most companies, however, do not screen the contributions, but require registration, as this gives them the opportunity to remove users when needed. Additionally, many sites offer participants the opportunity to report inappropriate content. Besides the fact that the cases show that control from the company is in most cases not needed, it can also be harmful. When participants have a sense of control over their contributions and what is done with them, they will be more motivated to share their knowledge and ideas (Nambisan 2002).
Proposition 16a: Too much input control from the compan decreases the probability of success in crowdcasting and crowdstorming, because is decreases the motivation of participants (CIMO)
Proposition 16b: Too much output control from the company decreases the probability of success in crowdfunding and crowdstorming, because it decreases the motivation of participants (CIMO)
Efforts
Finally, Nambisan (2002) mentions three other conditions that could increase the success of crowdsourcing, which were already mentioned in chapter four. Evidence of the importance of these conditions was also found in the case studies. The first condition is providing information to the participants. According to Nambisan, this is not only important to help participants make high quality contributions, but it also shows the reciprocity of the organization, as the organization is prepared to do something in return for the ideas and knowledge that is shared by the participants. According to many practitioners, transparency is crucial for the success of crowdsourcing, as was shown for example in the interviews with Nabuur and NS (appendix B.5 and B.6). Although this represents an important concern for many companies, most examples show that the information is used with respect and that it results in high quality contributions. One clear example is the GoldCorp challenge, in which the company decided to publish all geological data, which is considered extremely valuable, and let the crowd analyze the data (appendix B.1). In the case of the Netflix challenge, all data about movie ratings was published for the participants to work with15. This is also argued by Porter and Donthu (2008), who found that providing quality content has positive effects on participants view on the company.
Proposition 17: Providing information increases the probability of success in crowdsourcing,
44 and Shell Game Changer45 provide participants with a time limit for feedback.
52
Proposition 18: Providing feedback increases the probability of success in crowdsourcing, because it increases the motivation of participants (CIMO)
Finally, active participation is considered an important condition. Showing commitment by active participation can also be important to motivate the participants, as this increases the feeling that they are being listened to. However, this is only applicable to cases with interaction. The importance of participation was found in Assignment Zero (appendix B.10), where the organization did not provide sufficient support for the participants. Even when participants are enthusiastic about your project, they need to be engaged. In some cases, active participation is mainly needed in the early stages of the initiative, as the crowd will be increasingly able to manage the community itself. This was for example the case with Nabuur (appendix B.5) and RedesignMe (appendix B.9). This is also argued by Li and Bernoff (2008), who argue that the organization must provide the website with initial content to drive traffic, and it will take at least a year before the community can maintain the activity themselves. Additionally, Magnusson et al. (2003) found that although users can provide more original ideas, interaction with professionals is needed to make these ideas producible.
Proposition 19: In crowdsourcing cases where interaction is encouraged, active participation increases the probability of success, because it increases the motivation of participants (CIMO)
Conclusion
In this paragraph, thirteen propositions were developed that consider the conditions that help crowdsourcing to be more successful, of which two are descriptive. These conditions provide important information for both theory and practice and are used to develop the protocol.
First of all, the motivations of participants have been discussed. As expected, financial rewards and interaction are both important motivators. Interaction can also be advantageous because it gives participants the opportunity to get inspired by each other’s ideas and to further develop them. However, interaction is not always possible, as there is the risk that competitors will benefit from the activities as well. Therefore, whether interaction is desirable depends on the field the activity is in. In fields with an analytical knowledge base, interaction is not desirable, but in fields with a synthetical knowledge base, this causes no problems. Since it is often easier for participants to come up with good ideas when a more specific request is placed, this seems to be an important condition. Because some guidance is needed to create a breakthrough idea, crowdcasting seems to be the more suitable approach for this type of innovation, while crowdstorming is better suited for incremental innovations. When marketing objectives are important and interaction between participants is desirable, using crowdstorming can be advantageous. The probability of success in crowd production is increased by modularity. However, this is only possible with an analytical knowledge base. Control is also an important issue in crowdsourcing, as many companies are concerned about losing control. However, since most cases so far experience that the crowd’s contributions are useful and generally positive, increasingly more companies are willing to give up some of the control over the communication towards and among the participants to be able to make use of the advantages that crowdsourcing can bring them. Giving the crowd some control is seen as an important motivator and success factor. Finally, providing knowledge, giving feedback and active articipation are considered important to make crowdsourcing successful. p
53
6.3 FACILITATING THE CROWDSOURCING PROCESS
In this paragraph, the fourth research question Which efforts are needed from the company to facilitate the crowdsourcing process? will be answered. As explained in chapter four, although as much quantitative information is included as possible, this question will be mainly answered qualitatively, since little information is available. While the previous two research questions were important for both theory and practice, this fourth research question is mainly relevant for the protocol. Because the objective of this question is to find activities that result in spending resources, instead of resulting in outcomes that are relevant to the success of crowdsourcing, no propositions will be developed for this research question. Instead, a list of activities is provided that are important when anticipating the amount of resources that are needed to start and maintain the crowdsourcing initiative. This is important information for practitioners, as Deloitte38 found that 45 percent of companies recognize that finding enough time to manage the community is one of the biggest obstacles to making communities work.
Technical issues
First of all, technical issues like website design can require significant efforts from the company. However, this is only relevant in the case of an own initiative. Even then, in many cases where a company does not have its own IT department, this task is outsourced to a specialist. When a mediator is used, the technical requirements are already in place. Therefore, this is an important reason for companies to choose a mediator over an own initiative.
Achieving and maintaining critical mass
Secondly, crowdsourcing will not work without a crowd. Therefore, efforts are needed to acquire participants. According to Deloitte38, 50 percent of the companies say that engaging the crowd is the most difficult part of making communities work. As mentioned in 6.1, the search for participants should not be too focused, as it is the diversity of people that makes crowdsourcing work. However, as promotion is needed, some thought must be put into reaching a target group. Although this target group can differ, extensive promotion of the initiative is needed. The phase in which promotion is mostly needed can differ. In the case of RedesignMe, for example, the first few hundred participants were acquired because several magazines and newspapers picked up the launch, causing much word of mouth marketing. Currently, they want to increase the number of participants, and more active promotion is needed (appendix B.9). However, in most cases, creating critical mass is the most difficult thing, and promotion is especially needed in the early phases. Besides more general promotional activities, it is also possible to have some kind of introduction event to trigger participants. IncSpring46, Lego Factory43 and CrowdSpring6 all held special contests in their launch phases to trigger participants to upload their contributions before a certain date. This can accelerate the process of achieving critical mass, since participants have an incentive not to postpone participation. Because companies that use a mediator can use an already existing community, this issue is only relevant for companies with an own initiative.
Community management
Because some of the conditions in the previous paragraph require efforts from the company, there is some overlap. This mainly concerns the final two propositions, concerning giving feedback and active participation. When the initiative fosters a community, these two can be combined, because they both concern community management. Farquhar and Rowley (2006) argue that active participation and engaging participants are important parts of community
54
management. When there is no community, only feedback is required. Because there generally is a deadline involved, the organization can take its time to process the results and provide feedback later on. This requires much less effort than with a community. With a community, this is different, because the organization needs to be available for questions, and they are expected to respond in a timely manner. This could be seen in the case of NS, where the organization decided to use a forum to let participants discuss about their ideas. However, because it often took more than two weeks to respond to questions or comments, the forum was (temporarily) discontinued (appendix B.6).
Processing results
When critical mass has been achieved and participants start contributing, the results have to be processed. While this processing can be different in different cases and types, this all needs significant effort from the company. The main difference in effort lies in the choice whether to use crowd voting or not. When the amount of contributions increases, at some point it becomes too much effort to go through all the content, especially because the amount of useless contributions is probably increasing as well. Another important issue is the input control, as the screening of contributions requires significant effort from the company. Processing results is often more challenging than it seems, since it often concerns getting people to get involved that are not directly involved in the initiative and have to do this next to their full‐time jobs.
Employees
All these activities need employees to perform them. For some cases there is information about how many employees are involved in the crowdsourcing activities, although it is hard to compare them. Although the contributions of crowdcasting and crowdstorming are generally used complementary to the organization’s activities, in crowd production often a task is really outsourced. In this case, many employees can be spared. An illustrating example is Wikipedia, which employs only five people, while Encyclopaedia Britannica was written by over four thousand paid contributors and one hundred full‐time editors (appendix B.8). Also cases like TomTom33, Theyworkforyou28, GalaxyZoo23, and Clickworkers24 represent cases that have saved a lot of time and resources on the part of the company. Another example is Threadless (appendix B.11), which has fewer than 20 employees, because they do not need any designers or marketeers. Nabuur made a conscious choice to be only facilitating, and they are running the organization with only five employees (appendix B.5). That a lack of sufficient employees can cause problems is evident from Assignment Zero. At the start of the project, only one person was available for contact for the 500 participants who visited the site the first week. This was obviously not working, and around 30 volunteer professional editors were assigned to manage the various topics (appendix B.10). Although it is clear that sufficient employees are needed to manage the community, it is hard to say how much effort is needed. For comparison, Dell started with about 30 employees representing all areas of the company, but has reduced its moderator staff to five to manage the more than 6000 participants of IdeaStorm (appendix B.3), KLM Bluelab47 also uses 40 employees to manage the community of about 1000 participants.
Although sufficient employees are needed, in many cases these employees have never been involved in such an initiative and therefore need training. As explained in chapter three about open innovation, new roles have to be created, like community manager or a web care team. The Deloitte survey38 found that the role of the Chief Marketing Officer (CMO) is being revolutionized through communities, since the CMO is often the person to initiate and manage the community. 42 percent of the respondents in the survey indicate that the marketing organization is now responsible for driving online communities. With communities becoming a central focus, marketing is now required to participate in nontraditional functions such as
55
customer support, idea generation and employee communications. That training is needed is evident from several cases. For example, in Assignment Zero, the 30 people assigned to manage the topics did not have any experience in online collaboration, which made their job difficult (appendix B.10). At Nabuur, facilitators and local representatives are trained (appendix B.5). At Gannett, a publisher of more than 80 newspapers that started several crowdsourcing initiatives, the organizational structure was completely changed. It was realized that instead of editors, they needed programmers and community managers, therefore many people needed additional training (Howe 2008).
Finally, management support is needed to make the crowdsourcing effort succeed. First of all, especially when the organizational structure has to be changed or new roles have to be created, support from top management is crucial. The company needs to take the necessary time and effort to change the traditional way of thinking. Especially the realization that part of the control of the company is handed over to the crowd needs time and effort (Li and Bernoff 2008). This is acknowledged by Dell, where CEO Michael Dell is one of the community’s biggest proponents (appendix B.3). Second, employees have to be given the time and resources to find their way in their new roles and get involved with the community. Especially with innovative projects, it is important that the project team receives the necessary space and resources to experiment. For example, the project team at NS had the opportunity to focus on the crowdsourcing initiative full time (appendix B.6).
Conclusion
In this paragraph, several activities were identified that need significant effort from the company to make the crowdsourcing initiative successful. Effort is needed for technical issues, achieving and maintaining critical mass, community management (which includes for example providing content and moderation), processing results (which can be higher when screening and lower with crowd voting), training, and management support. Sufficient employees are needed for these tasks, although it is hard to say specifically how many. While there are likely to be other activities that need to be considered when starting a crowdsourcing initiative, it is believed that the activities discussed in this paragraph have the most dominant influence on the total amount of resources that is needed.
6.4 CONCLUSION
In this chapter, the second, third and fourth research questions were discussed. In total, 19 propositions were developed to answer these research questions. Some of the propositions are descriptive, but most of them are explanatory. All propositions need further testing, although initial testing has been done through the case studies as far as possible. Since the propositions use the outcome ‘probability of success’ in contrast to more specific outcomes, it is important to investigate which more specific measures influence the success. The explanatory propositions are used for the protocol, and are therefore as much as possible formulated according to CIMO‐logic, including context, intervention, mechanism and outcome variables. However, because not all propositions could include an intervention variable, there were two types of explanatory propositions, namely CIMO or just CMO. The propositions without an intervention (CMO) are used to determine the crowdsourcing type that is most suited in a particular context, while the CIMO propositions are used to determine the necessary interventions. Although 6.3 did not include any propositions, the findings on the efforts that are needed from the company also provide important information for the protocol.
Although the propositions were formulated as design propositions, the underlying theoretical propositions are easy to identify. The relations between the components of the propositions should be further tested in future research. While the design propositions contain three or four components that are interrelated, the dual relations between the components should be further tested. These relations are shown are figure 3. The upper part shows the variables that are relevant for crowdsourcing in general, while the lower part shows the relations that have been found to be relevant for specific crowdsourcing types. From this figure can be seen that the proposition that diversity increases the success of crowdsourcing can be investigated separately from the proposition that interaction decreases diversity. Furthermore, the contexts in which these propositions are relevant can be investigated separately. For example, the interventions that are now related to crowdsourcing in general should be further investigated to determine whether this is the case for all types or whether the relation depends on a specific type or other context. From the figure can be seen that one of the most complex relations have to do with interaction, since there is a relation with crowdsourcing in general that depends on the knowledge base (proposition 8), but interaction also has a negative influence on diversity, which can be problematic in crowdcasting and crowd production. Therefore, it is important to investigate the relative influence of these relations to determine the best approach concerning interaction.
56
FIGURE 3: VISUALISATION OF THEORETICAL PROPOSITIONS
7 TOWARDS A PROTOCOL
In this chapter, the explanatory propositions from the previous chapter are translated into a protocol that companies can use for the application of crowdsourcing.
As is mentioned in the introduction, it is acknowledged that crowdsourcing can be relevant for other purposes than innovation, but this is the main application that is considered in this study. The protocol therefore only considers crowdstorming and crowdcasting, as they are both used for innovative purposes. Another reason is that this purpose is the most relevant one for Altuition, especially crowdstorming, as it concerns engaging customers in innovation. Protocols for crowd production and crowdfunding are not developed in this study, since they do not focus on innovation. Since relevant business models can be derived from these types, developing these protocols is important in future research.
Some of the propositions from the previous chapter involve intervention variables, while others do not. As is already explained, the CMO propositions will be used to determine whether to use crowdstorming or crowdcasting, while the CIMO propositions will be used to determine the necessary interventions.
The protocol consists of several phases, as can be seen in figure four, which represents the phases of crowdcasting and crowdstorming. The differences are mainly the results of the continuous nature of crowdstorming versus the batch nature of crowdcasting. While both diagrams include an application phase, in crowdcasting the associated activities are included in the discussion, while this is not the case in crowdstorming. The reasons for this is that the activities are not really part of the crowdsourcing initiative, since it involves processing the results into the internal R&D process.
57
FIGURE 4: CROWDCASTING (LEFT) AND CROWDSTORMING (RIGHT) PROCESS
7.1 DEFINITION PHASE
The first phase is the definition phase, in which a choice is made for a type and associated characteristics. In the previous chapter, there were several propositions that suggest context attributes that can determine the crowdsourcing type that is best suited for a particular situation.
Choosing whether crowdcasting or crowdstorming would be the most suitable approach is not easy, as it depends on multiple characteristics that can sometimes conflict. In total, there are five reasons identified in the previous chapter to choose one type over the other, which are hown in figure five. s
58
FIGURE 5: THE CHOICE BETWEEN CROWDCASTING AND CROWDSTORMING
59
First of all, a choice for crowdstorming would mean that a company has to create their own community. This requires a large customer base, as it is expected that only one to five percent of the target group will participate, and several thousand participants are needed to maintain activity (proposition 5c). Second, the target group not only has to be sufficiently large, but also contain the right type of participants. According to proposition 4, crowdcasting needs relatively many creators, while in crowdstorming critics and joiners are more important. Third, in many cases marketing objectives play an important role in the choice to introduce crowdsourcing. Although crowdcasting can be beneficial for the company’s image as well, the marketing objectives would lead to a choice for crowdstorming, since it can be used as a way of communication with the end users (proposition 15). Fourth, as stated in proposition 12, crowdcasting is more suited for breakthrough innovation, because the request is more specific, while crowdstorming is likely to result in more incremental innovations. So the choice for crowdstorming or crowdcasting depends on the objectives of the organization. Fifth, the NPD phase can influence the decision, although this is only based on theory and not on the case studies. In concept and testing, crowdcasting would be more suited, while crowdstorming would be the best choice for development and customer support (proposition 13).
Because five separate characteristics are considered, it is possible that they can be conflicting. Because there is not enough information about how and to what extent these characteristics influence success, it is a matter of judgment which type would be most suitable in the case of conflicting arguments. The first two arguments are requirements, as it is very unlikely that a crowdstorming initiative would be successful if the target group is not large enough, and it is very unlikely that a crowdcasting initiative would be successful if there are not many creators in the target group. The other three characteristics are in some way controllable by the company. If for example the fact that marketing objectives are considered important is conflicting with the other choices, it is possible to adjust the expectations, or for example to use both types in parallel for different purposes. This is very important to notice, since the protocol focuses on choosing one of these types. However, the choice should be based on different purposes, and therefore allows parallel use of crowdstorming and crowdcasting.
When a choice has been made, most of the characteristics are already determined, according to table 3. Although there may be some variations possible in some of the attributes, the protocol will only consider the most common characteristics. However, in crowdcasting there are still two choices left to make, namely whether to use a mediator or organize everything yourself and whether to allow interaction. In some way these choices are interrelated, as in the case of a mediator it depends on this mediator whether interaction is allowed or not. However, the choices will be discussed separately here.
Type
Role of the company
Type of participation
Party that takes
initiative
Whose contributions
are used
Financial rew
ards
Interaction
Input control
Output control
IP
Crowdcasting mediator/
own ini e tiativ
upload company in itiative individual yes yes/no
contributions not public; registration
company company
Crowdstorming own initiative forum crowd
initiative combination no yes registration company company/ public
TABLE 3: CHARACTERISTICS OF CROWDCASTING AND CROWDSTORMING
60
According to proposition 8, this choice for interaction depends on the field the activity is in, whether it has an analytical or synthetical knowledge base. When the activity is mainly based on analytical knowledge, interaction would decrease the probability of success, as the information presented on the website would be useful for other organizations as well. Because of IP issues, the company posting the challenge needs protection, and they will buy the rights from the contributor. When the activity is mainly based on synthetical knowledge, the company posting the challenge does not need this kind of protection. The interaction can then be beneficial, because it allows for the further development of the initial contribution and provides an extra motivation for the participants (proposition 7b). However, the interaction could lead to a decrease in diversity, which is needed in crowdcasting (proposition 2). Although this relation is not evident from the cases, theory suggests that interaction makes the crowd more homogeneous. The choice for interaction also has an influence of the amount of effort that is needed. For example, community management is needed to show the company’s commitment,
g ealthou h th effort is less than with crowdstorming, because the initiative is not continuous.
There are several advantages and disadvantages of using a mediator instead of an own initiative. First of all, proposition 3c states that focusing on a specific expertise and target group increases the probability of success, and this can be facilitated by using a mediator. A mediator can collect challenges from several companies, and specializes in crowdsourcing. Similar to outsourcing, it is easier for a mediator to maintain the quality of results and create a group of participants that is knowledgeable and committed. The professionalism that is needed in crowdcasting (proposition 3b) can be achieved more easily than with an own initiative, because the target group is more likely to gather on a website where challenges are collected that they are interested in. Furthermore, the amount of effort that is needed changes, as the company does not have to develop the website, but even more important, the participants that are needed are already in place. Furthermore, the company can benefit from existing knowledge about how a challenge should be formulated, for example how specific or simple it should be (propositions 10 and 11), thereby decreasing the uncertainty of the expected outcomes. However, when choosing for an own initiative, the effect on the company’s image is likely to be stronger, and the company has more control over the initiative, because they can choose how to organize it, without being dependent on another organization.
When a type and possible additional characteristics have been chosen, the definition phase is completed. The remaining four phases will be discussed separately for crowdcasting and crowdstorming in the next two pa aphragr s.
7.2 CROWDCASTING
In this paragraph, the remaining four phases in crowdcasting will be explained. Although there are some different possibilities in cases with or without interaction and with or without a mediator, most of the activities will be relevant for all crowdcasting cases. The differences will also be discussed. In figure 6, the activities in each phase are shown schematically.
Preparation phase
First of all, a suitable challenge has to be found and formulated. When a mediator is used, this organization will usually provide assistance in these activities. This phase can take a significant amount of time, for example in the case of Battle of Concepts about six months (appendix B.7). It is important that the challenge is not too complex, because the participants must understand it quickly enough to remain their interest. Furthermore, the request should be specific enough to
FIGURE 6: ACTIVITIES IN CROWDCASTING PHASES
inspire the participants and make it easier for them to provide more detailed contributions. For the contributions to be more detailed, it can help to provide the participants with some relevant information (proposition 17). For example, GoldCorp published all their geological data, which is considered very valuable to the company. But as it turned out, how people use this information was even more valuable, as new sources of gold were found which would never have been possible without providing the crowd with the data (appendix B.1). Since many companies are hesitant to inform the crowd, and especially their competitors, about their activities and associated sensitive information, this often requires a trade‐off.
When a mediator is used, finding a suitable organization would be the first step in this phase, as it can help with formulating the challenge. Because the mediator already has the necessary participants, it is important to determine the diversity of the audience (proposition 2). By making sure the participants, and thereby the contributions, are diverse, it is ensured that the choice for crowdsourcing is used optimally. Furthermore, as professionalism is important in
61
crowdcasting (proposition 3b), the background of the participants should be considered.
When the initiative has to be integrated in the company’s own website, it can be adapted to the choices that are made. There are several companies that are specialized in wikis, forums or social networking that could be useful when interaction is desirable. However, in many crowdcasting cases, a simple upload possibility is sufficient, which would not require much effort.
62
Evaluation phase
As said in chapter five, many crowdcasting cases without a mediator are periodic competitions. But also in cases with a mediator, often more than one challenge is posted, because these organizations are likely to have an established collection of companies that place the challenges. Therefore, it is important to evaluate the project after the competition has ended and the results
Making potential participants aware of the project and getting them to participate is not an easy task. Although less participants are needed compared to crowdstorming – in fact only one brilliant contribution would be sufficient – generally many contributions are required for the project to be successful. Although it would be easier to address some specific target group, since diversity is important it would be better to address a broader and more diverse audience. Therefore, significant effort is needed for advertising and promotion.
Although the challenge itself must appeal to the potential participants, an important motivator will be the financial reward. Because the available time to participate is often limited, the financial reward can make sure that people are triggered to visit the site (proposition 7c). Therefore, it is important to determine the size of the reward, the criteria, and the deadline for the contributions. The prizes are usually significant, ranging from $2000, to $1 million in the Netflix prize. Although the financial reward will influence the quantity of the contributions because more people are likely to be motivated to participate, it will also increase the quality of contributions, as they know only the best contribution(s) will be rewarded (proposition 7a).
Implementation phase
In this phase, the challenge is posted and if all goes well, the participants start contributing. When there is no interaction between the participants, which generally means the contributions are not published on the website, the company does not have to do much. However, when the contributions are public, it is possible that the company has decided to screen the results before posting them. Although many cases show that this is not necessary, there are companies that screen all contributions. This would take significant efforts from the company. Furthermore, when there is interaction involved, the company can participate in discussions, giving feedback about contributions, or suggesting adjustments to improve the result, which is all considered community management. This can improve the final result of the contributions, as the participants gain more insight about the specific requirements of the company. This can be seen for example at CrowdSpring6, where the company gives feedback on which designs or part of it they like, making other participants respond to it by better designs in the style the company is looking for. This participation and feedback requires significant efforts from the company, although it is not comparable to the effort that is needed for crowdstorming, since it is not needed to keep a discussion going, because the main contributions are the initial designs, and because the activity is only temporary.
Application phase
When the deadline is reached, it is time for the company to process the results and choose a winner. Therefore they have to go through all contributions, which takes significant efforts. An advantage of publishing the contributions on the site is that it is possible to include crowd voting. This is especially useful when a high number of contributions are expected, because it will significantly decrease the effort that is needed. However, in most crowdcasting cases this is not needed, as the amount of contributions is generally less than 100. As in most cases the results from the competition need further development, it provides input for the internal NPD process. This is also the phase where feedback is provided about the results of the competition and how the company is going to use the results (proposition 18).
have been implemented. Maybe the quality or quantity of the results was not satisfactory. It is for example possible that the request should be more or less specific or complex, or that the organization should participate more in the discussions and give feedback. A company can also experience that more or less employees are needed. If it is decided that more similar projects will be initiated, the evaluation can lead to changes in the preparation phase or even the definition phase.
From this description can be seen that there are many activities that the company has to anticipate when deciding to use crowdcasting. In figure 7, an indication is given of the relative effort that is needed in each phase. As can be seen from the dotted line, the amount of effort that is needed in the preparation and implementation phases depends on the choices for a mediator and interaction. The preparation phase needs less effort in case a mediator is used, because of the promotional activities and the help from the mediator with the formulation of the challenge. The execution phase needs more effort in case there is interaction, because more timely feedback and active participation is needed. Also the possible screening of results would take ore effort. m
63
FIGURE 7: R NG PHASES ELATIVE EFFORT IN CROWDCASTI
7.3 CROWDSTORMING
In this paragraph, the four remaining phases will be discussed for crowdstorming. Although some of the activities will remain the same, there are many differences, because the characteristics are in many cases opposite to the characteristics of crowdcasting. Unlike crowdcasting, there are no further choices to make concerning the characteristics, therefore all crowdstorming cases are considered to be similar. Compared to crowdcasting, the transition from one phase into the other is more fluent. In figure 8, the activities in crowdstorming are hown schematically. s
FIGURE 8: ACTIVITIES IN CROWDSTORMING PHASES
Preparation phase
Since crowdstorming concerns an own initiative, it involves technical issues like website design. Although this can be outsourced to a specialized organization, the requirements should be carefully determined. The social technographics profile should be used to determine whether it is expected that the crowd will contribute new ideas (creators) or that they will mainly respond to other people’s contributions (critics). When the target group contains many spectators, a combination with a weblog would be an option, while social networking features are useful when there are many joiners in the target group (proposition 4). However, a crowdstorming initiative generally involves a forum, and the other options are mainly additional. Since the choice for crowdstorming is (partly) determined by the target group, it will mainly consist of customers. It is therefore likely that some demographic information is available of the target
m
64
group, on which the social technographics profile can be deter ined.
This target group must be made aware of the initiative and their interest has to be obtained. Because the forum needs sufficient activity to gain the interest of participants, and it is expected that only a small portion of the participants are really active, while the remaining part only occasionally participates (proposition 5), several thousand participants are needed. Because the target group is generally known to the company, there are possibilities to contact these potential participants. Sufficient advertising and promotion is needed to achieve critical mass. As Li and Bernoff (2008) argue, it is expected that this will take at least one year. Therefore,
65
Evaluation phase
Already soon after the launch, it is important to evaluate continuously, and make adjustments when needed. Many crowdstorming initiatives provide participants the opportunity to suggest improvements for the website. This is an easy tool for evaluation. It is important to research the participants and their motivations, so the activities can be adjusted to this information. There are several performance measures for the crowdstorming initiative, for example the number of participants, the amount of contributions in terms of new ideas, comments and votes, and the amount of contributions that are being implemented. As can be seen in figure 8, there are feedback loops from evaluation to implementation and to maintenance, which implies that the results of the evaluation should be already used in the implementation phase and should be continued after the transition to maintenance has been made.
these activities remain to be important for the implementation phase. In the preparation phase, it is mostly important to prepare for the launch. It is for example possible to start with an introduction event, as explained in the previous chapter. This introduction can be a competition that can be organized as a one‐time crowdcasting event. This could help to trigger potential participants to visit the website.
Although active participation is not relevant until the implementation phase, it is important to provide the forum with initial content. This can be done by the company itself, but it is also possible to select a few volunteer participants to manage various topics, possibly for a small reward.
Furthermore, the organization must prepare itself for the introduction of the crowdsourcing initiative. As explained in the previous chapter, it is likely that new roles have to be created, such as community manager. The employees involved are therefore likely to need additional training. Management support is needed to provide the involved employees sufficient resources to make the crowdsourcing initiative a success. This is also important for control issues, as the organization needs to acknowledge that the community will determine much of the communication about the company and its products.
Implementation phase
The implementation phase is entered as soon as the initiative is officially launched. The most important objective of this phase is to create the critical mass that is needed to get the discussions going and make sure there is less effort needed from the company in the future. Therefore, significant effort is needed for advertising and promotion, to invite potential participants to join the discussions and provide their ideas. Community management is crucial in this phase of the process, as participants need to be welcomed, their questions answered, and feedback given so that participants feel their contributions are meaningful (propositions 18 and 19). In crowdstorming, it is very important to encourage interaction between participants, since the community‐related benefits are the main motivation of participants to contribute (proposition 7b). It will take some time for the crowd to become a community, but after a while, the community will be the reason for participants to keep coming back (proposition 7c). This means the first results need to be processed, selecting the most useful contributions. Compared to crowdcasting, there are generally much more contributions, causing difficulties in making a selection. Therefore, crowd voting is often included as a possibility on the website. It would be good for the ambiance in the community if some suggestions, however incremental, would be used, because this would show the organization’s commitment towards the community. Following the votes would also help giving meaning to the participant’s activities. Giving the community the control to post what they want is important, especially in this initial stage (proposition 16).
Maintenance phase
The transition from implementation to evaluation and maintenance is continuous. The maintenance stage means that a critical mass of participants is achieved, and less employees are needed for community management, because the community can mainly manage itself. Therefore, there is at least an evaluation moment between implementation and maintenance, as it can be decided during the evaluation that less employees are needed to manage the community. The most important activities include processing and implementing the results and giving feedback about it. However, evaluation is still needed in the maintenance stage, shown by the feedback loop from maintenance to evaluation.
Similar to figure 7, the amount of effort that is needed in the various stages of crowdstorming is shown in figure 9. As can be seen, most effort is needed in the implementation phase, when critical mass is not yet achieved and the community management takes significant efforts. Through the evaluation to the maintenance phase, increasingly less effort is needed, as the community is able to manage itself. As the initiative matures during the maintenance phase, evaluation is likely to take less and less time and effort.
66
FIGURE 9: RELAT EFF RMING PHASES IVE ORT IN CROWDSTO
7.4 CONCLUSION
In this chapter, the findings of the study were translated into a protocol that can be used by practitioners. Both crowdcasting and crowdstorming were discussed, since they are used for innovation, which is the focus in this study. The process starts with a definition phase, in which a choice is made between these two types and their associated characteristics. After this, the two types have a slightly different process, mainly due to the fact that crowdstorming has a more continuous nature, while crowdcasting works in batches. The activities that are needed in each of these four phases are explained in this chapter, as well as the relative effort that is needed. Figure 6 and 8 show the process of each of the types, while figure 7 and 9 show the effort that is needed in each phase. From these figures can be seen that the effort that is needed in each phase differs substantially between the two types, which is important information for practitioners. All explanatory propositions that were developed in chapter six were used to
67
develop the protocols. Therefore, the protocols has shown the relevance of this study’s findings for practice.
However, there are some limitations to the protocol. First of all, the propositions on which it is based need further testing. Furthermore, additional protocols should be developed for crowd production and crowdfunding. The protocol should not be considered as a detailed plan, but more as support that should be tailored to a specific situation. Nevertheless, this study provides important information for practitioners, since it provides an overview of possible choices in crowdsourcing and the activities that are needed to make a crowdsourcing initiative work.
68
69
8 CONCLUSION
In this concluding chapter, this study will be evaluated in hindsight. It contains a summary of the findings, as well as a discussion of the implications for theory and practice.
In essence, this study has made a first start at looking at crowdsourcing from a research perspective. The results from this study consist of four parts, namely a definition and context, a typology of crowdsourcing, propositions that can be used for further research, and a protocol that can be used to support practitioners when they want to apply crowdsourcing.
The study has placed crowdsourcing in perspective with several related concepts including user driven innovation, open innovation, open source software development and online communities. Although all these concepts are important to understand crowdsourcing, none of them embraces the entire topic. Therefore, this study has shown that crowdsourcing is a
orelevant new business m del that is an improvement or extension of already existing concepts.
These related concepts have provided important information to guide data collection and analysis in this study. Data was collected from over 100 crowdsourcing cases, that were used to answer the research questions. Studying their characteristics resulted in the development of a typology of crowdsourcing, including crowdcasting, crowdstorming, crowd production and crowdfunding. The typology is based on the structural characteristics of the cases, representing the intervention attributes as explained in chapter four. This provides significant advantages, especially for companies that want to apply crowdsourcing, since it focuses on organizational design choices.
Combining literature and the case studies revealed important information regarding the research questions on the expected participants, the conditions that are needed to make crowdsourcing a success and the efforts that are needed from the company. Propositions were developed that were subjected to initial testing through the available information on the case studies. The propositions describe and explain relations between interventions, outcomes and the context in which this relation is relevant. The findings revealed the importance of both diversity and ability of participants for the success of crowdsourcing, as well as the existence of different amounts and types of participation, represented by the difference between creators, critics and spectators. The study revealed that both financial rewards and interaction increase the motivation of participants, while interaction, simplicity and specificity increase the ability of participants. Other conditions that increase the probability of success in crowdsourcing include providing information and feedback and active participation of the organization. The activities that need the most effort from the company include technical issues, achieving and maintaining criticial mass, community management, processing results, and training of employees.
These findings have provided important information for the protocol that was developed for crowdcasting and crowdstorming. While the process is slightly different for each type, they both consist of five phases, starting with the definition phase, in which a choice must be made between the two types. The study provides information on which type is most relevant in which context, helping organizations to make a choice. Furthermore, the protocol provides information on the activities that are relevant in each phase of the process, and the relative effort that is needed for each phase.
This study provides important information for both academics and practitioners. During this study, the objective was to maintain a balance between implications for theory and practice.
70
According to Van Aken (2004), management research suffers from a utility problem, since there is too much focus on rigorous results, instead of relevance for practitioners. Although some researchers have referred to this as the rigor‐relevance dilemma, Van Aken argues that the two can coexist. Therefore, explanatory research should be complemented with prescriptive research. This study attempted to include both perspectives, by considering descriptive, explanatory and prescriptive purposes. First of all, the typology represents a description of crowdsourcing applications. Secondly, theoretical propositions were developed for explanatory purposes. The relations that are implied by these propositions represent important information for organization theory. Finally, these propositions were formulated as design propositions, so the descriptive and explanatory research could be used for prescriptive purposes, as been shown in the protocol, thereby contributing to management theory. In the next section, the contributions to both organization theory and management theory will be discussed, as well as the limitations of this study.
As mentioned before, significant contributions have been made to organization theory, since the typology can be considered a theoretical framework. Besides that the typology itself is important information on crowdsourcing, it can also be used to structure future research, as it can be focused on a specific type. Furthermore, although the propositions are formulated as design propositions, they can be considered as translations and combinations from theoretical propositions that describe the relation between two variables. Since the propositions were developed from information on several case studies, they can be considered initially tested.
eHowever, further research is needed to test th propositions.
Also significant contributions have been made to management theory. Although the propositions can be seen as theoretical propositions, they are formulated as design propositions in order to be able to use them directly for the protocol. The protocol was developed for practitioners that want to introduce crowdsourcing in their company. Since there is no preliminary research on this subject, this study represents important information for practitioners, such as which possibilities exist in crowdsourcing and how crowdsourcing is related to their current activities and objectives. All intervention variables have indeed proven useful. Two variables have been used for the categorization of the cases, while the remaining variables are all used in the propositions.
For Altuition in particular, this study has provided information on this emerging trend and shows them how other companies are embracing it. With the protocol, Altuition can advise their clients about the application of crowdsourcing. Although crowdcasting could also be relevant, crowdstorming is the most relevant type for them.
However, there are also limitations to this study. Although much information has been collected, from which many important findings could be developed, the results should not be considered definitive. Because of the emerging nature and broad range of the topic, this study forms a
sub ascientific basis for the ject to mature from further studies nd experiences.
Because the purpose of this study was theory building, as opposed to theory testing, further testing of the propositions is needed to validate the findings. Furthermore, more quantitative research should be conducted to investigate the relative importance of the interventions. For example, it is important to further investigate the influence of interaction, since it was found that it can have a direct positive influence on success, but at the same time has a negative influence on diversity, and is also dependent on the knowledge base. The same can be concluded for the contexts in which the propositions are relevant. Especially the context in which the different crowdsourcing types should be applied requires further attention, for example which type is most suited for which NPD phase and which type of innovation. Another area that needs
71
further research is the motivation of participants. Although the expectation that financial and community‐related benefits are important was confirmed, there are likely to be other sources of motivation that can contribute to the success of crowdsourcing. Since one of the worries of companies is their loss of control, the influence of both input and output control should be further investigated. More generally, since this study has only focused on crowdsourcing itself, future research should investigate the results of crowdsourcing compared to more traditional approaches.
Since the protocol was developed from these propositions, the protocol needs to be validated as well. The protocol should not be considered a detailed plan, but more as support that should be tailored to a specific situation. Furthermore, the protocol only considers crowdstorming and crowdcasting. The possibilities of the other two types should be further researched. In this emerging field, practitioners need to experiment to gain further knowledge about the possibilities and limitations of crowdsourcing.
Despite the rather exploratory nature of this study, it has provided important findings for both theory and practice. It is important to recognize that crowdsourcing is still an emerging trend, and it is likely that organizations will increasingly experiment with this new business model, allowing study of the topic in more detail. This study has shown that crowdsourcing is more than just a hype, and can be considered a new business model that is likely to be relevant for many organizations in many fields. The limits of crowdsourcing are not nearly in sight, as new applications appear every day. With this report, organizations can get ahead of the game, by taking their advantage of the knowledge and guidance that this study provides. Crowdsourcing is likely to continue to influence the means of economic production. Concepts that were once clearly understood like ‘amateur’ and ‘consumer’ now need different definitions. Companies have started to rethink their strategies towards innovation, knowledge management and control. Crowdsourcing has already transformed industries such as photography and ournalism. Only future can tell which one will be next. j
72
73
REFERENCES
Websites
ember 2, 2008) 1 www.crowdsourcing.com (last visited Dec2 www.kiva.org (last visited January, 2009) 3 www.sellaband.com (last visited January 23, 2009) 4 www.myfootballclub.co.uk (last visited January 23, 2009)
y 23, 2009) 5 www.fluevog.com/files_2/os‐1.html (last visited Januar
) 6 www.crowdspring.com (last visited January 25, 20097 www.99designs.com (last visited January 30, 2009)
0, 2009) 8 www.project10tothe100.com (last visited January 39 www.ideascale.com (last visited January 23, 2009)
y 23, 2009) 10 www.getsatisfaction.com (last visited Januar11 www.spot.us (last visited January 23, 2009) 12 www.bandstocks.com (last visited January 30, 2009) 13 www.peugeot.com/en/design/design‐contest/5th‐edition/2008‐theme.aspx (last
visited January 30, 2009)
y 30, 2009) 14 www.mturk.com/mturk/welcome (last visited Januar15 www.netflixprize.com (last visited January 23, 2009)
9) 16 www.mystarbucksidea.com (last visited January 23, 200
9) 17 www.marketocracy.com (last visited January 30, 200
) 18 www.nowpublic.com (last visited January 30, 200919 www.webook.com (last visited January 30, 2009)
09) 20 www.amillionpenguins.com (last visited January 30, 2021 www.beerbankroll.com (last visited January 25, 2009)
009) 22 www.ideaconnection.com (last visited January 25, 223 www.galaxyzoo.org (last visited January 23, 2009) 24 clickworkers.arc.nasa.gov (last visited January 23, 2009) 25 http://www.birds.cornell.edu/netcommunity/Page.aspx?pid=708 (last visited
January 25, 2009)
09) 26 www.forrester.com/Groundswell/profile_tool.html (last visited January 23, 20
ry 23, 2009) 27 mindstorms.lego.com/community/default.aspx (last visited Janua28 www.theyworkforyou.com/video (last visited January 23, 2009)
74
29 www.digg.com (last visited January 23, 2009) 30 www.showusabetterway.co.uk (last visited January 23, 2009) 31 www.google.com/corporate/tech.html (last visited January 23, 2009)
09) 32 www.sensenetworks.com/citysense.php (last visited January 23, 20
ited January 23, 2009) 33 www.tomtom.com/page/mapshare (last vis34 current.com (last visited January 23, 2009) 35www.businessweek.com/innovate/content/jun2007/id20070611_139079_page_2.ht
m (last visited January 25, 2009) 36 www.creativecrowds.com (last visited January 25, 2009) 37 www.crowdsourcingdirectory.com/?p=124 (last visited January 25, 2009) 38 www.deloitte.com/dtt/article/0,1042,sid=108577%2526cid=219449,00.html (last
visited January 25, 2009) 39 www.nvohk.com (last visited November 12, 2009) 40 http://www.businessweek.com/magazine/content/08_17/b4081000030457.htm
(last visited January 25 2009) 41 www.robeco.com/connect (last visited January 25, 2009)
y 25, 2009) 42 www.associatedcontent.com (last visited Januar43 factory.lego.com (last visited January 25, 2009) 44 brands.kraftfoods.com/innovatewithkraft (last visited January 25, 2009) 45www.shell.com/home/content/global_solutions/innovation/innovative_thinking/gam
e_changer (last visited 23 January 2009) 46 incspring.com (last visited January 25, 2009) 47 www.klmbluelab.nl (last visited January 30, 2009)
75
Literature
Albors, J., Ramos, J.C. and Hervas, J.L. (2008) “New learning network paradigms: Communities of objectives, crowdsourcing, wikis and open source”, International Journal of Information Management, Vol. 28, Iss. 3, p. 194.
Amant, K.S. and Still, B. (2007) “Handbook of research on open source software: Technological, economic, and social perspectives” Information Science Reference, London.
Antikainen, M. and Väätäjä, H. (2008) “Rewarding in open innovation communities – How to motivate members?”, Proceedings of XIX ISPIM Conference “Open Innovation – Creating Products and Services through Collaboration, Tours, France.
Asheim, B.T. and Coenen, L. (2005) “Knowledge bases and regional innovation systems: Comparing Nordic clusters”, Research Policy, vol. 34, Iss. 8, p. 1173.
Asheim, B.T. and Gertler, M. (2005) “The geography of innovation: Regional innovation systems”, in Fagerberg, J., Mowery, D., and Nelson, R., Oxford University Press.
Bagozzi, R.P. and Dholakia, U.M. (2002) “Intentional social action in virtual communities”, etingJournal of Interactive Mark , Vol. 16, Iss. 2, p. 2.
Barnard, W. and Wallace, T.F. (1994) “The innovation edge: Creating strategic breakthroughs Junction, using the voice of the customer”, Essex Omneo.
Berkhout, A.J., Hartmann, D., Van der Duin, P. And Ortt, R. (2006) “Innovating the innovation process”, International Journal of Technology Management, Vol. 34, Iss. 3/4, p. 390.
Bjelland, O.M. and Wood, R.C. (2008) “An inside view of IBM’s ‘Innovation Jam’”, MIT Sloan Management Review, Vol. 50, Iss. 1, p. 32.
Bloem, J. and Van Doorn, M. (2007) “Open voor business: Open‐source‐inspiratie voor i s innovat e”, Verkenningsin tituut Nieuwe Technologie ViNT.
Boxall, P. and Purcell, J. (2003) “Strategy and human resource management”, Palgrave Macmillan, Basingstoke.
Brabham, D.C. (2008) “Crowdsourcing as a model for problem solving: An introduction and cases”, Convergence: The International Journal of Research into New Media Technologies, Vol. 14, Iss. 1, p. 75.
Brockoff, K. (2003) “Customers’ perspectives of involvement in new product development”, International Journal of Technology Management, Vol. 26, Iss. 5/6, p. 464.
Bughin, J., Chui, M. and Johnson, B. (2008) “The next step in open innovation”, The McKinsey Quarterly, June 2008.
Casalo, L.V., Flavian, C. and Guinaliu, M. (2008) “Fundaments of trust management in the s pdevelopment of virtual communities”, Management Research News, Vol. 31, I s. 5, . 324.
Chesbrough, H.W. (2003) “Open innovation: An imperative for creating and profiting from technology”, Harvard Business School Press, Boston.
Chesbrough, H.W., Vanhaverbeke, W.P.M. and West, J. (2006) “Open innovation: Researching a new paradigm”, Oxford University Press, Oxford.
76
Chesbrough, H.W. (2007) “Why companies should have open business models”, MIT Sloan Management Review, vol. 48, Iss. 2, p. 21.
Chiu, C.M., Hsu, M.H. and Wang, E.T.G. (2006) “Understanding knowledge sharing in virtual communities: An integration of social capital and social cognitive theories”, Decision Support
. Systems, Vol. 42, Iss. 3, p 1872.
Chua, A. (2002) “The influence of social interaction on knowledge creation”, Journal of Intellectual Capital, Vol. 3, Iss. 4, p. 375.
Ciccantelli, S. and Magidson, J. (1993) “From experience: Consumer idealized design: Involving consumers in the product development process”, Journal of Product Innovation Management, Vol. 10, Iss. 4, p. 341.
, ods”, Cooper D.R. and Schindler, P.S. (2003) “Business research meth McGrawHill: New York.
Coyne, K.P, Clifford, P.G. and Dye, R. (2007) “Breakthrough thinking from inside the box”, Harvard Business Review, Vol. 85, Iss. 12, p. 145.
Dearstyne, B.W. (2007) “Blogs, mashups, & wikis: Oh, my!”, Information Management Journal, Vol. 41, Iss. 4. p. 25.
Denyer, D., Tranfield, D. and Van Aken, J.E. (2008) “Developing design propositions through Sresearch synthesis”, Organization tudies, Vol. 29, Iss. 3, p. 393.
Desouza, K.C., Awazu, Y., Jha, S. and Dombrowski, C. (2008) “Customer‐driven innovation”, Research Technology Management, Vol. 51, Iss. 3, p. 35.
Dholakia, U.M., Bagozzi, R.P. and Pearo, L.K. (2004) “A social influence model of consumer participation in network‐ and small‐group‐based virtual communities”, International Journal of Research in Marketing, Vol. 21, Iss. 3, p. 241.
Dul, J. and Hak, T. (2008) “Case study methodology in business research”, ButterworthHeinemann: Amsterdam.
Farquhar, J. and Rowley, J. (2006) “Relationships and online consumer communities”, Business Process Management Journal, Vol. 12, Iss. 2, p. 162.
Füller, J., Bartl, M., Ernst., Mühlbacher, H. (2006) “Community based innovation: How to integrate members of virtual communities into new product development”, Electronic Commerce Research, Vol. 6, Iss. 1, p. 57.
Füller, J. and Matzler, K. (2007) “Virtual product experience and customer participation – A chance for customer‐centred, really new products”, Technovation, Vol. 27, Iss. 6/7, p. 378.
Grand, S., Von Krogh, G, Leonard, D. and Swap, W. (2004) “Resource allocation beyond firm boundaries: A multi‐level model for open source innovation”, Long Range Planning, Vol. 37, Iss. 6, p. 591.
Granovetter, M.S. (1973) “The strength of weak ties”, The American Journal of Sociology, Vol. 78, Iss. 6, p. 1360.
Griffin, A. and Hauser, J.R. (1993) “The voice of the customer”, Marketing Science, Vol. 12, Iss. 1, p. 1.
77
Gulley, N. (2001) “Patterns of innovation: A web‐based MATLAB programming contest”. URL: a 0http://www.starchamber.com/gulley/pubs/contest‐paper.pdf (last visited Janu ry 24, 20 9).
Gulley, N. (2004) “In praise of tweaking: A wiki‐like programming contest”. URL: http://www.starchamber.com/gulley/pubs/tweaking/tweaking.html (last visited January 24, 2009).
Hertel, G., Niedner, S. and Herrmann, S. (2003) “Motivation of software developers in open source projects: An Internet‐based survey of contributors to the Linux kernel”, Research Policy, Vol. 32, Iss. 7, p. 1159.
Hielkema, R., Van Vugt, T. and Bosveld, K. (2008) “9e Grote Marketing Enquête: marketeer heeft nog steeds weinig online kennis”, Tijdschrift voor Marketing, Vol. 42, Iss. 7/8, p. 8.
Hof, R., Browder, S. and Elstrom, P. (1997) “Internet communities‐forget surfers: A new class of netizen is settling right”, Business Week, Iss. 5, p. 38.
Howe, J. (2006) “The rise of crowdsourcing” Wired, Iss. 14.06. URL: http://www.wired.com/wired/archive/14.06/crowds.html last visited 24‐01‐2009 (last visited January 24, 2009).
Howe, J. (2008) “Crowdsourcing: Why the power of the crowd is driving the future of business” Crown Business: New York.
Hung, C.L., Chou, J.C‐L. and Shu, K.Y. (2008) “Searching for lead users in the context of web 2.0”, Proceedings of ICMIT, 4th International Conference on Management of Innovation and Technology, Bangkok September 24th 2008.
Janzik, L. and Herstatt, C. (2008) “Innovation communities: Motivation and incentives for community members to contribute”, Proceedings of ICMIT, 4th International Conference on
mber 24th 2008. Management of Innovation and Technology, Bangkok Septe
Kano, N. (1984) “Attractive quality and must‐be quality”, The Journal of the Japanese Society for Quality Control, Vol. 14, Iss. 2, p. 39.
Kristensson, P., Gustafsson, A. and Archer, T. (2004) “Harnessing the creative potential among f nusers”, Journal o Product Innovation Ma agement, Vol. 21, Iss. 4, p. 4.
Kristensson, P. and Magnusson, P.R. (2005) “Involving users for incremental or radical innovation – a matter of tuning”, paper for the 12th International Product Development management Conference in Copenhagen, Denmark.
Laestadius, S. (1998) ”Technology level, knowledge formation, and industrial competence in paper manufacturing”, in Eliasson, G. and Green, C. “Microfoundations of economic growth – A Schumpeterian perspective”, Ann Harbour, Michigan.
Lakhani, K.R. and Jeppesen, L.B. (2007) ”Getting unusual suspects to solve R&D puzzles”, Harvard Business Review, Vol. 85, Iss. 5, p. 30.
Lakhani, K.R., Jeppesen, L.B., Lohse, P.A. and Panetta, J.A. (2006) “The value of openness in scientific problem solving”, Harvard Business School Working Paper, No. 07050.
Lakhani, K.R. and Panetta, J.A. (2007) “The principles of distributed innovation”, Innovations: Technology, Governance, Globalization, Vol. 2, Iss. 3, p. 97.
78
Lauglaug, A.S. (1993) “Technical‐market research – Get customers to collaborate in developing products”, Long Range Planning, Vol. 26, Iss. 2, p. 78.
Leadbeater, C. and Miller, P. (2004) “The pro‐am revolution: How enthusiasts are chaning our Demos: economy and society”, London.
Lenhart, A., Horrigan, J. and Fallows, D. (2004) “Content creation online”, Pew Internet and American Life Project, February 29.
Li, C. and Bernoff, J. (2008) “Groundswell: Winning in a world transformed by social technologies”, Harvard Business Press: Boston.
Liyanage, S., Greenfield, P.F. and Don, R. (1999) “Towards a fourth generation R&D management model‐research networks in knowledge management”, Vol. 18, Iss. 3/4, p. 372.
Magnusson, P.R., Matthing, J. and Kristensson, P. (2003) “Managing user involvement in service innovation”, Journal of Service Research, Vol. 6, Iss. 2, p. 111.
Mascitelli, R. (2000) “From experience: Harnessing tacit knowledge to achieve breakthrough innovation”, The Journal of Product Innovation Management, Vol. 17, Iss. 3, p. 179.
Miles, M.B. and Huberman, A.M. (1994) “Qualitative data analysis: An expanded sourcebook”, Sage: London.
Mintzberg, H. (1979) “The structuring of organizations: A synthesis of the research”, PrenticeHall: Englewood Cliffs.
Nambisan, S. (2002) “Designing virtual customer environments for new product development: Toward a theory”, Academy of Management Review, Vol. 27, Iss. 3, p. 392.
Nobelius, D. (2004) “Towards the sixth generation of R&D management”, International Journal t .of Projec Management, Vol. 22, Iss. 5, p. 369
Nonaka, I. (1994) “A dynamic theory or organizational knowledge creation”, Organization Science, Vol. 5, Iss. 1, p. 14.
Nonaka, I. And Toyama, R. (2005) “The theory of the knowledge‐creating firm: Subjectivity, rate Change, Vol. 14, Iss. 3, pobjectivity and synthesis”, Industrial and Corpo . 419.
Nov, O. (2007) “What motivates Wikipedians”, Communications of the ACM, Vol. 50, Iss. 11, p. 60.
Page, S.E. (2007) “The difference: How the power of diversity creates better groups, firms, i n .schools and societies”, Princeton Univers ty Press: Pri ceton
Porter, C.E. and Donthu, N. (2008) “Cultivating trust and harvesting value in virtual communities”, Management Science, Vol. 54, Iss. 1. p. 113.
Prahalad, C.K. and Ramaswamy, V. (2004a) “Co‐creation experiences: The next practice in value creation”, Journal of Interactive Marketing, Vol. 18, Iss. 3, p. 5.
Prahalad, C.K. and Ramaswamy, V. (2004b) “Co‐creating unique value with customers”, Strategy & Leadership, Vol. 32, Iss. 3, p. 4.
Robertson, A. (1984) “Characteristics of the successful inventor: Some notes on the nature of creativity and the creative mind”, Technovation, Vol. 2, Iss. 2, p. 141.
79
Rose, F. (2006) “Commercial Break”, Wired, Issue 14.12. URL: http://www.wired.com/wired/ archive/14.12/tahoe.html (last visited January 24, 2009)
Rothwell, R. (1994) “Towards the fifth‐generation innovation process”, International Marketing Review, Vol. 11, Iss. 1. p. 7.
Sawhney, M., Prandelli. and Verona, G. (2003) “The power of innomediation”, MIT Sloan Management Review, Vol. 44, Iss. 2, p. 77.
Shah, S.K. (2003) “Community‐based innovation & product development: Findings from open logysource software and consumer sporting goods”, Massachusetts Institute of Techno .
Sicilia, M. and Palazon, M. (2008) “Brand communities on the Internet: A case study of Coca‐Cola’s Spanish virtual community”, Corporate Communications: An International Journal, Vol. 13, Iss. 3, p. 255.
Surowiecki, J. (2004) “The wisdom of crowds: Why the many are smarter than the few and how Ycollective wisdom shapes business, economies, societies, and nations”, Doubleday: New ork.
Tapscott, D. and Williams, A.D. (2006) “Wikinomics: How mass collaboration changes everything”, Penguin Books: London.
open mind”, Leader to Leader, Vol. 2007, Iss. 43, p. 32. Taylor, W.C. (2007) “Leading with an
Toffler, A. (1980) “The Third Wave”, Random House Value Publishing: New York.
Van Aken, J.E. (2004) “Management research based on the paradigm of the design sciences: The quest for field‐tested and grounded technological rules”, Journal of Management Studies, Vol. 41,
9Iss. 2, p. 21 .
Van Aken, J.E. (2007a) “Developing organization studies as an applied science using a triple learning approach”. URL: http://www.egosnet.org/jart/prj3/egosnet/data/uploads/OS_2007/ W‐057.pdf (last visited January 24, 2009).
Van Aken, J.E. (2007b) “Design science and organization development interventions: Aligning business and humanistic values”, The Journal of Applied Behavioral Science, Vol. 43, Iss. 1. p. 1.
Van Wendel de Joode, R. (2005) “Understanding open source communities: An organizational perspective”, Enschede.
University Pres New York. Von Hippel, E. (1988) “The sources of innovation”, Oxford s,
Von Hippel, E. (2005) “Democratizing innovation”, MIT Press, Cambridge.
Weiss, R.S. (1994) “Learning from strangers: The art and method of qualitative interview studies”, Free Press: New York.
Yin, R.K. (2003) “Case study research: Design and methods”, Sage: London.
80
81
APPENDIX A
In this appendix, the data that is collected for each of the cases is shown. The data consists of several tables, organized according to type. The first set of tables contains the information on the intervention and context attributes, while the second set of tables present the information that was found on the outcome attributes, as shown in table 4.
Intervention and ntext attributes co Outcom ributese att
Crowdcasting A.1 A.6
Crowdstorming A.2 A.7
Crowd production A.3 A.8
Crowdfunding A.4 A.9
Hybrids A.5 A.10
TABEL 4: ARRANGEMENT OF APPENDIX A
The sources in which the information was found are included in the table. However, this needs some additional explanation. An important category of sources are websites where collection of examples can be found. For these sources, not the total URL is given, because this would be compromising the readability of the tables. Therefore, an additional explanation on several sources is given here.
Assignment Zero: This is a project that tempted to collect information on crowdsourcing using crowdsourcing. More on this project can be found in appendix B.10. In total, about 80 interviews were conducted with crowdsourcing practitioners and experts. These interviews can be found at zero.newassignment.net.
Business Week: This magazine also provides information on many cases. Articles can be found at www.businessweek.com.
Crowdsourcingdirectory: This information can be found on www.crowdsourcingdirectory.com. It represents a weblog from the Dutch company CreativeCrowds and collects examples on crowdsourcing.
Crowdsourcing.com: this refers to the weblog of Jeff Howe, the person who coined the term crowdsourcing. Information on many cases can be found, as well as more general information about the topic.
Openinnovators: Some information on the cases can be found at www.openinnovators.net
Springwise: This information can be found on www.springwise.com. This is a Dutch company that collects new business ideas.
Wired: This refers to Wired Magazine, the articles can be found on www.wired.com.
A.1 CROWDCASTING INTERVENTION AND CONTEXT ATTRIBUTES
Organisation
Role of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
99 designs mediator upload; commenting
company initiative individual yes yes registration company company support design springwise; www.99designs.com
Battle of concepts mediator upload; weblog company initiative individual yes no
contributions not public;
registrationcompany company concept general crowdsourcingdirectory;
www.battleofconcepts.nl; interview
Chevroletown
initiative upload company initiative individual yes no no control company company support commercial
video Wired; crowdsourcing.com
Converseown
initiative upload company initiative individual no no registration company company support commercial
videoWired; Howe 2008; crowdsourcing.com
CrowdSpirit mediator upload; commenting
company initiative individual yes yes registration company company development/
design design
Assignment Zero; business week; crowdsourcingdirectory;
crowdsourcing.com; springwise; openinnovators; www.crowdspirit.com
Crowdspring mediator
upload; commenting; voting; forum;
weblog
company initiative individual yes yes registration company company support design
business week; crowdsourcingdirectory;
crowdsourcing.com; springwise
Dewmocracyown
initiative upload; forum company initiative individual no yes registration company company support consumer
goods www.dewmocracyvoltage.com/
Doritosown
initiative
upload; commenting;
forum
company initiative individual yes yes registration
company decides final 5, crowd decides winner
company support commercial video
crowdsourcing.com; www.crashthesuperbowl.com
GeniusRocket mediator upload; weblog; commenting
company initiative individual yes yes
registration, company can
choose whether to directly post or to
screen
company company support design
www.crowdspring.com; www.geniusrocket.com;
https://www.vator.tv/pitch/show/user‐generated‐advertising‐0
Goldcorpown
initiative upload company initiative individual yes no contributions not
public company company production science wikinomics; crowdsourcingdirectory; openinnovators
Google 10^100own
initiative upload; voting company initiative individual yes no contributions not
public
company chooses best 100, crowd decides best 20, company chooses best
5.
company concept general http://www.project10tothe100.com/
IdeaConnection mediator upload; wiki company initiative individual yes yes
registration; sign non‐disclosure
and Confidentiality
Agreement before sensitive
information is shared.
company company development/ design science http://www.ideaconnection.com/
InnoCentive mediator upload company initiative individual yes no
contributions not public;
registrationcompany company development/
design science
wired; Howe 2008; assignment zero; business week;
crowdsourcingdirectory; crowdsourcing.com
Innovation Exchange (IX)
mediator upload; social networking
company initiative individual yes no
contributions not public;
registrationcompany company concept
product, service,
process and business model innovation
www.openinnovators.net; www.innovationexchange.com; www.canadianbusiness.com
Kidekoown
initiative upload company initiative individual no, product
prize no contributions not public company company development/
design designSpringwise;
http://www.kideko.com/pages.php?pageid=18 (last visited 16 dec 2008)
Kraftfoodsown
initiative upload crowd initiative individual yes no
extensive policy; contributions not
publiccompany crowd
concept & development/
design
consumer goods
business week; crowdsourcingdirectory;
www.kraftfoods.com/innovatewithkraft
local motorsown
initiative upload; forum company initiative individual yes yes
registration, screening for following guidelines
company company development/ design design
crowdsourcingdirectory; www.local‐motors.com;
http://mashable.com/2008/04/09/local‐motors/
Mechanical turk mediator upload company initiative individual yes no
contributions not public;
registrationcompany company production general
Assignment Zero; crowdsourcingdirectory;
crowdsourcing.com; springwise; wired; http://bjoern.org/projects/catbook/
NameThis mediator post names; comments; voting
company initiative individual yes no
registration; control by type of contribution
crowd decides winner, company decides
what to do with
outcomes
public support general crowdsourcingdirectory; springwise; www.namethis.com
82
Organisation
Role of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
Netflixown
initiative upload; forum company initiative individual yes yes
contributions not public;
registration
company decides (but based on measure)
company development/ design science www.netflixprize.com,
www.wikinomics.com
Ninesigma mediatorrespond to RFP, work under contract
company initiative individual yes no
contributions not public;
registrationcompany according to
contractdevelopment/
design science www.ninesigma.com
Peugeotown
initiative upload company initiative individual yes no
contributions not public;
registrationcompany company development/
design designhttp://www.peugeot.com/en/design/d
esign‐contest/5th‐edition/2008‐theme.aspx
ReDesignMe RDM challenges
mediatorupload;
commenting; voting
company initiative individual no yes registration company public development/
design design springwise; www.redesignme.com; interview
Shell Game Changerown
initiative upload crowd initiative individual yes no contributions not
public. company crowd concept science
Dutch Innovation Seminar; http://www.shell.com/home/content/global_solutions/innovation/innovativ
e_thinking/game_changer/
Stray Cinemaown
initiativeupload; weblog;
forumcompany initiative individual no yes registration
crowd decides top 5, then
screening by company
publicdevelopment/ design & production
video editing assignment zero; http://straycinema.com/
TalpaCreativeown
initiative upload company initiative individual yes yes
registration (also for reading);
contributions not public
crowd
share between crowd and company
conceptideas for
unscripted TV shows
springwise; www.talpacreative.com
TekScout mediator upload company initiative individual yes no
contributions not public;
registrationcompany company development/
design science openinnovators
Threadlessown
initiativeupload; forum; commenting
company initiative individual yes yes
registration. Participants can choose whether to
submit it for critique before submitting for
scoring.
crowd company development/ design design
wired; Howe 2008; assignment zero; crowdsourcingdirectory;
crowdsourcing.com; springwise; www.threadless.com
Ukgovernmentown
initiative
upload; commenting;
voting
company initiative individual yes no screening company crowd concept government www.showusabetterway.co.uk
Walkers (do us a flavour)
own initiative upload company
initiative individual yes no registration
company chooses final six, which
get produced, crowd decides winner
company concept consumer goods
http://www.walkers.co.uk/flavours/default.html?ver=hi
YourEncore mediator work under contract
company initiative individual yes no
contributions not public;
registrationcompany according to
contractdevelopment/
design science www.yourencore.com
83
A.2 CROWDSTORMING INTERVENTION AND CONTEXT ATTRIBUTES
Organisation
Role of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
Dell's Ideastormown
initiative forum; voting crowd initiative combination no yes registration company company concept consumer
goods
Howe 2008; crowdsourcing.com;
http://sync.nl/prosumptie‐leek‐hype‐maar‐is‐een‐blijvertje/;
http://www.itbusinessedge.com/blogs/tve/?p=381
Gannett's governance malfeasance
investigations in Fort Myers
own initiative forum crowd
initiative combination no yes no control company company production journalism wired; crowdsourcing.com
Gemeente Smallingerland
own initiative forum; voting crowd
initiative combination no yes registration company company concept & development government www.wijbouweneenwijk.nl;
www.dutchcowboys.nl
GetSatisfaction mediator forum; voting crowd initiative combination no yes registration company public customer service consumer
goods Business week
HPown
initiativeforum; voting; social network
crowd initiative combination no yes
registration; moderation on
off‐topic content, but no screening beforehand
crowd public customer service consumer goods
marketingfacts; http://forums11.itrc.hp.com/service/forums/home.do?admit=109447626+1227089567102+28
353475
ideaadayown
initiative forum crowd initiative combination no yes
new ideas are screened before
posting, registration for commenting.
company public concept general crowdsourcingdirectory
IdeaScale mediator forum; voting crowd initiative combination no yes
registration. companies can choose whether to moderate or
not.
company public concept general www.ideascale.com
Instructables restaurant
own initiative forum crowd
initiative combination no yes registration company publicconcept &
development/ design
consumer goods assingment zero; springwise
KLMbluelabown
initiative forum crowd initiative combination no yes
registration; exclusive
membershipcompany company concept industrial
products
www.klmbluelab.nl; http://blog.adforesult.nl/site/entries/marktplein‐dm‐van‐bluelab‐tot‐one‐night‐stand/;
www.adformatie.nl; http://www.favelafabric.com/Downloads/Zakenreisbluelab020
8.pdf
Lego Mindstormsown
initiative forum crowd initiative combination no yes registration company public
concept & development/ design & testing
consumer goods
Groundswell; Wikinomics; http://mindstorms.lego.com/en
g/NZ_dest_2/Default.aspx
Peer to Patent project
own initiative
weblog; forum; voting; upload
crowd initiative combination no, but in
future yes registration company public production patent applications
Howe 2008; assignment zero; crowdsourcing.com;
http://www.peertopatent.org/
philips Live Simplicity forum
own initiative forum; voting crowd
initiative combination no yesregistration (except for voting)
company company concept consumer goods
openinnovators; www.openeur.com
ReDesignMe Design Critique
own initiative
upload; commenting;
voting
crowd initiative combination no yes registration crowd public development/
designconsumer goods
www.redesignme.com; interview
robecoown
initiative forum crowd initiative combination no yes screening (no
registration) crowd public customer service consumer goods
marketingfacts; www.robeco.nl/connect
Starbucksown
initiative forum; voting crowd initiative combination no yes registration company company concept consumer
goodscrowdsourcingdirectory.com; www.mystarbucksidea.com
wepc.comown
initiative forum; voting crowd initiative combination no yes registration company public concept consumer
goods
http://www.technischweekblad.nl/de‐gebruiker‐
innoveert.60144.lynkx; www.wepc.com
84
A.3 CROWD PRODUCTION INTERVENTION AND CONTEXT ATTRIBUTES
OrganisationRole of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
A million penguinsown
initiative wiki; weblog crowd initiative crowd no yes registration crowd public production writing
assignment zero; crowdsourcing.com; springwise; www.amillionpenguins.com
Associated Contentown
initiative
upload; commenting; rating; forum;
weblog
crowd initiative crowd yes yes registration;
screening company company production journalism assignment zero; www.associatedcontent.com
big stock photoown
initiative upload crowd initiative p2p yes no registration;
screening crowd crowd support photo wired; crowdsourcing.com; http://www.bigstockphoto.com/
Cambrian Houseown
initiative
upload; commenting; forum; voting
crowd initiative crowd yes yes registration crowd public
concept & development/ design & production
software
wired; Howe 2008; assignment zero; business week;
crowdsourcingdirectory; crowdsourcing.com; springwise
Citizendiumown
initiative wiki; forum crowd initiative crowd no yes
registration; articles are guided and approved by experts
crowd public production generalwired; assignment zero; crowdsourcing.com; www.citizendium.org
CitySenseown
initiative upload crowd initiative crowd no no type of
participation crowd public production other springwise; www.citysense.com
Clickworkersown
initiative other crowd initiative crowd no no no control
(only pilot) crowd company production science Howe 2008; clickworkers.arc.nasa.gov/top
Cornell University's birdwatching
own initiative other crowd
initiative crowd no yesregistration; participations not public
company public production natureHowe 2008; crowdsourcing.com;
http://www.birds.cornell.edu/netcommunity/Page.aspx?pid=708
CurrentTVown
initiative upload; weblog crowd initiative crowd no no
registration; only broadcast on tv when it is reviewed well
on the Internet.
crowd public production TV Wired; Howe 2008; assignment zero; crowdsourcing.com; springwise
Diggown
initiativecommenting;
votingcrowd initiative crowd no yes registration crowd public production general
wired; Howe 2008; assignment zero; business week; crowdsourcing.com; springwise; crowdsourcing.com;
http://mashable.com/2008/10/08/digg‐bans/; www.digg.com
Drawballown
initiative other crowd initiative crowd no no screening in
registration crowd public production design assignment zero
Dreamstime mediator upload; forum crowd initiative p2p yes yes registration;
screening crowd crowd support photo wired; crowdsourcing.com; www.dreamstime.com
Elementsown
initiativeforum; social network
crowd initiative crowd yes yes registration company public
concept & development/
design
consumer goods
crowdsourcing.com; washingtonpost; springwise
ePluribus Mediaown
initiative weblog crowd initiative crowd no yes registration crowd public production journalism assignment zero;
http://epluribusmedia.net/
fotolia mediator weblog; wiki; forum; upload
crowd initiative p2p yes yes registration;
screening crowd crowd support photo wired; crowdsourcing.com;
GalaxyZooown
initiativeweblog; forum;
other crowd initiative crowd no yes
registration; participations not public
company company production science crowdsourcingdirectory; www.galaxyzoo.org
Gannett's The Enquirer voting problems
own initiative
phonecalls and emails
crowd initiative crowd no no screening crowd public production journalism crowdsourcing.com
google image labelerown
initiative other crowd initiative crowd no no no control crowd public production other images.google.com/imagelabeler
Google PageRankown
initiative other crowd initiative crowd no no no control crowd public production other Howe 2008; crowdsourcing.com
85
86
Organisation
Role of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
IncSpring mediatorupload;
comenting; voting
crowd initiative p2p yes yes
registration, except for voting;
participations are screened for quality,
originality, and source files.
crowd crowd support design springwise; www.incspring.com
IstockPhoto mediatorupload;
commenting; voting; forum
crowd initiative p2p yes yes registration;
screening crowd crowd support photo
wired; Howe 2008; assignment zero; business week;
crowdsourcingdirectory; crowdsourcing.com; www.istockphoto.com
Lego Factoryown
initiative upload crowd initiative p2p yes no registration;
screening crowd crowd other consumer goods
wired; marketingfacts; factory.lego.com; http://mass‐
customization.blogs.com/mass_customization_open_i/2005/08/lego_fact
ory_ch.html
Marketocracyown
initiative voting; forum crowd initiative crowd yes yes registration company public decision making investment
wired; Howe 2008; assignment zero; crowdsourcingdirectory; crowdsourcing.com
NewAssignmentown
initiative weblog crowd initiative crowd no yes
registration; screening of comments.
crowd public production journalismWired; Howe 2008; Assignment Zero; crowdsourcing.com;
allbusiness
NewsTrustown
initiative review; rating crowd initiative crowd no yes registration crowd public production journalism assignment zero; www.newstrust.net
Newsvineown
initiative
post links; commenting;
voting
crowd initiative crowd no yes
registration for posting
links (seeding), not
for commenting and voting
crowd public production journalism wired
NowPublicown
initiativeupload; forum; commenting
crowd initiative crowd no yes
registration for
contributing, not for
commenting
crowd public production journalismWired; Howe 2008; assignment zero;
crowdsourcing.com; www.nowpublic.com
Nufotoown
initiativeupload;
commentingcrowd initiative crowd no yes registration crowd public production journalism www.nufoto.nl
Nujijown
initiative
post links; commenting;
voting
crowd initiative crowd no yes
registration for comments, not for voting
crowd public production journalism www.nujij.nl
OhMyNewsown
initiative
upload; commenting;
voting
crowd initiative crowd yes yes
registration for
contributing, not for
commenting. Two sections: approved or not approved
articles
company public production journalism crowdsourcingdirectory; springwise; english.ohmynews.com
openstreetmapown
initiativeupload; wiki; forum; weblog
crowd initiative crowd no yes registration crowd public production other assignment zero
platialown
initiative upload crowd initiative crowd no no registration crowd public production other assignment zero; www.platial.com
Politicopiaown
initiative wiki crowd initiative crowd no yes registration crowd public production politics assignment zero
Shutterstock mediator upload; forum crowd initiative p2p yes yes registration;
screening crowd crowd support photo
assignment zero; crowdsourcing.com;
www.shutterstock.com; http://submit.shutterstock.com/
Swarmsketchown
initiative other crowd initiative crowd no no type of
participation crowd public production design assignment zero
Theyworkforyouown
initiative other crowd initiative crowd no no
type of participation; registration only for reputation system
crowd public production other springwise; www.theyworkforyou.com/video
TomTomown
initiative upload crowd initiative crowd no no registration crowd company production other
crowdsourcingdirectory; http://www.jouwnav.nl/content/s/441/TomTom‐Map‐Share‐hoe‐+amp‐
wat.html; crowdsourcing.com
Webookown
initiativeupload; forum;
wiki crowd initiative crowd yes yes registration crowd
develop agreement when chosen
for publishing.
production writing Emerce; springwise; www.webook.com
Wikipediaown
initiative wiki crowd initiative crowd no yes
registration only for more complex editing.
crowd public production generalwired; Howe 2008; assignment zero;
crowdsourcingdirectory; crowdsourcing.com; springwise
Wordiaown
initiativeupload;
commentingcrowd initiative crowd no yes
registration; moderation on quality and abuse, but no screening beforehand
crowd public production video springwise
A.4 CROWDFUNDING INTERVENTION AND CONTEXT ATTRIBUTES
87
OrganisationRole of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
A swarm of angelsown
initiative
commenting; voting; financial
company initiative crowd no yes registration crowd public production filmmaking
wired; Howe 2008; assignment zero;
crowdsourcing.com; springwise;
www.aswarmofangels.com
Bandstocks mediator financial; forum; voting
company initiative crowd yes yes
screening; first voting, then raising money
company
rights belong to
bandstocks, only after 5
years reverted.
decision making music crowdsourcing.com;
www.bandstocks.com
BeerBankrollown
initiative
financial, forum and voting
company initiative crowd yes yes registration crowd public decision
making beer springwise; www.beerbankroll.com
Catwalk Genius mediator financial company initiative crowd yes no registration; type
of participation company public decision making fashion Dutch Innovation Seminar;
www.catwalkgenius.com
Kiva mediator financial; forum
company initiative crowd no yes registration; type
of participationborrowers are screened not relevant production development
aid
Howe 2008; Assignment Zero;
crowdsourcingdirectory; crowdsourcing.com;
www.kiva.org
MyFootballClubown
initiativefinancial;
forum; votingcompany initiative crowd yes yes registration; type
of participation crowd not relevant decision making sports
Howe 2008; assignment zero;
crowdsourcingdirectory; crowdsourcing.com;
springwise; http://www.myfootballclu
b.co.uk/
Nvohkown
initiative
upload; voting; forum; weblog;
financial
company initiative crowd yes yes registration crowd not relevant decision
making fashion springwise; www.nvohk.com
Sellaband mediator financial; forum
company initiative crowd yes yes registration; type
of participation company
shared for first 12
months, then crowd
decision making music
Howe 2008; Assignment Zero;
crowdsourcingdirectory; crowdsourcing.com; springwise; wired; http://www.vue‐
royale.nl/index.php/site/crowdfunding_delen_en_verm
enigvuldigen/
Slicethepie mediator financial; forum; rating
company initiative crowd yes yes registration; type
of participation company public decision making music
wired; Dutch Innovation Seminar;
www.slicethepie.com
Spot.us mediatorpost tips and pitches; financial
crowd initiative crowd no no (but option
for future)registration; type of participation
fact‐checking editors before publishing
public, except when news organization claims rights
production journalism crowdsourcing.com; www.spot.us
Boober mediator financial crowd initiative p2p yes no registration; type
of participation company not relevant support p2p loans Crowdsourcingdirectory; www.boober.nl
first giving mediator financial crowd initiative p2p no no control by type
of participation company not relevant support p2p loans crowdsourcingdirectory; www.firstgiving.com
Microplace mediator financial; weblog
crowd initiative p2p yes no registration; type
of participation company not relevant support development aid
Wired; www.microplace.com
prosper mediator financial crowd initiative p2p yes no registration; type
of participation company not relevant support p2p loans crowdsourcingdirectory; Assignment Zero
Zopa mediator financial crowd initiative p2p yes no registration; type
of participation company not relevant support p2p loansCrowdsourcingdirectory;
Assignment Zero; springwise; www.zopa.com
A.5 HYBRIDS INTERVENTION AND CONTEXT ATTRIBUTES
OrganisationRole of the
company
type of
participation
party that takes
initiative
whose
contributions
are used
Financial
rewards
interaction
input control
output control
IP phase
field
Sources
Buzzer mediator commenting; weblog
company initiative combination no no registration company public testing consumer
goodswww.adformatie.nl; www.buzzer.nl
Fluevogown
initiativeupload;
commentingcrowd initative individual no yes no control company public development/
design design
wired; crowdsourcing.com; openinnovators;
http://www.fluevog.com/files_2/os‐1.html
Gannett's The Enquirer 'Get Published'
own initiative upload crowd
initative individual no no screening, but only for abuse. company public production journalism crowdsourcing.com
Have You Got The Nerve TV
own initiative upload; financial crowd
initative individual yes yes registration crowd company concept TVcrowdsourcingdirectory;
springwise; www.thenerve.tv
IBM idea jamown
initiative forum company initiative combination no yes registration company public concept consumer
goods
business week; Bjelland & Wood 2008;
https://www.collaborationjam.com/
MijnProefstationown
initiative
voting; commenting;
upload
company initiative combination
no (except for design contest)
no
registration, except for voting;
contributions not public
company public concept consumer goods www.mijnstation.nl
Mob4Hire mediator commenting; forum; weblog
company initiative combination yes yes contributions not
public; registration company company testing consumer goods
crowdsourcingdirectory; www.mob4hire.com
Nabuur mediator forum company initiative combination no yes registration company public
concept & development/
design
development aid
Dutch Innovation Seminar; interview
Nokia Beta labsown
initiativecommenting; rating; weblog
company initiative combination no no
registration; contributions not
publiccompany public testing consumer
goodsmarketingfacts; afstudeerverslag
Ryzown
initiative upload; forum crowd initative individual yes yes registration crowd company development/
design design springwise; www.ryzwear.com
Scooptown
initiativeupload;
commentingcrowd initative individual yes yes registration company public production journalism
Howe 2008; assignment zero; springwise; dutchcowboys; www.scoopt.com
Talking Points Memoown
initiative weblog company initiative combination no yes no control company public production journalism
assignment zero; crowdsourcing.com;
www.talkingpointsmemo.com
TNOown
initiative
upload; commenting;
rating
company initiative individual no yes registration, except
for voting; screening company company concept general crowdsourcingdirectory
uTest mediatorweblog; report on bugs in software;
forum
company initiative combination yes yes
registration; contributions not
publiccompany public testing consumer
goodscrowdsourcing.com; www.utest.com
88
A.6 CROWDCASTING OUTCOME ATTRIBUTES
O
89
rganisation Composition participantssize and activity of crowd
nr and quality of contributions Company effort Sources
99 designs 25,989 designersmore than 1,2 million designs in 16,631 contests. average of 72 design per contest.
springwise; www.99designs.com
Battle of concepts
Only Dutch; students (2/3) and young professionals (1/3) (under 30). Diverse range of studies
3000 members; 400 more than 1 battle
30 companies; community 1200 contributions (average 40) Good quality.
regional managers; promotion/advertising
crowdsourcingdirectory; www.battleofconcepts.nl; interview
Chevrolet winners were students30000 contributions, 620.000 visitors of the website
more than 30,000 contributions. Many negative contributions, but 22,000 contributions, 629.000 visitors of website, which spent on average 9 min, and 2/3 visisted chevy.com, which was more than Google or Yahoo did, more sales. The most popular video subverted the intent of the campaign. Negative publicity is also good publicity, the new Tahoe started outselling it's counterpart at Ford by a margin of 2 to 1. There were many good quality video's
promotion (in Apprentice episode), provided information (content to work with about the car)
Wired; crowdsourcing.com
Converse 2.000within three weeks 750 submissions, later ran in thousands. good quality
Wired; Howe 2008; crowdsourcing.com
CrowdSpirit
Assignment Zero; business week; crowdsourcingdirectory; crowdsourcing.com; springwise; openinnovators; www.crowdspirit.com
Crowdspring
75% or more are designers. Designers over 130 countries, buyers from 30 countries. Half of the designers who've received awards are US designers. Nearly 500 different designers have been paid by crowdSpring
in the first 45 days 2,900 registered users, well over 750 designers already submitted concepts to projects.
700 buyers have posted creative projects. 6100 designers work on crowdSpring. some have submitted 100+, to 30+ projects each. Average 69 designs per assignment. According to one client, 95% of the designs posted are total garbage.
charge buyers a 15% commission on top of the awards given to creatives. Lot of work for the client having to rate, review and discuss each design as it is posted. Don't screen entries, provide a simple template (it's merely a canvas that allows the entries to be displayed in the best way possible).
business week; crowdsourcingdirectory; crowdsourcing.com; springwise
Dewmocracy93 submissions by 19 november 2008 http://www.dewmocracyvoltage.com/
Doritos 1100 contributions crowdsourcing.com; www.crashthesuperbowl.com
GeniusRocketprofessionals and semi‐pro creators. 83 countries
over 6000. one client: 500 submissions from 69 different creators.
more than 50 assignments by october. One client: 500 submission, 50 were good, 10 were briljant
discussion on crowdspring; www.geniusrocket.com; https://www.vator.tv/pitch/show/user‐generated‐advertising‐0
Goldcorpgeologists, but also students, consultants, mathematicians, militairy officers.
more than 14000 qualified participants dowloaded GoldCorp's data. More than 140 of them submitted detailed drilling plans.
110 drilling targets identified. at 80% of the places that were suggested by the crowds gold was actually found, in total 8 million ounces of gold were found (that's $3 billion). Exploration time shortened by 2‐3 years.
advertising competition; providing information
wikinomics; crowdsourcingdirectory; openinnovators
Google 10^100 over 100,000 ideas submitted http://www.project10tothe100.com/IdeaConnection 5 employees http://www.ideaconnection.com/
InnoCentive
professionals and amateurs. 175 countries, fewer than one third live in North America. More than a third have doctorates. Scientific generalists.
160000
June 2006: more than 30% of the problems posted on the site have been cracked. 75% of successful solvers already knew the solution to the problem.
wired; Howe 2008; assignment zero; business week; crowdsourcingdirectory; crowdsourcing.com
Innovation Exchange (IX)
community of professionals, start‐ups, executives, small businesses, consultants, academics, and other innovative people and organizations. Diverse personal and professional backgrounds.
size is growing by about 500 people a week.
generates five 'quality' solutions each day. 10 employees
www.openinnovators.net; www.innovationexchange.com; www.canadianbusiness.com
Kideko kidsSpringwise; http://www.kideko.com/pages.php?pageid=18 (last visited 16 dec 2008)
Kraftfoods only US residents
have built the capabilities required to integrate the ideas and inventions into winning business propositions. Provide company information to guide participants. Problem can be that it is too complex, as there is an extensive policy. Provide feedback within 8 weeks.
business week; crowdsourcingdirectory; www.kraftfoods.com/innovatewithkraft
Organisation Composition participantssize and activity of crowd
nr and quality of contributions Company effort Sources
local motorsa strong and seasoned team of engineers, designers, machinists, programmers and management
crowdsourcingdirectory; www.local‐motors.com; http://mashable.com/2008/04/09/local‐motors/
Mechanicalturk more than 100 countries 200,000 signed up by november 2007
seems that creative tasks are more popular than intellectual tasks.
Amazon has assembled an enourmous, energetic and highly effective workforce. The challenge lies in finding them something worthwhile to do.
Assignment Zero; crowdsourcingdirectory; crowdsourcing.com; springwise; wired; http://bjoern.org/projects/catbook/; NY times
NameThis
the crowd is active. For TechCrunch, who weren't even planning to really change their name, some 185 ideas were posted and judged. Huge amount of bad names
crowdsourcingdirectory; springwise; www.namethis.com
Netflix179 countries, professionals and amateurs
24 sept 2008: 39381 contestants on 32272 teams.
30287 valid submissions from 3750 different teams provide data www.netflixprize.com; www.wikinomics.com;
Howe 2008
Ninesigma
span industries, geographies and technical disciplines. 60% other companies, 30% academic researchers, 10% research labs
1,5 million experts www.ninesigma.com; businessweek
Peugeot80% of 15‐35 years old, 80 nationalities in first year. 90 nationalities in 2nd year
in 2002 3800 projects
In first year, more than 2,000 submissions and 8000 votes. In 2nd year, 2800 submission and 93,000 votes, 1,150,000 visitors. In third year, 3800 submissions, 400,000 visitors.
http://www.peugeot.com/en/design/design‐contest/5th‐edition/2008‐theme.aspx (last visited 16 dec 2008)
ReDesignMe RDM challenges
50‐60% designers or young professionals, remaining part just users. 60% male, 40% female. 70‐80% Dutch
over 1000 members; 1/10/89 rule seems applicable
9 employees springwise; www.redesignme.com; interview
Shell Game Changer
the amount of ideas from outside the organization is much more than from inside, 70% of projects result out of participation
Dutch Innovation Seminar
Stray Cinema
submissions from 15 different countries (UK, US, Mexico, New Zealand, Australia, Italy, Estonia, Spain, France, China, etc.) amateurs, pro‐am
598 registered users 31 film submissions assignment zero; http://straycinema.com/
TalpaCreative tenthousands 3000 ideas springwise; www.talpacreative.com
TekScout
2000 universities, national labs and companies and thousands of individual scientists. From computer science and math, chemistry, life and the physical sciences, engineering and design, and alternative energy and sustainable products.
thousands
more than 150 projects on the site in march 2008, 200 in october 2008. until october no solutions yet.
openinnovators; www.tekscout.com; http://www.cioupdate.com/article.php/3735681
ThreadlessOrders are 30% international. Professionals and amateurs 700000
annual sales of USD 30 million and profit margin of 30%. Sells more than 90,000 T‐shirts per month. Profit margin of 35%. May 2007: three years ago we increased in size 5 times, two years ago we grew in size 4 times, and last year we doubled.
fewer than 20 employeeswired; Howe 2008; assignment zero; crowdsourcingdirectory; crowdsourcing.com; springwise; www.threadless.com
Ukgovernment
916 ideas submitted. 5 ideas will be built, 5 ideas will be supported, and 4 models they would like ot see built upon.
www.showusabetterway.co.uk
Walkers (do us a flavour)
over 800,000 flavour submissions http://www.walkers.co.uk/flavours/default.html?ver=hi
YourEncore Only from US and UK 30 client companies www.yourencore.com
90
A.7 CROWDSTORMING OUTCOME ATTRIBUTES
Org
anisation company effortcomposition participants
size and activity of crowd
nr and quality of contributions Sources
Dell's Ideastorm
One person is managing ideastorm, but about 40 employees, representing all areas of the company, participate on a team called communities and conversations.
customers 6,000 members
after only 5 days, 1348 ideas and 122388 votes. one of the first and many suggestions were about one topic, Dell listened to the community. Winning the very first Rheingold. 10,000 accepted solutions within 7 months. Implemented or partially implemented 160 ideas, of about 10,000 ideas, 75,000 comments, 660,000 votes.
Howe 2008; crowdsourcing.com; http://sync.nl/prosumptie‐leek‐hype‐maar‐is‐een‐blijvertje/; http://www.itbusinessedge.com/blogs/tve/?p=381; www.ideastorm.com
Gannett's governance malfeasance investigations in Fort Myers
posted hundreds of documents on the sitefor six weeks the story generated more traffic than it had ever received. city responded to the investigations
wired; crowdsourcing.com
Gemeente Smallingerland106 registered members 100 ideas http://www.wijbouweneenwijk.nl/;
www.dutchcowboys.nlGetSatisfaction Business week
HP
official site moderators. Their primary purpose is to help keep the discussion groups running smoothly, answer HP‐directed questions, and report any issues with the service. The moderators contribute to various discussions as they see fit, but they are free to contribute or not contribute. Our moderator staff is drawn from a wide variety of support folks who have knowledge in some, but not all, discussion areas.
marketingfacts; http://forums11.itrc.hp.com/service/forums/home.do?admit=109447626+1227089567102+28353475
ideaaday 3083 ideas published crowdsourcingdirectoryIdeaScale providing feedback www.ideascale.com
Instructables restaurant 9 employees springwise
KLMbluelab feedback, 40 employees involved. business customers
>1000 members in first two weeks, with 350 ideas. 800 after 4 months. Now about 4000 members, no new members allowed.
12 solutions implemented
www.klmbluelab.nl; http://blog.adforesult.nl/site/entries/marktplein‐dm‐van‐bluelab‐tot‐one‐night‐stand/; www.adformatie.nl; http://www.favelafabric.com/Downloads/Zakenreisbluelab0208.pdf
Lego MindstormsGroundswell; Wikinomics; http://mindstorms.lego.com/eng/NZ_dest_2/Default.aspx
Peer to Patent project
contributions useful. by spring 2008, nearly 33000 people had reviewed some 22 patent applications and had submitted 192 instances of prior art. november 2008: 2352 members have submitted 274 instances of prior art.
Howe 2008; assignment zero; crowdsourcing.com; http://www.peertopatent.org/; http://www.todaysengineer.org/2008/Dec/peer2patent.asp
philips Live Simplicity forum
the reactions and opinions on the platform are indeed of high quality. openinnovators; www.openeur.com
ReDesignMe Design Critique
50‐60% designers or young professionals, remaining part just users. 60% male, 40% female. 70‐80% Dutch
over 1000 members; 1/10/89 rule seems applicable
9 employees www.redesignme.com; interview
robeco marketingfacts; www.robeco.nl/connect
Starbucks48 idea partners, which are specially trained to act as hosts of the discussion
75% USA 9% Canada 3% United Kingdom 13% Other. Customers
37000 contributions in first month
nearly 75,000 ideas in less than six months, with many ideas receiving thousands of votes nd hundreds of comments
crowdsourcingdirectory.com; www.mystarbucksidea.com; businessweek
wepc.comhttp://www.technischweekblad.nl/de‐gebruiker‐innoveert.60144.lynkx; www.wepc.com
91
A.8 CROWD PRODUCTION OUTCOME ATTRIBUTES
Or
ganisation company effortcomposition participants
size and activity of crowd nr and quality of contributions Sources
A million penguins over 130 countries
Nearly 1500 people have contributed to the writing and editing, 75000 people have visited the site and there have been more than 280,000 page views.
not a coherent novel with a beginning, a middle and an end, characters came and went, so it was more a series of sketches possibly than one ongoing narrative. There was an overall tone in the contributions. 11000 edits, 1030 pages in total
assignment zero; crowdsourcing.com; springwise; www.amillionpenguins.com
Associated Content55,000 content producers.
over 850,000 pieces of content published
assignment zero; www.associatedcontent.com
big stock photo 2,817,000 stock photos wired; crowdsourcing.com; www.bigstockphoto.com
Cambrian House50 employees dec 2006. May 2008: 25 employees.
May 2008, 35.000 members, 2.25% of members provides ideas.
Dec 2006: 3700 ideas generated, of which 4 implemented. May 2007: 4973 ideas generated of which 206 were turned into businesses although only 72 businesses have more than one participant. May 2008: more than 7000 ideas generated. Only 3 ideas actually commercialized.
wired; Howe 2008; assignment zero; business week; crowdsourcingdirectory; crowdsourcing.com; springwise
Citizendiumuse of expert editors, but also amateur contributors.
thousands of registered contributors, with a few hundred active every month.
struggled to build critical mass. Over 8900 articles.
wired; assignment zero; crowdsourcing.com; www.citizendium.org
CitySense springwise; www.citysense.com
Clickworkers several thousand
first 88000 images analysed in 2 years. The same done in 1 month by several thousand contributors, in comparable degree of accuracy
Howe 2008; clickworkers.arc.nasa.gov/top
Cornell University's birdwatching
11000 people participanted in the first bird count in 1996, in 2007 more than 80,000 people participated
collected well over 100,000 sightings in 2003, but over 1,15 million in 2008.
Howe 2008; crowdsourcing.com; http://www.birds.cornell.edu/netcommunity/Page.aspx?pid=708
CurrentTVWired; Howe 2008; assignment zero; crowdsourcing.com; springwise
Digg
600,000 users, 1% responsible for driving 32% of pageviews, 224.1 million page views a month, of which 156 million come from the US
it accounts for 1% of the total internet traffic in the US. As of october 2008, roughly 10,000 to 15,000 stories are submitted to Digg per day.
wired; Howe 2008; assignment zero; business week; crowdsourcing.com; springwise; crowdsourcing.com; http://mashable.com/2008/10/08/digg‐bans/; www.digg.com
Drawball assignment zero
Dreamstime56,130 photographers, 1,179,914 users 4750646 images wired; crowdsourcing.com;
www.dreamstime.com
Elements 2 employees 484 (29 oct 2008)
it has built a loyal base of customers interested in patronizing the restaurant. Take advantage of a broad range of expertise. Crowdsourcing hasn't saved money.
crowdsourcing.com; washingtonpost; springwise
ePluribus Mediaassignment zero; http://epluribusmedia.net/
fotolia 961,157 photographers 4,888,189 stock images online wired; crowdsourcing.com; www.fotolia.com
GalaxyZoo over 100,000 by dec 2007
every picture of galaxies that was uploaded on the site was analyzed and rated 30 times reaching high levels of validity. Unexpected research results. Results of amateurs as good as professionals.
crowdsourcingdirectory; www.galaxyzoo.org
Gannett's The Enquirer voting problems
about 45 contributions crowdsourcing.com
google image labeler images.google.com/imagelabelerGoogle PageRank Howe 2008; crowdsourcing.com
IncSpring introduction event professionals; pro‐am 663 members; 13 more than 10 submissions; 459 no submissions 815 brands available springwise; www.incspring.com
IstockPhoto 96% amateurs. 45 percent from outside US.
more than a million members, of which 33,000 contributors in jan 2007, 1.8 million members in june 2007 (1.2 regular customers purchasing photographs), 3.6 million in august 2008. 27,000 image contributors by november 2006. 2000 exclusives in may 2007. 35,000 contributors
every 1.3 seconds an image or video is downloaded of the 3.2 million stock images
wired; Howe 2008; assignment zero; business week; crowdsourcingdirectory; crowdsourcing.com; springwise; www.istockphoto.com
92
Organisation company effortcomposition participants
size and activity of crowd nr and quality of contributions Sources
Lego Factoryevaluates all models. Introduction contest
customers, now two designs of 9 year olds
In the 90s Lege was in a bad period, developing the Lego Factory got them back on track. 3000 to 6000 models per week
wired; marketingfacts; factory.lego.com; http://mass‐customization.blogs.com/mass_customization_open_i/2005/08/lego_factory_ch.html
Marketocracy
pro‐am, experience in specific industry. 130 countries and all walks of life.
400 portfolios in m100 or considered. 65000 portfolios in july 2007
400 portfolios that are in or being considered for m100. the average age is 5,5 years, beaten the market, on average, by 19 percent per year. 90% return for master 100 fund, versus 51% of S&P500.
wired; Howe 2008; assignment zero; crowdsourcingdirectory; crowdsourcing.com
NewAssignmentprofessionals and amateurs 500 in the first week
highly satisfying failure, 28% worked according to Jay Rosen, 75% according to Jeff Howe
Wired; Howe 2008; Assignment Zero; crowdsourcing.com; allbusiness
NewsTrust
About 76% of members have a college degree, and about 25% have 5 years or more of news experience. Wide range of political viewpoints
4500 in May 2007 over 3500 new story reviews per month (May 2007). Quality just as good as professionals.
assignment zero; www.newstrust.net
Newsvine wired
NowPublicdistribution deal with Associated Press.
140 countries, 16‐35, skewing on 25‐35 June 2008: 130,000 users They're become a news gathering
service for mainstream media.
Wired; Howe 2008; assignment zero; crowdsourcing.com; www.nowpublic.com
Nufoto www.nufoto.nl
Nujij www.nujij.nl
OhMyNewsfacts are checked by a staff of 40 Koreans
26,000 citizen reporters, 1‐3 million active readers a day (2003). 41,000 citizen reporters (2006)
14 million pageviews a day, citizen reporters make up 80% of all content.
crowdsourcingdirectory; springwise; english.ohmynews.com
openstreetmapmost from UK, Denmark, Holland and Germany
we have maps you can argue are as good as Google's in certain places. We have a certain maturity of toolchain and of community
assignment zero
platial 9 employees
from around the world and have truly diverse interests; 25‐44 years old
millions of entriesassignment zero; www.platial.com; http://www.netmag.co.uk/zine/how‐we‐built‐1/platial‐com
Politicopia active participation assignment zero
Shutterstock amateurs 140,289 photographers 5709192 photosassignment zero; crowdsourcing.com; www.shutterstock.com; http://submit.shutterstock.com/
Swarmsketch assignment zero
Theyworkforyoumost active contributor has made 8000 entries.
two months after launch all 42,018 video clips were fully searchable
springwise; http://www.theyworkforyou.com/video/
TomTom provide software 3,5 million registered users, 16000 new users a day.
10,000 uploads, 15 million downloads in six months.
crowdsourcingdirectory; http://www.jouwnav.nl/content/s/441/TomTom‐Map‐Share‐hoe‐+amp‐wat.html; crowdsourcing.com
Webook
many professionals, but also amateurs. More than 100 countries.
thousands of participants, hundreds of books. 500,000 unique visitors (sept 2008)
first book on market, sales about 1000 copies in september, reviews are positive. Goal for 2008 5 books, goal 2009 15 to 25 books. 20% growth every week in the first 4 months. 20,000 books in development, 250 ready, 4 books published (sept 2008)
Emerce; springwise; www.webook.com
Wikipedia
More than 1500 participating administrators who have earned special privileges to prevent edits on certain articles, usually to stop vandals who have targeted them. Five full‐time employees
over 250 languages
few hundred super‐contributors. Tens of thousands of contributors. 1% actieve gebruikers, 9% reacties, 90% kijker (Ruben Nieuwenhuis op amtenaar 2.0.nl). 1 percent of all Wikipedia contributors are responsib le for writing about half of the 1 billion words in the English version
perceived as very succesful, the topics that tend to have larger fan bases also tend to better covered. Authority problem. 2.2 million entries (aug 2008). 17 articles within 3 weeks, 150 a month later, quadrupled by the end of April, 3700 by the end of August.
wired; Howe 2008; assignment zero; crowdsourcingdirectory; crowdsourcing.com; springwise; wikinomics
Wordialaunch site has 76,000 words and 120,000 definitions, and another 21,000 word thesaurus entries.
springwise
93
A.9 CROWDFUNDING OUTCOME ATTRIBUTES
Org
anisation Composition participants size and activity of crowdnr and quality of contributions
Company effort Sources
A swarm of angels1000 subscribers by july 2007. 1‐3% is active, 10‐25% are critics.
evaluating contributions
wired; Howe 2008; assignment zero; crowdsourcing.com; springwise; www.aswarmofangels.com
Bandstocks 219 artists registered crowdsourcing.com; www.bandstocks.com
BeerBankroll 5 countries by june 2008 50 members after 30 days springwise
Boober Dutch €2,472,899 loaned Crowdsourcingdirectory; www.boober.nl;
Catwalk Genius 29 designers19 designers supported, of which 9 more than $50
Dutch Innovation Seminar; http://www.catwalkgenius.com/
first giving only US non‐profits 1,468,736 people have donated to 19,536 nonprofits over $80 million raised crowdsourcingdirectory;
www.firstgiving.com
Kiva
10% of lenders are from Europe, 90% from Canada and the USA. Most are Caucasian Americans who watch Frontline. The lenders are thirtysomething Americans all across the political spectrum, and generally are wealthy and white. The entrepreneurs come from about 30 countries. The majority of businesses are in Africa.
after three years, 65,000 micro‐lenders and about 9,000 entrepreneurs. Distributed more than $13 million in loans, at an average size of about $500. Circulates $1 million every 7 days
Howe 2008; Assignment Zero; crowdsourcingdirectory; crowdsourcing.com; www.kiva.org
Microplace 29 different countries to invest in 28,097 loans enabled Wired; www.microplace.com
MyFootballClub
After a club has been purchased, a Board will be formed to help run it, containing existing supporters of the club, existing board members, new directors and a member of the MyFootballClub web team. EA sports sponsors the venture.
75% from UK, many from Germany, Scandinavia, Argentina, Canada, USA and Spain, also from New Zealand, United Arab Emirates and Iceland
22500 members after two weeks, 45247 members in nine weeks, 50000 by august 2, 2007
it took less than three months to get 50000 potential owner‐managers to sign up. In november 2007, they agreed to buy a controlling stake in Ebbsfleet United FC.
Howe 2008; assignment zero; crowdsourcingdirectory; crowdsourcing.com; springwise; http://www.myfootballclub.co.uk/
Nvohk 3000 members by july 2008 261 contributions for logo within a week springwise
prosper 830,000 membersdistributed $178 million in loans at an average of about $5000 per loan.
crowdsourcingdirectory; Assignment Zero
Sellabandboth bands and believers from more than 60 countries over 8600 artists
In 2 years, more than 30,000 people have invested over $2.5 million. 29 bands reached the $50,000 mark
Pim Betist teamed up with two former Sony Music big cheeses.
Howe 2008; Assignment Zero; crowdsourcingdirectory; crowdsourcing.com; springwise; wired; http://www.vue‐royale.nl/index.php/site/crowdfunding delen en vermenigvuldigen/
Slicethepie
at launch 800 pre‐registered bands. This converted into 1300 bands signed in the first two weeks.
22 bands have reached $30,000, of which two in the first week.
heavy‐weight PR campaign (40% response)
wired; Dutch Innovation Seminar; www.slicethepie.com; http://www.graphico.co.uk/CaseStudy.aspx?cid=60
Spot.us
the project funded 3,5 stories in its beta phase. One of the biggest examples raised $2500 from 74 different donors
crowdsourcing.com; www.spot.us
Zopa of all ages and income brackets. 76,000 UK members 25 employees Crowdsourcingdirectory; Assignment Zero; springwise; www.zopa.com
94
A.10 HYBRIDS OUTCOME ATTRIBUTES
Orga
nisation company effort composition participants size and activity of crowdnr and quality of contributions
Sources
Buzzer 9 employees www.adformatie.nl; www.buzzer.nl
Fluevog422 contributions, of which 12 are chosen for production
wired; crowdsourcing.com; openinnovators; http://www.fluevog.com/files_2/os‐1.html
Gannett's The Enquirer 'Get Published'
several dozen submissions every day crowdsourcing.com
Have You Got The Nerve TV 650 members by feb 2008 crowdsourcingdirectory; springwise; www.thenerve.tv
IBM innovation jam
some 50 senior executives and professionals spent a week reviewing the output
104 countries 150,000 employees and stakeholders
46,000 ideas, resulted in 31 'big ideas', in total 10 businesses funded as a result of the Jam
business week; Bjelland & Wood 2008; https://www.collaborationjam.com/
MijnProefstation station visitors and travellers
100,000 unique visitors in 2008. 15,000 people have voted in polls. Almost 1000 people have registered for the newslettter. More than 5000 surveys have been filled out.
www.mijnproefstation.nl
Mob4Hire87 countries, 226 networks/carriers 2061 members 400 different handsets in
participants crowdsourcingdirectory; http://www.mob4hire.com/
Nabuurfrom 14 employees back to 5. promotion; training
neighbours from 150 countries.
193 villages and more than 18,000 neighbours. between 30 and 60 neighbours for each village. About 10‐20% of neighbours are active
30 successful projects in 2008 Dutch Innovation Seminar; interview
Nokia Beta labsactive participation, blogging customers
every week, 10,000 returning visitors, 45,000 new visitors. 66% only rate (jan/feb 2008) 39% only rate (mar/apr 2008).
over 1300 submissions each month. Montly blog comments about 400. jan/feb 4% spam, mar/apr 22% spam.
marketingfacts; master thesis report Peltola
Ryz 4024 members 7 contests, 144 entries on average springwise; www.ryzwear.com
Scoopt sales have been modest Howe 2008; assignment zero; springwise; dutchcowboys; www.scoopt.com
Talking Points Memo
While some of them were repetitious, the readers made a lot of interesting and useful observations. Attorneys were fired
assignment zero; crowdsourcing.com; www.talkingpointsmemo.com
TNO136 concepts submitted, 5000 votes and comments
2007: 38000 unique visitors, 200.000 times the concepts were viewed, 5000 votes, 5000 comments, 136 concepts submitted, 88 published, 48 denied.
crowdsourcingdirectory
uTest 148 countries 13747 testers
all ten of the pilot participatns have been converted into paying clients. Revenue grew 150% over the last three months. 185 testing cycles, 10156 bugs detected
www.crowdsourcing.com; www.utest.com
95
96
97
APPENDIX B
In this appendix, all cases that are used for illustrative purposes in the text will be explained further. In total, eleven cases will be presented, including the four cases for which an interview was conducted. These four cases, Nabuur, NS, Battle of Concepts and RedesignMe, will be described more extensively, according to the topics of the interview. The remaining cases will be discussed more briefly, mainly describing the issues that were essential for the arguments made in the main text.
B.1 GOLDCORP CHALLENGE
The goldCorp challenge is organized as a one‐time crowdcasting event. In 2000, the mining company GoldCorp launched the GoldCorp challenge. Inspired by open source software development, the company decided to open up to the public to see whether they could help them found the next six million ounces of gold. They published all their geological data, which is considered a very precious and normally carefully guarded resource, on Goldcorp’s Web site. However, GoldCorp’s CEO acknowledged that this data was of no use to anyone if GoldCorp’s internal geologists were ill equipped to make sense of it. As an incentive, they made $575,000 in prize money available to participants with the best methods and estimates. And indeed, within weeks submissions came in from all over the world. Next to the expected geologists, participants included graduate students, consultants, mathematicians and military officers. Because they had no experience in geology, they came up with new methods and analyses that Goldcorp had never thought of before, such as math, advanced physics, intelligent systems, computer graphics and organic solutions. In total more than 14000 people from over fifty countries downloaded the data and more than 140 of them submitted detailed drilling plans. They identified 110 drilling targets, of which 50 percent was new to they company. At 80 percent of these targets, actual gold was found, in total eight million ounces, which is worth about $3 billion. It was estimated that the challenge has shortened exploration time by two to
n was slashed by over 60 percent in four years. three years, and the cost of productio
Source: Tapscott and Williams 2006.
B.2 INNOCENTIVE
InnoCentive is a crowdcasting platform for all kinds of scientific challenges, which was founded in 2001. Research has found that even for the toughest R&D problems, there are people who already know the solution. The challenge for companies is to gain access to those individuals. At InnoCentive, companies with R&D problems – seekers – post challenges for the community – solvers – to work on. Financial rewards are offers for the best solution. More than a third of the more than 400 problems posted have been cracked. These scientific problems previously remained unsolved. On average, 10 solutions are submitted for each challenge. Until recently, only chemistry and biology problems were posted, and in 2007 engineering challenges were added. Currently there are eight disciplines including computer science and information technology, and business and entrepreneurship.
98
There are about 50 seeker companies, and a community of 160,000 scientists from over 175 countries., only one third coming from the US. Research into InnoCentive has found that the further someone is away from the problem the most likely it is that this person will have a solution to the problem. 75 percent of successful solvers already knew the solution to the problem beforehand. Problem‐solving success was found to be associated with the ability to attract specialized solvers with diverse scientific interests. The more diverse the problem‐solving population, the more likely a problem is to be solved. Especially radical innovation often happens at the intersection of disciplines. However, diversity does not trump ability in this case. While some members of InnoCentive’s labor force might qualify as amateurs in the strict sense that they are not professionally trained scientists, most are passionately curious Ph.D.s who are eagerly willing to expend their free time for the sake of an intellectual challenge. 65.8 percent of successful solvers hold Ph.D.s, compared to about one third of the total solver population. Rewards are generally in the range between $10,000 and $100,000. Although financial rewards proved necessary, intrinsic rewards are also important motivators for successful solvers. The financial rewards are high for crowdsourcing standards, but a seeker company still earns about
twenty times the fee paid to a solver from the solution.
Seekers and solvers remain anonymous to one another throughout the process. InnoCentive assists seeker firms in breaking up and generalizing their problems such that no company specific information nor the nature of the ultimate application is revealed. The problem statement includes the problem’s background and the solution requirements, as well as deliverables which outside solvers are expected to provide. A firm that finds a suitable solution to its broadcasted problem acquires the IP from the solver in exchange for the agreed prize. InnoCentive ensures that solutions viewed but not acquired by seeker firms do not somehow show up in the firm’s IP portfolio, thereby also protecting non‐winning solvers.
Research acknowledges that the innovative capability of such as distributed network is limited by preventing transparency and openness of solutions. Currently it is not possible for solvers to discover complementary approaches through other scientists solutions, since seeker firms are concerned about IP transfer.
Sources: Howe 2008; Lakhani and Jeppesen 2007; Lakhani et al. 2008; Lakhani and Panetta 2008; Lakhani and Jeppesen 2007; Howe 2008; www.innocentive.com.
B.3 DELL’S IDEASTORM
Launched in February 2007, Dell’s IdeaStorm is a typical example of a successful crowdstorming initiative. On www.ideastorm.com, Dell’s customers are encouraged to provide advice on how to improve Dell’s products and services. At the time of writing, more than 11,000 ideas and over 85,000 comments have been posted and more than six hundred thousand votes were cast. However, the results were not that positive from the start, as many contributions were negative. Only after a while the tone got more positive, changing complaints into ideas. One of these ideas was a PC running Linux, the open‐source operating system, instead of Windows. Some 30,000 users voted for this idea, which remained the number one idea on the site for months. With the confidence that came from IdeaStorm, Dell tapped the community to decide on features, support methods, even the type of Linux to install. As a result, the Linux PC went from idea to finished product in just two months, versus the nine to fifteen months it usually takes to ship a new achine at Dell. By May 2007, Dell released three models using Linux. m
99
Dell has taken advantage of the ideas coming from the community. In total, Dell has implemented or partially implemented 160 ideas from the community so far. Dell has been able to get around 6,000 of its customers into IdeaStorm, enough to make the site a productive source of ideas for the company. However, although the total number of members may be large, a smaller subset are active participants in the community.
Management support is considered very important in this case. Dell’s CEO made sure that his executive team was paying attention to the ideas that would impact the business and assigned a cross‐departmental team to review results from IdeaStorm every week. Corporate strategy was deeply involved and helped figure out which ideas got the green light – and who would be responsible for them. The company is keeping the community up to date about the activities concerning IdeaStorm and provides a full summary of ‘ideas in action’ on the site. Since the start, about 160 ideas from the community are implemented. At the start of the initiative, thirty employees were responsible for moderating the site, representing all areas of the company. Currently, the staff is reduced to five employees. While the discussion mainly evolved around
co pproblems at the start, the tone has be me more ositive, focusing on new ideas.
Sources: Li and Bernoff 2008; Tapscott and Williams 2006; www.ideastorm.com; http://www.itbusinessedge.com/blogs/tve/?p=381
B.4 ISTOCKPHOTO
iStockphoto is one of the many microstock photo sites, who sell stock photographs for just a few dollars. The company started out in 2000 as a non‐profit, consisting of a website where designers could share photographs. Hereby they avoided paying for stock photographs and created the possibility to improve their skills. A community of mostly amateur photographers grew up around the site. Because of the rising costs of maintaining the site, the founder decided, together with the community, which includes about 500 people, to start charging a small fee for the photos. Charging a fee of 25 cents, later changed to $1, iStockphoto was undercutting the big stock‐photo agencies by 99 percent. As of February 2008, iStockphoto has 1 million regular customers and a database of over 3 million photographs.
Currently about 50,000 photographers are contributing to iStockphoto, of which about 2000 are exclusive. Only 4 percent of them claim photographer as their chief occupation. However, more than half have had at least one year of formal schooling in art, design, photography, or related creative disciplines. Fewer than 10 percent of contributors earn enough from iStock to live on. One of the top contributors has already sold over 500,000 photos. About 55 percent of iStock photographers is from the US. The mentality of iStock photographers is different from professionals, since they do not have to making a living out of this.
The contributors of iStockphoto are forming an enthusiastic community that share their knowledge and is therefore continually improving their skills. People who only consider photography their hobby – what it just would be a decade ago ‐ can now earn some money and recognition for their work, which gives it meaning. Therefore, iStock photographers consider themselves amply compensated, although they are only paid 20 percent of sales (40 percent for exclusives).
Respecting the community is important at iStockphoto, since the community is the company. The contributors all want a voice, and the organization gives it to them. Any shift in the way the product is priced, sold and marketed needs the community’s approval.
100
The company has experienced double‐digit growth nearly every month of its existence. Recognizing that iStock’s growth came at the expense of its own business, in 2006, Getty Images bought iStockphoto for $50 million. Getty estimated that iStock’s earning would nearly quadruple by 2012, while its earnings from traditional stock offerings would decline from 50 percent of its overall earnings to less than 30 percent.
Sources: Howe 2008; www.istockphoto.com; Assignment Zero
B.5 NABUUR
Nabuur is an organization that uses the crowd for development aid to Third World countries. It was started by Siegfried Woldhek, former CEO of the Dutch part of WWF, at the end of 2001. At WWF, Woldhek noticed that there were increasingly more people who wanted to do something themselves, instead of only donating money, which led to the idea of Nabuur.
Villages in Third World countries that are working on projects can post tasks on the Nabuur website to ask ‘neighbors’ to help them with the project by performing these tasks. Forums are used to let neighbors discuss possible solutions. Nabuur is built on the neighbor concept, where also the name relates to, because people who are neighbors will help each other when their help is needed, even though they do not always know each other that well. This metaphor is very important in all decisions that are made in the organization. Neighbors represent a group of people ready to help each other when needed, because they are connected through a common characteristic, in this case their geographical location. As in the metaphor, neighbors are never really experts, but they are committed and therefore often the right people to solve the task. Like with ‘real’ neighbors, a project is not executed by one person, but by division of tasks among the neighbors, according to self‐organization.
Participants
Because forums are used, the neighbors form a community, as is explained in chapter three. In the case of Nabuur, the social aspect is very important, and it is gaining even more importance. At the start the idea was that the contact between neighbors would be primarily about the content, which is still the case at the moment. However, the idea has emerged to give neighbors the opportunity to connect on a social basis and get to the content part later on. They think it can work both ways, as this is also the case with ‘real’ neighbors. When you don’t know your neighbors and something happens, you will help and the social relation follows. However, if you already have a good relationship with your neighbors, you will also help them when needed. Therefore, social network features were developed, and neighbors are increasingly interacting on topics other than the projects. Besides purely social interaction, there are expertise groups who discuss about their expertise, so that knowledge is shared that could help several villages.
Every village has its own group of neighbors. Most villages have somewhere between 30 and 60 neighbors. The activity among the neighbors differs per period. However, every neighbor can also be active in more than one village. In most villages, there is only a small percentage really active. There are different roles in the community. All villages have a local representative, who manages the project and is the main contact for the village. However, because of practical problems like poor Internet connections and to help local representatives with formulating tasks, each village also has a facilitator. This is a volunteer, someone from the community who has applied to become a facilitator. Neighbors can also perform different roles. For example, some neighbors are committed and active in one village, while others are connected to several
101
villages, recognizing the similarities between them. This shows the community’s ability to self‐organization.
Everyone is welcome at Nabuur, no one is excluded from participation. However, because the website is in English, only English speaking countries will participate. In total, there are 193 villages and more than 18,000 neighbors. In total, neighbors come from 150 different countries. About 10‐20 percent of the neighbors can be considered an active participant, although there is no distinction between contributors and reviewers. The top 5 neighbors are India, Kenya, US, UK and the Netherlands, South Africa usually follows. The top 5 of villages consist of Uganda, Kenya, India, Congo, and Zambia.
One of the things Nabuur has learnt is that being an online organization does not have to exclude offline contact, which is sometimes very valuable. Some groups of neighbors have offline meetings, and some local representatives have visited the Netherlands last year. That the neighbors represent a real community is shown again by the fact that all meetings but one were initiated by the community itself.
Conditions
A solution is often formed through collaboration between neighbors. Experience has taught Nabuur that about 30 people are needed to reach a solution, because it is enough to get multiple ideas without too much redundancy. More than one person is needed for both creating the idea and developing it. The diversity among the neighbors is very important, as diverse ideas are needed to get to a solution. For example, the expertise of developed countries are useful for initiating new ideas, but for the further development of the idea it is useful that at least some of the neighbors are themselves from Third World countries, because they are familiar with the environment and the context of a project. The fact that India and Kenya are present with both villages and neighbors is therefore very helpful.
Although not all tasks are suitable to be performed online, the possibilities are wider than most people initially would think. However, a change in the way of thinking about development projects is sometimes required. For example, when most people think of starting a school, they think about the physical building. Because this would not be possible within Nabuur, the focus is on for example the materials, to acquire books or computers, help the teachers with their lessons, find a way to involve the parents in the program, find funding, etc. This means many projects involve a new way of thinking, although there are also projects that require existing solutions to be found. The type of tasks ranges from Internet research to social work, physical science and design.
An important part of the motivation of neighbors is the fact that they get the opportunity to create a global network. They can share their knowledge and even gain management experience. The fact that more opportunities are created to engage with each other on a social basis and that the importance of offline contact is recognized is likely to increase the motivation of the neighbors to participate.
Related to the motivation of participants is their ability. To enable more people to participate, including people with little time or who need some more guidance, smaller tasks are needed. The organization has learnt that until now there have not been enough small tasks. Furthermore, the present‐day volunteer prefers clear tasks, with a clear start and end, and preferably a clear result. However, the smaller the task, the harder it is to formulate, because it has to be taken out of its context. Therefore, the small tasks are usually not specifically tied to a village, and involve for example Internet searches or translating pieces of text. This represents a
102
challenge for Nabuur, as it is the facilitator’s job to formulate these tasks, but they are tied to the villages.
One of the most important conditions for the projects to be successful is the participation of the local representatives. This shows commitment to the project and the neighbors, which is very important in motivating the neighbors to remain active in the project. Nabuur therefore encourages them to participate to keep the discussions going and to give feedback. Knowing what the impact will be of their work can increase the motivation of the participants. In most cases, the whole village can benefit from the project, not just one person is involved. Previously the organization of Nabuur would participate in the discussions as well. However, because Nabuur strives to be only facilitating, they stopped participating themselves.
With the increasing volume of posts, the need for structure has emerged. Some structuring is provided by tagging, so that neighbors can search for the tasks that are best suited for them, and that neighbors with a particular expertise can be contacted for projects that are suited for them. Because of the amount of content for a project, facilitators write summaries about the discussion, but because there is no way to control this, this doesn’t happen often enough. This enables new neighbors to engage in the discussions that are sometimes running for over two years. Recently, a rating system is introduced, enabling neighbors to structure the posts according to relevance. However, because this is a new feature, its usefulness is not yet proven. Neighbors have to understand the importance of rating and use the feature properly for it to be useful. Although rating could also help villages to see which contributions are considered most useful, the rating is only meant for structuring the content, the villages can choose how to use the contributions. The rating feature also provides a simple reputation system, because the importance of individual’s postings are determined. The need for a reputation systems was identified in a study among the neighbors with a remarkable result, namely that although they did not need recognition or a reward for their work themselves, they could imagine that others might need it.
Control is not needed at Nabuur, there never has been any abuse. An important reason for this is likely to be that there is no money involved. As is mentioned, “transparency is the new control”, when everything is visible, there is no use of posting something that is not useful for the project. As mentioned before, there is some social control, as the organization receives messages from facilitators and neighbors right away when something is posted that does not comply with the rules. An example is a village that wanted to collect money for a project, which Nabuur is not meant for, within half a minute the community warns the organization of this problem.
Efforts
As Nabuur want to be merely a facilitator, most of the functions within the organization are related to the website. Other tasks are related to promotion and training. At one point, Nabuur consisted of 14 employees, now they are back to 5. For example, there were two employees for monitoring purposes, which also included keeping the discussions going and helping with finding solutions. This is not the case anymore, as this is not the intention of Nabuur. Although sometimes it is still necessary, the neighbors are increasingly able to do this on their own.
The recruitment of villages is not hard, they are easy to find because they also collaborate with other organizations, and word‐of‐mouth is important. The recruitment of neighbors is more difficult. Everyone is welcome at Nabuur, although some target groups had to be defined because of promotional purposes. For example, they are now focusing on recruiting more students, as their knowledge and their connections with for example teachers are very valuable. This mostly occurs trough advertisements, although the possibilities of weblogs and social
103
networks are increasingly used. Additionally, they have received funding through Google, which results in many site‐visits.
Because being a facilitator is such a time‐consuming job, it is hard to find enough volunteers for this role. Until now they have managed to recruit sufficient facilitators, but since their objective is to work with thousands of villages, this could become harder in the future. The facilitators receive training, for example in the working of the website, some project management, which includes how to divide a project into smaller tasks, and about cross cultural communication. What works and what does not work is taught based on examples. The local representative also receives a similar training.
Currently, the neighbors do not receive any training. However, Nabuur wants to start with educating the neighbors a little to improve the process and the structure of the site. So for example, when you have a question, do not send an email to the facilitator or the local representative, but post it on the site, so that everyone can benefit from it and also that neighbors can answer each other’s questions. Neighbors can also be educated about the importance of tagging and rating, as this will improve the structure of the content, and therefore facilitates the process for the neighbors.
Finally, collaboration with other organizations is important, not only for recruitment of villages, as already mentioned, but also because for some projects, both online and offline tasks are relevant. Some projects are for example only partly done by Nabuur. Although Nabuur specifically searches for possible collaborations, this is not always welcomes. Some traditional organizations view Nabuur as competition, in the sense of competing for funding. Some of these organizations are trying to initiate something similar for online participation, but then realize this is not as easy as it seems, and therefore respect Nabuur more.
Results
Nabuur’s main objective is to work with thousands of villages and millions of neighbors. As this is not the case yet, they are not yet satisfied with the results. This does not mean that no significant results have been achieved. However, defining the success of Nabuur is difficult. The measure of successful projects is subjective, but this year about 30 successful projects are finished. These projects consists of several smaller tasks. The tasks and the measure of results could be improved, since currently it is not always clear even when a particular task can be considered as finished, which makes it even more difficult to determine whether an entire project is finished.
The villages consider Nabuur a success. Nabuur provides them with a ‘window to the world’. For them, the possibility to connect with other people for help is already very important, since they often do not even have contact with other villages in their region, let alone people all over the world. The local representatives get the opportunity to do something useful for their village, as
ave been for example unemployed. they h
Type
Nabuur is one of the cases that represents a hybrid between crowdcasting and crowdstorming. Because specific assignments and requests are posted, crowdcasting seems the dominant type. However, the website is organized as a collection of forums where people can discuss each other’s contributions, which means in the end a combination of the contributions are used, like in crowdstorming. Similarly, there are no financial rewards for participants, and the community is very important. Nabuur even acts as a social network, as participants are able to friend with each other, and keep a personal profile.
104
Sources: www.nabuur.com; interview with Gerdie Schreuders, marketing manager of Nabuur.
B.6 NS: MIJNPROEFSTATION.NL
The NS has a project to involve the crowd in the development and restructuring of one of the Dutch train stations, Leiden. The project is meant to realize the ideas that have already been developed and to present them to the audience. NS wants active participation from the crowd to be able to learn how to improve the Leiden station, but as main goal to learn for the restructuring of the six international train stations that are being build and will be build in the next five years in Breda, Arnhem, Utrecht, Amsterdam Zuid, Rotterdam and The Hague.
The ideas for these stations were already presented in a book. Although this all looked fine, it was not clear whether these ideas corresponded to the needs of the audience. A project team was assigned to test the assumptions in practice by making station Leiden a ‘test station’ to learn from the experiences for the six international train stations. A smaller project team was put in place to brainstorm about how to make improvements, and started in December 2007 with the website, where a trailer was placed to ask people to participate.
There were several activities in which the crowd could participate. The project team had big plans, but they had not a big finance deposit, so they started out small by providing information via the website and letting people give their opinion about the existing ideas in polls. Besides these closed questions, there were more open ones like “What is the first thing you would change about the station?”. These questions reveal insights in the needs of the travelers and how they experience the station.
An opportunity for a larger initiative emerged when there had to come a plan for the toilets. Experts had provided a plan that was good, except for the background and logo. Initially they planned to hire another agency for this task, but Dirkx suggested to bring in the audience. Within four weeks, a design contest was organized, involving young and creative talent in the project by contributing designs and letting a wider audience vote for these designs. The participants could win €5000, both for the background and the logo.
In February, there will be another design contest. In the new station, there will be several ‘worlds’, where retailers in different categories (such as media, food, or beauty and health) will be grouped. Glass walls will be separating these worlds, on which matching pictures will be placed, which will be designed by the crowd.
There were several other experiments to involve the audience in the development of the station. There has been a forum to let people discuss about ideas for the station. However, people from outside the project team were needed to answer the audience’s questions. Because these people had to do this on top of their regular jobs and they often did not understand the usefulness of the forum, it often took two weeks before questions were answered. Therefore, the forum was discontinued at that moment.
The forum however showed one topic that was very important to the audience, namely the gates. About 80 percent of the conversation was about these gates. They are placed on the station to be able to check people’s tickets, but because the station also acts as an important passage between two parts of the city, it is often used by people without tickets. With the gates blocking the passage, people often have to make a detour of about one kilometer. However, NS can not change this and is obliged to place these gates. The only thing they can do is provide
105
information about why they are there. Therefore, when the forum was stopped, a FAQ about the e qgates was placed on the website, with the opportunity for p ople to add their own uestions.
Another initiative was an information column on the station. The column provides the opportunity to visit the website and give their opinion while being at the station, and to see and experience the station of the future in 3D.
Additionally, they started a weblog. There are ten NS employees who write about their work and experiences within NS on a daily basis. Since four months, also four customers of NS have been given the opportunity to blog about their experiences.
Apart from the website, there are some other experiments done on the station. One experiment was done with the lighting on the station. A similar experiment that is currently running is about improving the flow of travelers through the station. Furthermore, another test will be done in December, which investigates the influence of sound on the station. Although 90 percent of respondents answered negatively to the question whether there should be music on the station, NS sees opportunities to influence the atmosphere of the station.
There are several objectives of this project, such as to inform, investigate, involve and let consumers experience the NS. Another important objective was promotion of the brand NS. Because NS had a rather negative image, NS wanted to show first of all that they are more than just the trains, but also that they work with the latest technologies, use web 2.0 tools and are open to things like co‐creation and transparency.
Participants
There are several target groups and several measures for the community. In the past year, there were 100,000 unique visitors on the site. 15.000 people have voted in polls. Almost 1000 people have registered for the newsletter. More than 5000 surveys have been filled out. The NS also has a consumer panel of 100,000 people, who regularly answer online surveys. 2500 people from this panel have participated in the lighting experiment. In the first design contest, there were 250 design submitted.
The main target group are travelers and other visitors of the station, who are interested in the development of the station. Although the information will also be used for other stations, the main target group are the travelers of Leiden, because they are directly involved in the development of the new station. The target group for the design contests is different, mainly students of art schools were specifically targeted. However, the winning design turned out to be from a professional agency. Because this was not the objective of the initiative, professionals will be excluded from participation in the second design contest.
Because no registration is needed for participating in polls and surveys, it is not clear who are participating and whether the results are representative for the audience. However, there is information available on the participants of for example the lighting experiment. Because the results from this group do not deviate from the results from the polls and surveys, it is believed that the results can be considered useful. For the same reason, not much information is available about the relative activity of the participants. Although it is not researched, the percentages that are found in open source software communities are probably also applicable to this case.
Currently, there is not much interaction between the participants, except for the comments on the weblogs. Therefore, the participants are not considered a community. However, because the main target group are the travelers in Leiden, and much of the activities are organized apart
106
from the website, they participants nevertheless have the opportunity for interaction. Therefore, they could be considered a community because of their common geographical characteristic.
Conditions
Willingness to be transparent is one of the most important conditions to start with initiatives like mijnproefstation.nl. Transparency shows participants that you take them and their opinions and ideas seriously, which will result in serious contributions from their side.
Transparency is not always easy. A major challenge is informing about the planning, because it concerns a very complex and interrelated project, which causes the planning to change continuously. Transparency also involves feedback about the results of a poll or survey, but certainly also about what is done with these results.
NS also considered control issues. To involve the general audience in the design contest, they were able to vote which design they liked best. However, to let the audience completely determine the end result was considered too risky, because they wanted the final design to be innovative. Therefore, they let the crowd decide on a short list of ten designs, additionally giving the NS the opportunity to replace three of these designs. A jury of professionals would then decide on the final design. The voting was anonymous, the jury did not know whose design they were judging.
As a result, one of the designs that was added to the short list by NS was chosen, and it turned out that the winning design was developed by a professional agency. Therefore, in the next competition, professionals will be excluded from the contest. Another important conclusion was that the choice to have separate competitions for the background and logo had not turned out as they had hoped. As the jury did not find any of the logos fit the background, there was no winner for the logo competition. Transparency was also important here, because the jury had to be able to explain their choice and especially the choice not to use any of the logo submissions.
Considering control issues, the prejudices of many companies who are concerned about misuse of the opportunity are not justified. NS has experienced that if you take the crowd serious, they will respond seriously. Although NS has also been cautious about this issue, they have experienced that not much control is needed, because there is no misuse, all contributions are serious. One exception is the information column, about one third of the contributions is not serious, because the column can be used day and night and is not supervised. But these contributions are not open tot the public. Every question is answered individually. Because these experiences, new initiatives like a forum are easier accepted within the organization. The consumer is also involved in the process in increasingly earlier phases of the innovation process, when there are many uncertainties.
Although the crowd can vote on many issues, they do not control the outcomes, NS still decides what is done with the contributions. An important reason for this is that the polls are used to get an idea about the opinion of the audience. The results are used as input for further experiments, after which it is implemented in real life, but still for further testing. For example, in the polls 91 percent of the participants answers they wanted more vegetation on the platforms. Experiments with vegetation were done at two stations, with positive results. Therefore, more vegetation is present at the station now. Often the results of the consumer participation are input for the regular innovation process. Because the NS is such a large organization, it can still take some time before an idea can be realized. Transparency is also important in this case, because the audience needs to know how the process is organized.
107
Because of new technologies, the power is increasingly with the consumer. The organization must make sure that consumers use this power to help you. When you take them seriously, they will help you. Again, transparency is one of the most important things to achieve this. Because the NS shows they respect the opinions of the audience, the contributions are serious and mainly positive. Of course the consumers express criticism, but it is usually constructive.
Marketing objectives also play a role for NS, therefore it is more logical to organize their own initiative than to use a mediator. Although the importance of existing communities is recognized, it is important to get people to your website. Furthermore, there is more control, because you can organize the website the way you need to achieve your objectives. When using an existing community, you have to play by their rules.
Efforts
The project is being run by a small group of people that were enthusiastic about it from the start. They have been taken off their regular jobs to be able to make this a success. Additionally, there are new organizational roles like writing weblogs and responding on forums. Although the NS does not engage in discussions with existing communities, this would be a good thing. This would also require a new function like web care, where someone searches for conversations on the web about the organization and responds when needed.
The project team is however in many ways dependent on the rest of the organization. Especially at the start, there were many setbacks because the organization was not open to this kind of initiatives. Therefore, they started out small and convinced the rest of the organization of the usefulness of the project. Currently, it is much easier to launch new initiatives, because the organization has accepted the project and is more open to new experiments.
Although it was important to promote the project, there was not much budget available. For example, it was too expensive to put posters on the station etc. Because of the construction work, this is easier now because the walls can be used for communications and reference to the website. For the design contest, art school were actively contacted by phone to explain the contest and to ask for participation of the students.
Results
In general, the team is satisfied about the response. The goal was to have about 100,000 visitors on the site in a year, which is about the number of actual visitors. The quality of the contributions is also important, and they are also satisfied about that. But there are also other goals that are important, because it is mainly an experiment. To create support within the NS for initiatives like mijnproefstation.nl and that people within the organization know who to contact when they want to start something similar.
The design contest was only partly successful, as no winner could be chosen for the logo. In total there were 250 submissions of design, which were considered high quality. While some results are very concrete, the results from the weblog and forum are mainly that it shows which topics are considered important by consumers.
No research has been conducted into the success of one method over another, so there are conclusions to be drawn from this. It is probably too early and too experimental to be able to determine whether one method is better suited than the other.
A very important additional effect of the project is that the NS has a more positive image. With the project, they show that the NS is innovative and open to new approaches. Because more
108
interaction is possible with the organization, consumers are more aware of the organization and know what they are working on.
Type
This case represents a hybrid between crowdcasting and crowdstorming. In fact, this case represents a collection of initiatives, of which some resemble crowdcasting, while other are more similar to crowdstorming. The design competition is for example a typical case of crowdcasting, while the forum to discuss ideas for the development of the station is a typical case of crowdstorming. Other parts of MijnProefstation are a combination between the two types, as there are specific questions broadcasted to the crowd, while a combination of contributions are used by the organization.
Sources: www.mijnproefstation.nl; interview with Coen Dirkx, senior consultant communication for NS.
B.7 BATTLE OF CONCEPTS
Battle of Concepts is an organization that acts as a mediator between companies with specific challenges and students and young professionals. Companies can place challenges, named battles, on the website, where participants can upload concepts in response to these challenges, competing for the financial rewards that are divided among twenty winners. The concepts are not public, they are sent anonymously to the company that placed the battle, only after the winners are announced, the company receives information on the winning participants. Besides the financial reward, winners receive battle points, which are used to rank the participants.
Participants
Battle of Concepts has a specific target group, namely students and young professionals (younger than 30). Currently, more than 3000 people have registered, of which 460 have been in the top 20 of at least one of the battles. About two thirds of the participants are students, while the remaining one third are young professionals. According to Dekkers, working with a specific target group makes the battles easier to sell to companies.
Because the concepts are not public, and there is no opportunity for interaction between the participants, the participants do not form a community. However, the organization already has a plan to change this, but according to Dekkers it is too early for any changes, because most companies are not ready for this type of open communication.
The target group is from a range of studies, which accounts for the diversity in the concepts. There is not one specific group responding to specific requests, people with different types of perspectives contribute concepts for the same battle. For example, one battle was placed by a company who had already brainstormed about this issue internally before broadcasting it to the students. They were surprised that there were still five concepts with totally different perspectives that they had never thought of themselves. These five concepts all came from participants with a different background. This diversity is very important for the success of Battle of Concepts. This is also shown in the ranking of universities and study associations, there is not one ‘expertise group’ that is always on the top of the list, because they are all important.
The fact that the target group consists of students is often an important reason for companies to decide to place a battle. Apart from the possibilities in recruitment, companies are provided
used (first) for other types of challenges than the current battles.
To use the participants not only for the concepts but also for the further development, companies are encouraged to invite the winners to present and further discuss their concepts. Dekkers expects that increasingly more battles will continue with an offline event for further development. Although it was expected that participants would not be interested in these offline events, they have turned out to be provide an extra incentive for the participants as well as the companies.
109
with an opportunity to present themselves in a new way. For example, by placing a battle the company sh it is innovative, in both products and innovation processes. ows that
Conditions
One of the important conditions that are important in a battle is that is works best in the concept phase of the development process, when a company is still crowdstorming. The main goal of Battle of Concepts is to provide companies with a collection of ideas. The diversity of ideas is therefore very important to companies.
The battles generally have a technical or social focus, although this is not a conscious choice. Because Battle of Concepts is still in an early phase, companies can decide what type of battle they would like. There have not been any battles that were not suited for Battle of Concepts. When the number of battles increases, there are plans to categorize the battles, also providing an opportunity for companies to collaborate in placing battles.
The fact that there is a financial reward is definitively important. Although the amount of money does not seem to play a role, there have been two battles without a financial reward, and only ten concepts were uploaded, compared to an average of 47 concepts. However, there are also other motivations that are important for the participants. Besides a financial reward, all twenty winners also receive points for the battle ranking. For some participants, this reputation is very important, especially when they start winning battles, they want to get even higher in the ranking. Some students mention the battle ranking in their CV to show that they are creative and innovative. While the ranking does not imply any other benefits, Battle of Concepts has ideas to change this. Prizes like an iPod or a holiday could be given away for people with a certain amount of battle points. There are also plans to give members with a high ranking access to a battle club, where they can interact and can be involved with the organization of Battle of Concepts. However, while these motivations are important, the content of the battle itself is most important, as participants should be inspired by the description and encouraged to participate.
The fact that Battle of Concepts works with a specific request, instead of a more general call for ideas is important for the quality of the concepts. There are few people who can come up with a good idea from scratch. Much more people can come up with a good idea based on a request, because they can get inspired by the request itself.
Dekkers realizes that people can also get inspired by other people’s contributions and by interaction, but in his experience companies are not ready for this, because this would require more openness. This would also create intellectual property issues, because competitors could use the ideas when the crowdstorming would be done online. The fact that the concepts are not public is an advantage for many companies. Although there already is a plan to provide possibilities for interaction on Battle of Concepts, it is better to wait with the implementation, because it is expected that this would not be saleable to companies. The ideas for a community approach would be more suited for creative tasks, like designing flyers or making commercial videos, because this causes no intellectual property issues. Therefore, the plan will probably be
110
Providing feedback about what is done with the results is encouraged, because it is perceived as a motivator for the participants. This is generally done through the battle blog, a weblog where Battle of Concepts itself can let the participants know what is going on within the organization, and for example when Battle of Concepts is mentioned in media. Currently, it is increasingly used by companies that have placed the battle, to inform participants about their experiences
oand outc mes.
Another important success factor is the format in which the participants can upload the concepts. Currently, they can choose between a text format or a PowerPoint format. While the text format is mostly used, companies prefer the presentations, because it requires participants to be more to the point and use visual representation, and it takes less time to go through all the contributions. Therefore, the participants will be asked increasingly more to provide a PowerPoint presentation instead of a text format.
There are a few reasons why using Battle of Concepts would be a logical choice for companies, compared to setting up such an initiative for themselves. First of all, companies should not underestimate the money and effort that is needed to develop a website and form a community. Second, you need a large enough target group, because often only a very small part of this group will respond. The target group of Battle of Concepts is about 950,000 people, of which 50,000 have visited the site, of which 3,000 have registered. Although participation is relatively easy, it still requires a significant effort from participants, because they have to visit the site, register, find an interesting battle, have good idea, write a concept about this idea and then upload it. Because several companies place battles, the chances of finding an interesting battle are better than when a company would launch its own initiative. Finally, there have been only serious contributions on Battle of Concepts and they are written by higher educated people, so the quality is relatively good. With a community of customers, especially when the request is less specific, the quality of the contributions would probably be lower and the amount of contributions which you have to go through that are not really useful is likely to be higher.
Efforts
Recruitment of the students was not easy. They started with directly contacting universities and colleges, but because it was difficult to directly reach the students and every institution was organized differently, this was not very effective. Currently, there are regional managers, which are students who use their network and build contacts within their own region. For each new battle, promotion is done by these regional managers.
Study associations are very important, because they can directly reach the students and it is possible to organize activities around the battles. For example, one study association has done a battle event, for which Battle of Concepts sponsored the association for each registration. Because this increased the quantity but not the quality of the participation, they want to organize another battle event, but instead providing sponsoring for each top ten concept.
Selling battles to companies is considered the main activity for Battle of Concepts. They started with several launching partners, who each signed for four battles, and also get the opportunity to promote themselves for example by placing their logo on the website. Generally, the first contact is with recruiters. But convincing them is only the first step, as other departments are needed to develop the battles. Therefore, these other departments are contacted more directly, or people from both recruitment and innovation are involved. It takes about six months before a battle is really posted on the site.
111
Results
Since the start of Battle of Concepts in June 2007, the results have been promising. Until now, 28 organizations have placed a battle, of which eight have placed more than one, resulting in a total of 46 battles. Currently, this represents about one new battle each week, hopefully this will soon increase to two new battles a week.
The average number of concepts that are uploaded for a battle is 47, with a minimum of 10 and a maximum of 122. The concepts are generally of high quality, even surprising to some of the companies that have placed the battle. In general, about 30 to 40 percent of the concepts is really good, about 20 percent is considered not useful in itself. However, often there are some trends that can be seen in the concepts, so every contribution can be important in finding this trend. This information is very important for companies, because these trends can confirm the direction that the company has already chosen or can cause a change in direction when the concepts suggest something different.
The quality has also increased over time, people learn from participating in several battles. The number of twenty winners was chosen because it was expected that about twenty contributions would be useful. For the first battles, it turned out that ten would be a better choice. However, just before this change would have been implemented, the quality of the contributions had improved so much that even more than twenty concepts were considered very useful. The quality of the concepts is often so good that even the skeptics within the organization become very enthusiastic. Some are surprised about the results, one company even asked whether the concepts were not made by professionals.
Besides the concepts itself, there are some other important results for these companies. First of all, marketing is very important. Each battle is read about 4‐5000 times, and downloaded 1000 times. These are all people who take the time to read about a company in detail, a result that is often not achieved with traditional advertising. Furthermore, there are changes within the organization, because many people are involved in the whole process and they can see how good the results are.
Type
Battle of Concepts is a typical crowdcasting case, since specific challenges are broadcasted to the crowd. The initiative is organized as a competition with a financial reward and a specific deadline. Contributions are uploaded without publishing them on the website, sending them directly to the company posting the challenge.
Sources: www.battleofconcepts.nl; interview with Joost Dekkers, founder of Battle of Concepts.
B.8 WIKIPEDIA
Wikipedia is a web‐based user‐generated encyclopedia that runs on a wiki, allowing multiple users to share responsibility for creating and maintaining content. It is one of the most well‐known examples of crowd production. Wikipedia is the largest encyclopedia in the world, offered for free, and created entirely by volunteers. In less than seven years, Wikipedia has expanded to offer more than six million articles in over 250 languages. While the accuracy is certainly not 100 percent, changes are that whatever you’re looking for, you’ll find it described there.
112
meet Wikipedia’s standards can be edited, locked, or nominated for deletion by users.
Although the accuracy of Wikipedia is a topic of much discussion, one study has shown that Wikipedia contains four inaccuracies per entry compared to Britannica’s three. Although Britannica claims that Wikipedia’s errors are much more serious, these have already been
Wikipedia started out as Nupedia in 1998, an encyclopedia that was put together by expert volunteers, and would be available on the Internet for free, inspired by open source. To make sure the content would be reliable, a time‐consuming seven‐step review process was implemented. After a year, only a 24 articles were posted on the site. When in 2001 the founders heard about wiki software that allows an unlimited number of users to create and edit text on a single webpage, they were soon convinced to start using this software for a new encyclopedia, named Wikipedia.
Within three weeks, contributors had created seventeen articles. That number jumped to two hundred a month later, and quadrupled by the end of August. By the end of the year, Wikipedia consisted of eighteen thousand articles. Throughout the first years of its existence this exponential growth continued until only recently reaching a plateau.
About ten percent of the one million registered users have contributed ten or more entries. Within this group, a smaller group of about five thousand Wikipedians form a hard‐core group. A survey among this group has shown that the average level of contribution among these people is 8.27 hours per week. Studies have found that 1 percent of all Wikipedia contributors are responsible for writing roughly half of the 1 billion words in the English version. Another study has shown that the 1/10/89 rule of thumb from open source software is also applicable to Wikipedia.
Wikipedians really form a community, since there are community norms and values. The organization has facilitated the community aspect by including social networking features. The organization shows their understanding of the importance of the community by involving them in organizational decisions. For example, since the community resisted the idea of an advertising model, the organization has decided to find other sources of funding. Some of the community norms are stated explicitly, such as the aim to be unbiased and to present all points of view sympathetically.
Wikipedia self‐organizes around areas of expertise. In practice, Wikipedia entries are rarely written by one person, but by a small sub‐community within the larger group of contributors. Wikipedians discuss what should be included in the articles on the ‘talk pages’ that are parallel to each page.
Specificity and simplicity have been important for the success of Wikipedia. When Wikipedia founder Jimmy Wales was asked why his online encyclopedia had done so well, he said it was because everyone already knew what an encyclopedia entry was.
As of early 2007, Wikipedia employed only five people. By contrast, the Encylopaedia Britannica was written by over four thousand paid contributors and one hundred full‐time editors. Organizational tasks include administering pages, developing software, finding copyright‐free photos, moderating conflicts, and patrolling for vandalism. With only five paid staffers, volunteers perform most of it. Volunteers maintain watch lists of hundreds of articles and photos that they monitor every time a change is made. They help ensure the accuracy of editorial changes and can quickly remedy any vandalism. On occasion, ‘edit wars’ break out, in which users repeatedly reverse each other’s changes. In these rare cases, a Wikipedia staffer makes the final judgment. Wikipedia has recently moved to freeze entries that are obvious magnets for vandalism and tampering, such as political topics. Controversial entries that fail to
113
corrected, while those in Encyclopaedia Britannica remain. Another study has shown that a randomly inserted obscenity on Wikipedia is removed in an average of only 1.7 minutes. Each Wikipedia article has been edited an average of twenty times, and for newer entries that number is higher.
Wikipedia’s relevance and power can be shown by an illustrating example. When in July 2005 four bombs exploded in the transportation system in London, it took only 18 minutes for the first Wikipedia entry on the subject to emerge. By the end of the day, over 2500 users had created a comprehensive fourteen‐page account of the event that was much more detailed than the information provided by any single news outlet.
Sources: Nov 2007; Howe 2008; Tapscott and Williams 2006; Li and Bernoff 2008; Bughin et al. 2008.
B.9 REDESIGNME
After working several years learning and teaching about the usability of products, Maxim Schram wanted to use this knowledge to help improve product designs for companies. Originally, RedesignMe was used to collect examples of poorly designed products to use as illustration for both students and companies. RedesignMe was launched in June 2007 as a non‐profit organization, providing a platform for people to express their frustrations about poorly designed products and the opportunity to redesign them. But as increasingly more companies contacted RedesignMe to post their products on the site to let them be redesigned, a business model was developed and in March 2008 the new RedesignMe was a fact. It now includes two parts, design critique and RDM challenges. Design critique represents the non‐profit model, where designers and product users can brainstorm about product redesigns, without the involvement of any company. RDM challenges are posted by companies that need a redesign for a product, organized as a competition, but where it is still possible to respond to each other’s work. This works a little like a weblog, where someone can post something, including visual representation, and there is the possibility to post a redesign or to write a comment.
Participants
At the time of the interview, there were over 1000 members. 50 to 60 percent of them are designers or young professionals, while the remaining 40 percent are just users of the products that are posted. 60 percent is male, 40 percent is female. 70 to 80 percent is Dutch, the remaining part mainly consists of US and Asian participants.
Although the redesigners represent a community, because they interact with each other, the interaction is primarily about the content, and is not socially oriented. The possible interaction is therefore not directly a reason to participate.
Many people are not directly interested in the RDM challenges, or receiving rewards, they are just searching for a place to use their creativity, and to express their frustration about products. On the other hand, some people are specifically interested in the challenges, for example because they are looking for interaction with the manufacturer and to receive feedback on their designs. However, the people that emerge as good designers from design critique, are often also the ones in the top of the RDM challenges.
Currently, most participants do not care so much about what is done with their redesigns, they are already pleased that they can express their ideas somewhere. However, Schram expects that
114
redesigns, not for comments.
With the RDM challenges, it is important that companies are participating and giving feedback. First, the feedback was going through the organization of RedesignMe, and there was no direct contact between the participants and the company that posted the challenge. This is now encouraged, by giving the company an account to directly respond to the posts of the participants. However, these companies are responsible for their own success. Schram advocates the introduction of a community manager to be responsible for the communication
this will not remain to be the case, as participants will increasingly see companies making a profit from their ideas. However, because the redesigns are only considered concepts, the company still has to deal with a significant risk in further developing this concept. Therefore, it is not always fair to conclude that companies take advantage of the crowd. So as long as the crowd is not concerned about intellectual property, RedesignMe will not change its model. Currently, all the content is public property, anyone can use the redesigns and no one can claim the rights. Another reason for this is that the complexity of participation increases when intellectual property issues have to be dealt with, which could demotivate participants.
One of the most important reasons that RedesignMe does not consider any changes in this direction is that most companies do not seem to be ready to really integrate the crowd in their development process. Companies are therefore mostly using RedesignMe for marketing purposes, to show customers that they are listening. In reality, they are indeed listening, or at least hearing, but most companies do not know how to incorporate the ideas and designs. But these are even the companies that realize that co‐creation, open innovation and crowdsourcing are interesting concepts that they need to work with. Many other companies still need to make this step. They need to change their way of thinking about product development to be able to effectively use crowdsourcing. Until then, there is only a small group of companies that are expected to participate in RedesignMe.
Conditions
There are some challenges involved in making RedesignMe work. First of all, designing a product online presents challenges with taking products out of their context. For example, the challenge from DECT that is currently online is about a product that does not yet exist. For many people it is hard to imagine such a product, which can present limitations to the contributions that are possible. The fact that RedesignMe works with visual representations is certainly helping, but still this would be much easier to do offline than online. However, despite the challenges that working exclusively online presents, this is an opportunity that companies can almost not resist, because the information that can be achieved with letting such a large group of designers think qualitatively about your product is much more and better than could be achieved with for example a focus group.
Given that companies understand the concept of crowdsourcing and want to engage the crowd in their activities, there are two reasons to use RedesignMe. The first is quite general, because when a mediator is used, a company does not have to develop the website on their own, or to start a community. What is specifically related to RedesignMe, is that a company can not only access an existing community, but a community of designers. RedesignMe also helps companies formulate their challenge. The most important criterion is that participants should understand the challenge in about 30 seconds. So it is important to keep it simple.
Companies put much faith in RedesignMe and hand over much of the control over the process. Companies posting challenges do not feel the need to control the crowd, they are not concerned about for example negative comments about their products. RDM does not screen the results, according to their experience this is not needed, registration is only needed to post the
115
participants are discussing more themselves.
The process of recruiting the first few hundred participants was quite natural. The concept of RedesignMe was picked up by some magazines and newspapers, which triggered people who were already familiar with ideas like co‐creation to register. To get beyond this group, more effort is needed. Currently, standard marketing tools like Google AdWords are used, as well as activities like handing out flyers on exhibitions. Collaborating with companies can be very
with the crowd. This could be someone from public relations or customer service, who could take this on as a part time job. Companies do not realize this, they really outsource this task to the mediator, in this case RedesignMe. Feedback afterwards, about what is done with the contributions, is also encouraged. For example with the Vodafone challenge, of which the results are currently being processed, there is an arrangement that information about which ideas will be used and how will be shared with the community. This is considered an important incentive for the participants.
The fact that RedesignMe allows interaction between the participants, even in the competitive format, is important. According to Schram, it is interesting to see the interaction, because the designs are improved. Participants also challenge each other, especially in the design critique. Therefore, it was important to keep the design critique when the RDM challenges were added, because this was an important reason for participants to visit the site. Additionally, for branding a community is essential. So while companies increasingly understand the concept of crowdsourcing, the use of a community for marketing purposes is in many cases an important part of the deal.
RedesignMe intentionally focuses only on product design and on a community of designers. The reason for this is explained with a metaphor of a specialist store compared to a supermarket. Before there were supermarkets, there were specialist stores like bakeries and butchers, and when is became possible to buy the same products in a supermarket, it took a while for people to get used to this. The specialist stores could also offer better quality, because they could focus on a small range of products. With crowdsourcing, the same principle can be applied. In the future, it could be possible that one organization manages all kind of topics and participants, but people first have to get used to it and it is easier to do this with specialist stores. Supermarkets can be overwhelming, because there is too much choice. RedesignMe acts as a specialist store, and thinks this will be the best approach for the crowd, the organizations and RedesignMe itself. Because crowdsourcing in general is new, it is better for organizations like RedesignMe to specialize in one area, to learn how it works, before moving into a broader area.
Although RedesignMe focuses on concepts for product design, Schram can certainly imagine that crowdsourcing can be applied in other phases of the process, but this just needs time. For example, there are already organizations who provide tools for individuals to make a prototype of their design, which enables the crowd to be involved in the further development of the concept. Similarly, it can be imagined that increasingly more parts of the process that are traditionally done by companies, can now be performed by private individuals. Marketing functions are also changing because of Web 2.0, as the involvement of the crowd often already functions as marketing. Eventually, one could question the relevance of companies in general. One of the main challenges in this view are the many risks that companies usually have to deal with in the entire process. These are issues that individuals can not handle by themselves.
Efforts
RedesignMe has nine employees, which are responsible for, among other things, the website design and marketing and sales. At the start, the organization regularly participated in discussions. Besides the fact that there is less time to do this now, there is less need, because the
116
useful, as it could increase the number of participants significantly by accessing their database of customers. Although everyone is allowed to join, RedesignMe prefers targeting specific groups. This includes designers, but also regular users. Although on the one hand it would be best to reach out to as many people as possible, just asking people on the street would probably not be very effective.
There are also some technical challenges. For example, the choice for RDMs, which participants can use to buy products, instead of a financial reward has to do with the technical challenges that come with payment. This is an option for the future.
Type
The design critique and the RDM challenges are considered separate cases, because their characteristics are fundamentally different. Design critique is considered crowdstorming, because there is continuous input from the participants, while the results represent a combination of the products, the redesigns and the comments. The RDM challenges section is considered crowdcasting, since specific challenges are broadcasting to the crowd. Interaction between participants is encouraged, which is not surprising since it concerns creative tasks. Although this could be problematic because the participants could consider it unfair, this has not caused any problems so far.
Source: interview with Maxim Schram, Founder and CEO of RedesignMe.
B.10 ASSIGNMENT ZERO
Assignment Zero is the first project of NewAssignment, a platform to spark innovation in citizen journalism, which is considered a typical example of crowd production. This twelve week project asked the crowd to investigate and report on crowdsourcing. A challenging project, with the objective of “producing the most comprehensive knowledge base to date on the scope, limits and best practices of crowdsourcing”, as well as finding out whether crowdsourcing could be a successful method for this kind of journalism. Although the knowledge base consisted of seven original essays and some 80 interviews (much of which is actually used in this study), the real value of the project was learning about how crowdsourcing works in practice.
First of all, the topic was too vague, causing confusion among participants instead of inspiration. Second, topics were developed by the organization for the crowd to work on, according to their interests, but the crowd had different interests. Third, there was no one to welcome the participants and only one person available to contact with questions. This required the participants to organize the whole project themselves. While there were about 500 potential contributors after the first week, these problems caused them to drift away.
It was learned that much more effort is needed to engage and inspire the crowd, such as having a clear request and having people welcoming and engaging the crowd. Therefore, around 30 volunteer professionals editors were assigned to manage various topics. This however presented new problems, because these volunteers had little experience in online collaboration. Therefore, they needed to be trained. About half way through the project, more possibilities for interaction were added, including social networking features and a forum for each topic. This immediately resulted in more active participation and collaboration, showing the importance of creating a community. Then the organization changed its focus from producing feature stories to interviews, as is appeared that the crowd was far more interesting in talking with people they admire and respect, instead of writing a story themselves. Furthermore, the much clearer
117
request of interviews instead of more general stories influenced the participation of the crowd significantly. The ability for self‐organization resulted in high quality results, as professionals often have limited time or commitment for advance research.
Sources: www.newassignment.net; Wired magazine, 16‐7‐2007.
B.11 THREADLESS
Launched in 2000, Threadless involves a website where people can submit designs for a cool T‐shirt and vote for their favorite shirt. The winner gets a free T‐shirt bearing his of her winning design, and everyone else gets to buy the shirt. Since the start, the revenue has nearly doubled every year. Threadless can be considered a typical example of crowdcasting, since the assignment of designing a T‐shirt is very clear and specific, and the winners receive a financial reward.
Threadless currently receives some thousand designs each week, which are voted on by the community, now six hundred thousand strong. Every week, nine shifts are selected from the top hundred to print. The winning designers are paid $2000 in cash and a $500 gift certificate, and Threadless keeps all the intellectual property. Each design sells out, since the community can not only rate the designs, but they’re also able to check an ‘I’d buy it!’ box. The community is important for the company, in fact, it is the company. Threadless shows its commitment to the community by giving them the opportunity to determine the company’s inventory.
According to the top designers, it is not about the money, it is mainly about reputation. Designing for Threadless is addictive, especially for people at design schools or companies where there are strict guidelines for the designs. Threadless gives participants the opportunity to use their creative potential.
Threadless currently sells about 90,000 T‐shirts a month. Threadless spends $5 to produce a shirt that sells for between $12 en $25. They don’t need advertising or marketing budgets, as the community performs those functions admirably: designers spread the word as they try to persuade friends to vote for their designs, and Threadless rewards the community with store credit every time someone submits a photo of themselves wearing a Threadless shirt or refers a
irt. friend who buys a sh
Source: Howe 2008.