HCI

People have interacted with computers fromthe start, but it took time for humancomputerinteraction (HCI) to become a recognized fieldof research. Related journals, conferences, andprofessional associations appeared in the 1970sand 1980s. HCI is in the curricula of researchuniversities, primarily in computer science, yetit has not coalesced into a single discipline.Fields with researchers who identify with HCIinclude human factors and ergonomics, infor-mation systems, cognitive science, informationscience, organizational psychology, industrialengineering, and computer engineering.

This article identifies historical, conceptual,and cultural distinctions among three majorresearch threads. One thread extended humanfactors or engineering psychology to comput-ing. Another developed when mainframesspawned business computing in the 1960s. Thethird, focused on individual use, arose withminicomputers and home computers and bur-geoned with personal computing in the 1980s.

Although they share some issues and meth-ods, these research efforts have not converged.They emerged within different parent disci-plines, at different times, and comprised differ-ent generations of researchers. Approaches,attitudes, and terminology differed. Twocom-puter operation and information systems man-agementembraced the journal-orientedscholarly tradition of the sciences; the thirdcomprising cognitive and computer scientistshas placed greater emphasis on conferencepublication. In addition, each thread initiallyemphasized a different aspect of computer use:mandatory hands-on use, hands-off manageri-al use, and discretionary hands-on use.Designing for a use that is a job requirementand designing for a use is discretionary can be

very different activities. These often unvoiceddistinctions contributed to the current state ofHCI research and may shape its future.

Humantool interaction at the dawn ofcomputing

Highly specialized tools were developedthrough the centuries to support carpenters,blacksmiths, and other artisans. However,efforts to apply science and engineering toimprove the efficiency of work practicesbecame prominent only about a century ago,when time-and-motion studies exploitedinventions such as film and statistical analysis.Frederick Taylors principles of scientific man-agement1 had limitations and were satirized inCharlie Chaplins film Modern Times, but theywere applied successfully to assembly line man-ufacturing and other work practices.

World War I training requirements acceler-ated efficiency efforts in Europe and the US.World War II prompted intense interest in engi-neering psychology as a result of complexequipment used by soldiers, sailors, and pilotsthat tested human capabilities. Aircraftergonomic design flawsfor example, in theejection systems escape hatchled to thou-sands of casualties. After the war, aviation psy-chologists created the Human Factors Society.Two legacies of World War II were awareness ofthe potential of computing and an enduringinterest in behavioral requirements for designand training.2

Early approaches to improving work andwhat at the time were called manmachineinterfaces focused on nondiscretionary use.Assembly line workers were hired to use a sys-tem; pilots were given planesneither had achoice in the matter. If training was necessary,

2 IEEE Annals of the History of Computing 1058-6180/05/$20.00 2005 IEEE

Three Faces of HumanComputerInteractionJonathan GrudinMicrosoft Research

Humancomputer interaction is considered a core element ofcomputer science. Yet it has not coalesced; many researchers whoidentify their focus as humancomputer interaction reside in otherfields. I examine the origins and evolution of three HCI research foci:computer operation, information systems management, anddiscretionary use. I describe efforts to find common ground andforces that have kept them apart.

the workers and pilots were trained. Researchgoals included reducing training time, but mostimportant was eliminating errors and increas-ing the pace of skilled performance.

Three roles in early computingENIAC, arguably the first general-purpose

electronic computer in 1946, was 10 feet tall,covered 1,000 square feet, and consumed asmuch energy as a small town. Once a programwas written, several people loaded it by settingswitches, dials, and cable connections. Despitea design innovation that boosted vacuum tubereliability by enabling them to be operated at25 percent normal power, 50 spent tubes hadto be found and replaced on an average day.

Early computer projects employed people inthree roles: operation, management, and pro-gramming. A small army of operators was need-ed. Managers oversaw design, development,and operation, including the specification ofprograms to be written and the distribution ofresults. Each role eventually became a focus ofHCI research, and despite the continual evolu-tion of computers and the activities aroundthem, we still find that these roles reflectaspects of this early division of labor.

1945-1958: Managing vacuum tubesReducing operator burden was a key focus of

early innovation: eliminating the need to resetvacuum tubes, facilitating replacement ofburned-out tubes, and developing stored-pro-gram computers that could be loaded by taperather than manually with cables and switch-es. These endeavors were consistent with theknobs and dials human factors tradition. Bythe late 1950s,3 one computer operator coulddo the work that previously required a team.

The first engineers to design and build com-puters chose their vocations. They delegatedroutine tasks to human operators. As comput-ers became more reliable and capable, pro-gramming became a central activity. Peopletook it up because they enjoyed it. To improveprogrammers interfaces to computers meant todevelop languages, compilers, and constructssuch as subroutines. Grace Hopper, a pioneer inthese areas in the 1950s, described her goal asfreeing mathematicians to do mathematics.4

This is echoed in todays usability goal of free-ing users to do their work.

19581965: Transistors open newvistas

Early forecasts that the world would needfew computers reflected the limitations of vac-uum tubes. The arrival of commercial solid-

state computers in 1958 led to dramatic change.As computers were deployed more widely,attention to the operators job increased. Evenmore significantly, people could envision pos-sibilities that were unimaginable for barn-sizedmachines of limited capability.

Helping operatorsIn the beginning, the computer was so

costly that it had to be kept gainfully occupiedfor every second; people were almost slaves tofeed it. 5

Brian Shackel

Low-paid computer operators set switches,pushed buttons, read lights, loaded and burstprinter paper; they loaded and unloaded cards,magnetic tapes, and paper tapes, and so on.Teletypes were the first versatile mode of directinteraction. Operators typed commands andread printed computer responses and statusmessages on paper that scrolled up one line at atime. The first displays (called VDUs or VDTsfor visual display units or terminals, or CRTs forcathode ray tubes) were nicknamed glass ttysglass teletypesbecause they too scrolled upoperator commands and computer-generatedmessages. Most displays were monochromeand restricted to alphanumeric characters. Earlyterminals cost around $50,000 in todays dol-lars: expensive, but a small fraction of the costof a computer. A large computer might haveone console, used only by the operator.

Improving the design of console buttons,switches, and displays was a natural extensionof human factors. Experts in this field authoredthe first human-computer interaction papers,capturing the attention of some who were devel-oping and acquiring systems in industry andgovernment. In 1959, Brian Shackel publishedthe article, Ergonomics for a Computer,6 fol-lowed by Ergonomics in the Design of a LargeDigital Computer Console.6 Sid Smith pub-lished ManComputer Information Transferin 1963.6

Early visions and demonstrationsIn his influential 1945 essay As We May

Think, Vannevar Bush, who helped shape sci-entific research funding in the US, described amechanical device that anticipated many capa-bilities of computers.7 After transistors replacedvacuum tubes, a wave of creative writing andprototype building by several computer pio-neers and experts led to expanded and morerealistic visions.

J.C.R. Licklider outlined requirements forinteractive systems and accurately predicted

OctoberDecember 2005 3

which would prove easier (for example, visualdisplays) and which more difficult (for exam-ple, natural-language understanding). JohnMcCarthy and Christopher Strachey proposedtime-sharing systems, crucial to the spread ofinteractive computing. In 1963, IvanSutherlands Sketchpad demonstrated con-straints, iconic representations, copying, mov-ing, and deleting of hierarchically organizedobjects, and object-oriented programming con-cepts. Douglas Engelbarts broad vision includ-ed the foundations of word processing,invention of the mouse and other input devices,and an astonishing public demonstration of dis-tributed computing that integrated text, graph-ics, and video. Ted Nelson anticipated a highlyinterconnected network of digital objects, fore-shadowing aspects of Web, blog, and wiki tech-nologies. Rounding out this period were AlanKays descriptions of personal computing basedon versatile digital notebooks.8

Progress in HCI is perhaps best understood interms of inspiring visions and prototypes, wide-spread practices, and the relentless hardwareadvances that enabled software developers totransform the former (visions and prototypes)into the latter. Some of the anticipated capabili-ties are now taken for granted, some are justbeing realizedothers remain elusive.

Titles such as ManComputer Symbiosis,Augmenting Human Intellect, and AConceptual Framework for ManMachineEverything described a world that did notexist, in which people who were not computerprofessionals were hands-on users of comput-ers out of choice. The reality was that for sometime to come, most hands-on use would beroutine, nondiscretionary operation.

Discretion in computer useOur lives are distributed along a continuum

between the assembly line nightmare of ModernTimes and utopian visions of completelyempowered individuals. To use a technology ornot to use it: Sometimes we have a choice, othertimes we dont. When I need an answer byphone, I may have to wrestle with speech recog-nition and routing systems. In contrast, myhome computer use is largely discretionary. Theworkplace often lies in-between: Technologiesare recommended or prescribed, but we ignoresome injunctions, obtain exceptions, use somefeatures but not others, and join with colleaguesto advocate changes in policy or availability.

For early computer builders, their work wasmore a calling than a job, but operationrequired a staff to carry out essential but lessinteresting repetitive tasks. For the first half of

the computing era, most hands-on use was bypeople hired with this mandate. Hardware inno-vation, more versatile software, and steadyprogress in understanding the psychology ofusers and tasksand transferring that under-standing to software developersled to hands-on users who exercised more choice in whatthey did with computers and how they did it.Rising expectations played a rolepeople havelearned that software is flexible and expect it tobe more congenial. Competition among ven-dors produces alternatives. Today, more use isdiscretionary, with more emphasis on market-ing to consumers and stressing user-friendliness.

Discretion is not all-or-none. No one mustuse a computer. But many jobs and pastimesrequire it. True, people can resist, sabotage, usesome features but not others, or quit the job. Buta clerk or systems administrator is in a differentsituation than someone using technology forleisure activity. For an airline reservation opera-tor, computer use is mandatory. For someonebooking a flight, use is discretionary. This articleexplores implications of these differences.

Several observers have remarked on the shifttoward greater discretion. A quarter centuryago, John Bennett predicted that discretionaryuse would lead to more concern for usability.9

A decade later, Liam Bannon noted broaderimplications of a shift from human factors tohuman actors.10 But the trajectory is notalways toward choice. Discretion can be cur-tailed even as more work is conducted digital-lyfor example, a word processor is virtuallyrequired, no longer an alternative to a type-writer. Even in an era of specialization, cus-tomization, and competition, the exercise ofchoice varies over time and across contexts.

Discretion is only one factor, but an analy-sis of its role casts light on diverse HCI efforts:the early and ongoing human factors work,visionary writers and prototype builders, sys-tems management, performance modeling, therelentless pursuit of some technologies despitelimited marketplace success, the focus of gov-ernment research funding, the growing empha-sis on design, and unsuccessful efforts to bridgeresearch fields.

19651980: HCI before personalcomputing

In 1964, Control Data Corp. launched thetransistor-based 6000 series. In 1965, integratedcircuits arrived with the IBM System/360. Thesepowerful computers, later christened mainframesto distinguish them from minicomputers,brought computing into the business realm. Atthat point, each of the three roles in comput-

4 IEEE Annals of the History of Computing

Three Faces of HumanComputer Interaction

ingoperation, management, programmingbecame a significant profession.

Operators interacted directly with comput-ers for routine maintenance, loading and run-ning programs, filing printouts, and so on. Thishands-on category can be expanded to includedata entry, retrieval, and other repetitive tasksnecessary to feed the computer.

Managers variously oversaw hardware acqui-sition, software development, operation, androuting and using output. They were usuallynot hands-on users.

Programmers were rarely direct users untillate in this period. Instead, they flowchartedprograms and wrote them on paper. Keypunchoperators then punched the program instruc-tions onto cards. These were sent to computercenters for computer operators to run.Printouts and other output were picked uplater. Many programmers would use computersdirectly when they could, but the cost of com-puter use generally dictated an efficient divi-sion of labor.

Human factors and ergonomicsIn 1970, Brian Shackel founded the Human

Sciences and Advanced Technology (HUSAT)center at Loughborough University in the UK,devoted to ergonomics research emphasizingHCI. Sid Smith and other human factors engi-neers published through this period.11 In 1972,the Computer Systems Technical Group (CSTG)of the Human Factors Society formed, and soonit was the largest technical group in the society.

Leading publications were the general jour-nal Human Factors and the computer-focusedInternational Journal of Man-Machine Studies(IJMMS), first published in 1969.

The first influential HCI book was JamesMartins 1973 Design of Man-ComputerDialogues.12 A comprehensive survey of inter-faces for operation and data entry, it beganwith an arresting opening chapter describing aworld in transition. Extrapolating from declin-ing hardware prices, Martin wrote:

the terminal or console operator, instead of beinga peripheral consideration, will become the tailthat wags the whole dog ... The computer indus-try will be forced to become increasingly con-cerned with the usage of people, rather than withthe computers intestines.12

In 1980, two major HCI books on VDTdesign and one on general ergonomic guide-lines were published.13 German work on VDTstandards, first published in 1981, provided aneconomic incentive to design for human capa-

bilities by threatening to prohibit noncompli-ant products.

Information systemsBeginning in 1967, the journal Management

Science published a column titled InformationSystems in Management Science. Early defini-tions of IS14 included an integratedman/machine system for providing informationto support the operation, management, anddecision-making functions in an organizationand the effective design, delivery and use ofinformation systems in organizations. A his-torical survey of IS research identifies HCI as oneof five major research streams, initiated byRussell Ackoffs 1967 paper on challenges indealing with computer-generated information.15

Companies acquired expensive businesscomputers to address major organizational con-cerns. Managers could be virtually chained tothem almost as tightly as Shackels operatorand data entry slaves. However, operator orend-user resistance to using a system could bea major management concern. For example,the sociotechnical approach involved educat-ing representative workers in technology pos-sibilities and involving them in design toincrease acceptance of the resulting system.16

Cognitive style, a major topic of early ISresearch, focused on difficulties that managershad communicating with people knowledge-able about computers. IS researchers publishedin management journals and in the human-factors-oriented IJMMS.17

Programming: Subject of study, source of changeIn the 1960s and 1970s, more than 1,000

research papers on variables affecting program-ming performance were published. Most viewedprogramming in isolation, independent of orga-nizational context. Gerald Weinbergs landmarkThe Psychology of Computer Programming appearedin 1971. In 1980, Ben Shneiderman publishedSoftware Psychology, and Beau Sheil reviewed stud-ies of programming notation (conditionals, con-trol flow, data types), practices (flowcharting,indenting, variable naming, commenting), andtasks (learning, coding, debugging).18

Programmers changed their own fieldthrough invention. In 1970, Xerox Palo AltoResearch Center (PARC) was founded to advancecomputer technology by developing new hard-ware, programming languages, and program-ming environments. It drew researchers andsystem builders from the labs of Engelbart andSutherland. In 1971, Allen Newell of CarnegieMellon University proposed a project to PARC,launched three years later: Central to the activ-


ities of computingprogramming, debugging,etc.are tasks that appear to be within thescope of this emerging theory (a psychology ofcognitive behavior).19

Like HUSAT, also launched in 1970, PARChad a broad research charter. HUSAT focusedon ergonomics, anchored in the tradition ofnondiscretionary use, one component ofwhich was the human factors of computing.PARC focused on computing, anchored invisions of discretionary use, one component ofwhich was also the human factors of comput-ing. Researchers at PARC and a few other placesextended human factors to higher-level cogni-tion. HUSAT, influenced by sociotechnicaldesign, extended human factors by consideringorganizational factors.

19801985: Discretionary use comesinto focus

In 1980, Human Factors and Ergonomics(HF&E) and IS were focused more on improv-ing efficiency than on augmenting humanintellect. In contrast, many programmers werecaptivated by this promise of computation.Growing numbers of students and hobbyistsused minicomputers and microprocessor-basedhome computers, creating a population ofhands-on discretionary users. Twenty yearslater, the visions early pioneers had of peoplechoosing to use computers that helped themwork better began to come true. And as a result,the cognition of discretionary users became atopic of interest.

Human Interaction with Computers, a 1980book by Harold Smith and Thomas Green,perched on the cusp. It briefly addressed thehuman as a systems component (the nondis-cretionary perspective). One-third coveredresearch on programming. The remainderaddressed non-specialist people, discre-tionary users who were not computer special-ists. Smith and Green wrote: Its not enoughjust to establish what people can and cannotdo; we need to spend just as much effort estab-lishing what people can and want to do. [Italicsin the original.]20

The formation of ACM SIGCHIIn 1980, as IBM prepared to launch the PC, a

groundswell of attention to computer userbehavior was building. IBM had recently addedsoftware to hardware as a product focus.21

Several cognitive psychologists joined an IBMresearch group that included John Gould, whohad engaged in human factors research sincethe late 1960s. They initiated empirical studiesof programming and software design and use.

Other psychologists who led recently formedHCI groups included Phil Barnard at theMedical Research Council Applied PsychologyUnit (APU); Tom Landauer at Bell Labs; DonaldNorman at the University of California, SanDiego; and John Whiteside at DigitalEquipment Corp.

PARC and CMU were particularly influen-tial. In 1980, Stuart Card, Thomas Moran, andAllen Newell published Keystroke-Level Modelfor User Performance Time with InteractiveSystems and introduced cognitive elements ascomponents of the goals, operators, methods,selection rules (GOMS) model that was thebasis for their landmark 1983 book, ThePsychology of Human-Computer Interaction.22

Communications of the ACM initiated theHuman Aspects of Computing department in1980. Computing Surveys published a specialissue on The Psychology of the ComputerUser the next year, edited by Tom Moran. TheAssociation for Computing Machinery (ACM)Special Interest Group on Social and BehavioralScience Computing (SIGSOC) extended its1981 workshop to cover interactive softwaredesign and use; the group shifted to the latterfocus and adopted the name Computer-Human Interaction (SIGCHI) in 1982.

In 1983, the first CHI conference23 drewmore than 1,000 people. Cognitive psycholo-gists in industry dominated the program. Halfof the 58 papers were from the seven organiza-tions mentioned earlier. The 1983 Computer-Human Interaction Conference (CHI 83) wascosponsored by the Human Factors Society.Human factors contributors included programchair Richard Pew, committee members SidSmith, H. Rudy Ramsay, and Paul Green, andseveral presenters. Brian Shackel and societypresident Robert Williges gave tutorials the firstday. Human Factors in Computing Systemswas and remains the conference subtitle.

CHI and human factors divergeDespite the initial interdisciplinary coopera-

tion with human factors specialists, most cogni-tive psychologists were familiar with interactivesoftware but not the human factors research lit-erature. Many had turned to HCI after earningtheir degrees, when academic psychology posi-tions became scarce. The Human Factors Societydid not again cosponsor CHI, and its researchersdisappeared from the CHI program committee.Soon, few CHI authors identified themselveswith human factors.

Reservations about human factors were evi-dent in The Psychology of Human-ComputerInteraction:



Human factors specialists, ergonomists, andhuman engineers will find that we have synthe-sized ideas from modern cognitive psychologyand artificial intelligence with the old methodsof task analysis The user is not an operator. Hedoes not operate the computer, he communi-cates with it 22

Two years later, Newell and Card noted thathuman factors had a role in design but

classical human factors has all the earmarks ofsecond-class status. (Our approach) avoids con-tinuation of the classical human-factors role (bytransforming) the psychology of the interfaceinto a hard science.24

In a June 2004 email communication, Cardsaid Human factors was the discipline we weretrying to improve, and

I personally changed the (CHI conference) call in1986 so as to emphasize computer science andreduce the emphasis on cognitive science,because I was afraid that it would just becomehuman factors again.

Hard science, in the form of engineering,drives out soft science, in the form of humanfactors, wrote Newell and Card.24 Cognitiveengineering and usability engineeringappeared; human factors disappeared. MostCHI researchers who had published in theannual human factors conference and HumanFactors shifted to CHI, Communications of theACM, and the journal Human-ComputerInteraction established in 1985 by Tom Moran.

In the first paper presented at CHI 83,Design Principles for HumanComputerInterfaces, Donald Norman experimentedwith applying engineering techniques to dis-cretionary use, creating user satisfaction func-tions based on technical parameters.25 Onlyslowly would CHI stop identifying so stronglywith engineering.

Although highly respected, human per-formance modeling did not draw a large CHIfollowing. Key goals of the modelers differedfrom those of practitioners and otherresearchers. The central idea behind the modelis that the time for an expert to do a task on aninteractive system is determined by the time ittakes to do the keystrokes.26 This helps designfor nondiscretionary users, such as telephoneoperators engaged in repetitive tasks.27 But CHIfocused instead more on the first experiencesof new discretionary users: The early visionwas, two decades later, a pressing concern for

software and telecommunications companies.The shift was reflected at IBM T.J. Watson

Research Center. John Gould and ClaytonLewis authored a CHI 83 paper that beautiful-ly defined the CHI focus on user-centered, iter-ative design based on prototyping,28 andWatson cognitive scientists helped shape CHI.But Goulds principal focus remained humanfactors; he served as Human Factors Societypresident in 19871988. Symbolically, in 1984Watsons Human Factors Group faded awayand a User Interface Institute emerged.

Ruven Brooks, Bill Curtis, Thomas Green,Ben Shneiderman, and other CHI founderscontinued the psychology-of-programmingresearch thread. Watson researchers also con-tributed, I learned from John Thomas in anOct. 2003 email:

One of the main themes of the early work wasbasically that we in IBM were afraid that the mar-ket for computing would be limited by the num-ber of people who could program complexsystems so we wanted to find ways for non-pro-grammers to be able, essentially, to program.

Line editors displaced coding sheets, andprogramming became the first profession pop-ulated by discretionary computer users. Manystudies of programmers as new hands-on userswere published in the early conferences. In1984 at an INTERACT session I attended, ThomasGreen remarked that text editors are the whiterats of HCI. As personal computing spread andthe same methods were applied to studyingother discretionary use, studies of program-ming gradually disappeared.

CHI focused on novice use for several rea-sons. Initial experience is particularly impor-tant for discretionary users, and thus for themany vendors who sprang up to develop soft-ware for PCs, workstations, and minicomput-ers. Novices are a natural focus when studyingnew technologies that have few experts. Andinitial use is critical when more people take upcomputing each year than did the year before.

Routine or experienced computer use waswidespread in this period. Computer databaseswere extensively used by airlines, banks, gov-ernment agencies, and other organizations. Buthands-on activity was rarely discretionary.Managers oversaw development and readreports, leaving data entry and informationretrieval to people hired for those jobs. CHIstudies of database use were fewI count threeover a decade, all focused on novice or casualuse. Improving skilled data entry was a humanfactors undertaking.




With fewer European companies producingmass-market software, research remained morefocused on less discretionary in-house develop-ment and use. At Loughborough University,HUSAT focused on job design (the division oflabor between people and systems) and collab-orated with the Institute for ConsumerErgonomics, particularly on product safety. In1984, Loughborough initiated an HCI graduateprogram drawing on human factors, industrialengineering, and computer science. TheInternational Conference on Human-ComputerInteraction (INTERACT) conference, first held inLondon in 1984 and chaired by Shackel, drewHF&E and CHI researchers.

In a perceptive essay written later from aEuropean perspective, Bannon urged that moreattention be paid to discretionary use, whilealso criticizing the exclusive focus on initialexperiences that marked CHI.10

Figure 1 positions some HCI events and top-ics on a timeline. The top row represents theHuman Factors and Ergonomics, predomi-nantly nondiscretionary, HCI focus. In the cen-ter is HCI in MIS (or Information Systems),initially focused on use that was relativelynondiscretionary and hands-off. At the bottomare CHI and its logical antecedents as discre-

tionary use shifted from engineers to program-mers to other individuals and groups.29

James Martins comprehensive 1973 guideto designing for data entry and retrievalbelongs in the top row, although his farsight-ed introduction, describing a future in whichusers are in control, places him among thosewho anticipated discretionary use in the bot-tom row. The Psychology of Human-ComputerInteraction, which introduced cognitive mod-eling, is placed at the top despite being thework of CHI researchers, because it focused onexpert performance and the reform of humanfactors. Discretion was not in its scope. A lead-ing modeler discouraged publication of a 1984study of a repetitive task that showed peoplepreferred a pleasant but slower interactiontechniquea result significant for discre-tionary use, but not for modeling aimed atmaximizing performance.30

The visionaries were not well-known with-in CHI in 1983. The 633 references in the 58papers presented at CHI 83 included manyauthored by well-known cognitive scientists,but Bush, Engelbart, and Sutherland were notcited at all. Shared concern for discretionaryuse, conceptual continuity, and the legitimacybestowed by a luminous past led CHI to graft

Figure 1. Timeline for some events, topics, and publications discussed in the text.

these pioneers onto CHI history somewhatafter the fact.

19852005: New interfaces, Internet,and the Web

Humancomputer interaction in the per-sonal computing era has been marked by thespread of Internet and intranet use, graphicaluser interfaces, and the World Wide Web.Although Internet users doubled annually withremarkable regularity, it required decades tobecome a significant fraction of the population.

Graphics made hard-earned progressthrough the 1960s and 1970s. In 1981, theXerox Star was the first product with a full GUI.The Star, the Apple Lisa, and other early GUIsdid not do well. When the 1984 Macintoshfailed with corporate buyers, Apples survivalwas uncertain. Late in 1985, positive consumerresponse and niche use for graphics and desk-top publishing validated the Mac and theGUI.31 When the Web linked the nodes of asteadily expanding Internet, graphics werethere to provide compelling content.

These breakthroughs played out differentlyin the three HCI research domains. TheMacintosh appeal to discretionary users had animmediate, sweeping impact on CHI research.GUIs did not attract significant corporate atten-tion until Windows 3.0 succeeded in 1990,delaying the impact on HF&E and IS until thetechnology was better understood. CHI tookthe discretionary early Web activity in stride,although it raised new issues. Initially a returnto a form-based interaction style, the Web inter-face had less impact on HF&E. For IS, the Websdiscretionary appeal and economic significancebrought opportunities and challenges.

HF&E and the role of governmentUnderstanding the field of human factors

and ergonomics requires a look at the role ofgovernment as user and supporter of researchand development. HF&E research has respond-ed to military, aviation, and telecommunica-tions interests, with government often leadingthe way. Bureaucratic needscensus, tax, socialsecurity, health and welfare, power plant oper-ation, air traffic control, ground control forspace missions, military logistics, processingtext and voice data for intelligencecontributeto governments being the largest consumer ofcomputing.

With primarily nondiscretionary bureau-cratic use, small efficiency gains in individualtransactions yield large benefits over time. Forroutine data entry and information retrieval orcomplex speech recognition and natural-lan-

guage understanding, incremental improve-ments that may not register with discretionaryusers make a difference.

Government drove the development ofergonomic standards. Acquiring a novel inter-active system through a competitive biddingprocess is tricky. As customers formulaterequirements, they must remain at armslength from potential developers who knowmore about technical possibilities. Compliancewith standards can be specified in a contract.

In 1986, Sid Smith and Jane Mosier pub-lished the last in a series of government-spon-sored interface guidelines. They mentioned butdid not address GUIs in 944 guidelines organ-ized into sections titled Data Entry, DataDisplay, Data Transmission, Data Protection,Sequence Control, and User Guidance. GUIswould expand the design space tremendously.Interfaces came to be based on predefinedstyles rather than built from scratch; contractscame to specify design processes rather thanadherence to specific feature guidelines.32

Worldwide, research funding is directed bygovernmental initiatives and shaped by gov-ernment concerns. The result is a focus onmandatory use. The US National ScienceFoundations interactive systems programsubsequently renamed HumanComputerInteractionwas described in this way:

The Interactive Systems Program considers sci-entific and engineering research oriented towardthe enhancement of humancomputer commu-nications and interactions in all modalities.These modalities include speech/language,sound, images and, in general, any single or mul-tiple, sequential or concurrent, humancomput-er input, output, or action.33

Speech recognition and natural-languageunderstanding, strongly emphasized by theNSF, are useful when a phone system providesno alternative, when a disability limits key-board use, when hands are otherwise occupied,or for professional translators and intelligenceanalysts. But they have rarely been used bypeople who have much choice.

The Human Factors Society undertook a sur-vey that indicated little overlap with CHI,where high-tech commercial vendor companiesdrove research into discretionary use. NSF andDARPA HCI program directors rarely attendedCHI. Little on speech recognition or natural lan-guage appeared at CHI conferences. Anothersignificant NSF focus, the use of brainwaves todrive computer displays, may also have uses butperhaps not in many homes or offices.


A review panel that included CHI membersnoted that NSF-funded researchers (PIs, or prin-cipal investigators) did not come from theirmidst:

In reviewing HCI Program coverage we consultedthe on-line HCI Bibliography (www.hcibib.org).This heavily-used (over one million searches)public index of over 24,000 records covers thefull contents of 14 journals, 20 conferences,books and other materials. It lists 506 authorswith ten or more publications. No PI for the 10randomly selected FY1999-FY2002 HCI Programawards is on this list HCI program grants arenot fully reflective of the HCI literature 34

An official said: NSFs logic is that it shouldprimarily support research on difficult topics,often NOT those industry is heavily workingon (William Bainbridge, email to author, Nov.2003). But it may be differences in priority andperceived significance, not difficulty, that dis-tinguishes these efforts.

In the late 1990s, cognitive psychologistsbecame more influential within the HumanFactors and Ergonomics Society (Ergonomicswas added in 1992; [HFES]). The largest techni-cal group is now Cognitive Engineering andDecision Making (CEDM in Figure 1), whichformed in 1996. The Human PerformanceModeling technical group (HPM) was estab-lished in 2004 by Wayne Gray and CHI 83 pro-gram chair Richard Pew. The effort to reformhuman factors from the outside that accompa-nied the birth of CHI has moved within, led bysome of the same people.

Starting in 1987, a biennial HumanComputer Interaction International conferenceseries has drawn from industrial engineering,human factors, and government-contractedresearch and development. Despite its sizemore than 1,000 papers were presented in2003HCII has modest visibility in the CHIcommunity.

IS and the formation of AIS SIGHCIGUIs did not have a major impact on IS in

the 1980s, but business graphics did. Visual dis-play of information affects everyone. HF&Ehad long addressed manuals and displays, soft-ware psychologists considered flowcharts andcode organization, and IS focused on the pres-entation of quantitative data. Izak Benbasatand Albert Dexter wrote an influential paperthat contrasted tables and charts and consid-ered effects of color.35

IS research included the management of pro-gramming in organizations.36 Also, sociotechni-

cal and Scandinavian participatory approaches,initiated earlier to bring nondiscretionary usersinto design, gained recognition.37

Research into computer-supported meetingfacilities flourished in the mid-1980s, assistedby declining costs of interactive computing.38

Unlike most group support technologies com-mercialized in this period, they originated in IS,not software or computer companies. Theirexpense and managerial focus limited theirmass-market appeal.

Within enterprises, discretionary useincreased: Fewer employees were almost slavesfeeding the machine. Embrace of the Internetcreated more porous organizational bound-aries. Even when productivity benefits areuncertain, employees bring consumer softwaresuch as free instant messaging (IM) clients andmusic players inside the firewall. Free Web-based software that enables one to create aweblog in a few minutes is a different animalthan high-overhead applications of the past. Inaddition, home use of software reducesemployee patience with poor interactive soft-ware at work. Managers who were hands-offusers in the 1980s became late adopters in the1990s, and are now hands-on early adopters oftechnologies that benefit them.39

In 1989, Fred Davis introduced the influen-tial Technology Acceptance Model (TAM).Influenced by early CHI research, TAM identi-fies perceived usefulness and perceived utilityas key factors in improving white collar per-formance often obstructed by users unwill-ingness to accept and use available systems.40

This managerial view is reflected in the termacceptance, reflecting a lack of choice. In con-trast, CHI authors speak only of adoption.

The Web had a seismic effect in IS when e-commerce took off in the late 1990s. When theInternet bubble popped, organizations contin-ued building portals: The Web had become anessential business tool.

IS was where CHI had been 20 years earlier:IT professionals who had previously focused oninternal operations were now tasked with pro-viding interfaces to highly discretionary exter-nal customers.

In 2001, the Association for InformationSystems (AIS) established the Special InterestGroup in HumanComputer Interaction(SIGHCI). The founders defined HCI by citing12 works by CHI researchers and made it a pri-ority to bridge to CHI. In contrast, HF&E is notamong five key disciplines that are considered;it is the last of seven related fields.

SIGHCIs broad charter includes a range oforganizational issues, but published work



focuses on interface design for e-commerce,online shopping, online behavior especiallyin the Internet era, and effects of Web-basedinterfaces on attitudes and perceptions. Eightof 10 papers in special journal issues coveredInternet and Web behavior.41

CHI and the shifting focus of discretionary useCHI immediately took up issues raised by

GUIs, such as mouse manipulation, visual dis-play of information, and user interface man-agement systems (UIMSs). An influential 1986analysis by Edwin Hutchins, James Hollan, andDonald Norman concluded that it is too earlyto tell how GUIs would fare.42 Concluding thatGUIs could well prove useful for novices, theauthors said we would not be surprised ifexperts are slower with Direct Manipulation sys-tems than with command language systems.

Experts may well be faster using commandsand function keys, but in a rapidly expandingcommercial marketplace, novices outnumberedexperts. Once they are familiar with an inter-face, people often do not switch for a promiseof better performanceif they did, the Dvorakkeyboard would be more popular. Experiencedusers are continually adopting new featuresand applications. All in all, it was rational tofocus on initial experience.

More powerful networking and processingled to collaboration support, hypertext andhypermedia, and then mobile and ubiquitouscomputing. As each moved from research todiscretionary use, CHI increased coverage andsponsored relevant conference series: ComputerSupported Cooperative Work (CSCW, first in1986), Hypertext (1987), and Ubicomp (1999).CSCW represented a particularly significantshift, adding social theory and methods, includ-ing ethnography, to the previously cognitiveorientation.

Conversely, technologies that became rou-tine or confined to a niche faded from view atCHI. Papers on command languages, editors,UIMSs, and programmer support disappeared.

Color, animation, and sound added engage-ment and seduction to interface design in thecompetitive software industry. Interface designas a wholly scientific endeavor became unten-able. CHI has sponsored the DesigningInteractive Systems (DIS) conference seriessince 1995 and cosponsored Designing UserExperience (DUX) since 2003. DIS attracts bothsystems and visual designers; DUX focuses onthe latter.

Web site design introduced a new challenge.A site owner wishes to keep users at a given site;users may prefer to escape quickly. For a CHI

professional whose self-perception is useradvocate, designing for Web site owners intro-duces a stakeholder conflict. This dilemma didnot arise with individual productivity tools.Marketing, not an aspect of cognitive or com-puter science and often at loggerheads withR&D in organizations, has a foot in the door.43

Evolution of methods and theoryPsychologists who shaped CHI, like those

who formed HF&E 30 years earlier, were trainedto test hypotheses about behavior in laboratoryexperiments. Experimental subjects agree tofollow instructions for an extrinsic reward. Thisis a good model for nondiscretionary use, butnot for discretionary use. CHI researchers rela-beled them participants, which sounds voli-tional, but lab findings require confirmation inreal-world settings more often than is true forergonomics studies.

Traditional ergonomic goals applyfewererrors, faster performance, quicker learning,greater memorability, and being enjoyablebut the emphasis differs. For power plant oper-ation, error reduction is key, performanceenhancement is good. Other goals are less crit-ical. In contrast, consumers often respond tovisceral appeal at the expense of usability andutility. CHI slowly abandoned its roots in sci-ence and engineering, although the adoptionof the term funology suggests a wistful reluc-tance to do so. Will funology be researchedonly in the lab?44

Unlike HF&E, CHI embraced quick-and-dirtylab studies and time-consuming qualitativeapproaches. The former can guide real-worldstudies or help select among alternatives whenthe optimal solution is not needed. The lattercan provide deeper understanding of userbehaviors; challenges communicating suchunderstandings led to methodological innova-tions such as contextual design and personas.45

Some early CHI researchers worked on the-oretical foundations for design based on com-mand naming and line editing as referencetasks.46 GUIs curtailed interest in these topics.As the design space expanded, hope of estab-lishing an overarching theory contracted.Application of modern cognitive theory istoday more often found in cognitive science,HF&E, and IS. A recent compilation of HCI the-ory and modeling approaches includes severalwith a cognitive orientation and a few socialscience or cognitive-social hybrids.47 That onlyone chapter focuses on computer sciencereveals the atheoretical nature of CHIs shifttoward computer science over the past twodecades.


Moores law exempts software inventionfrom the usual tangled dance of engineeringand science. Faster, smaller, and cheaper hard-ware ensures a steady flow of new devices andfeatures, and more complex and layered archi-tectures. Mobile devices, remote sensors andactuators, higher resolution, color, animation,voice over IP, application program interfaces(APIs), user interface libraries, and communi-cation protocols spawn new choices. Researchopportunities arise from indirect effects ofspreading use on privacy, security, work-life bal-ance, and so on.

The evolution of CHI is reflected in theinfluential contributions of Donald Norman. Acognitive scientist who coined the term cogni-tive engineering, he presented the first CHI 83paper. It defined User Satisfaction Functionsbased on speed of use, ease of learning,required knowledge, and errors. His influential1988 Psychology of Everyday Things focused onpragmatic usability. Its 1990 reissue as Design ofEveryday Things reflected the broad refocusingon invention. Fourteen years later he publishedEmotional Design: Why We Love (or Hate)Everyday Things, stressing the role of aestheticsin our response to objects.48

Discussion: Cultures and bridgesDespite a significant common focus, there

has been limited interaction among the threethreads of human-computer interactionresearch. This has not been for lack of trying.This section outlines some obstacles to interac-tion and efforts to overcome them.

Three communities, two academic culturesThe first two HCI disciplines to emerge,

HF&E and IS, arose before discretionary hands-on use was widespread. Researchers in eachconsidered both organizational and technicalissues. They shared journals; the Benbasat andDexter paper published in Management Sciencecited five Human Factors articles.

HF&E and IS also share the traditional aca-

demic culture of the sciences: Conferences arevenues for work in progress, journals are repos-itories for polished work. In contrast, for CHIand other US computer science disciplines, con-ference proceedings are the final destination formost work. Journals are secondary. Outside theUS, computer science retains more of a journalfocus, perhaps due to the absence of profes-sional societies that archive proceedings.49 Thiscircumstance impedes communication acrossdisciplines and continents. Researchers in jour-nal cultures chafe at CHIs rejection rates; CHIresearchers are dismayed by the relativelyunpolished work at other conferences.

Table 1 presents figures obtained from edi-tors of leading conferences and journals.50 CHIconferences are selective. CHI journals receivefewer submissions despite higher acceptancerates. These patterns were confirmed in inter-views. Many CHI researchers state that journalsare not relevant. Only about 10 percent of workin CHI-sponsored conferences reaches journalpublication. In contrast, a Hawaii InternationalConference on System Sciences 2004 trackorganizer estimated that 80 percent of researchthere progressed to a journal.51

A linguistic divide also set CHI apart. HF&Eand IS used the term operator; in IS, user couldbe a manager who used printed computer out-put, not a hands-on end user. Within CHI, oper-ator was demeaning, user was always hands-on,and end user seemed a superfluous affectation.

In HF&E and IS, task analysis referred to anorganizational decomposition of work; in CHIit was a cognitive decomposition, such asbreaking a text editing move operation intoselect, cut, select, paste. In IS, implementationmeant deployment of a system in an organiza-tion; in CHI it was a synonym for develop-ment. System, application, and evaluation alsohad markedly different connotations or deno-tations. Significant misunderstandings andrejections resulted from failure to recognizethese distinctions.52

Different perspectives and priorities werereflected in attitudes toward standards. ManyHF&E researchers contributed to standards devel-opment and argued that standards contribute toefficiency and innovation. Widespread in CHIwas the view that standards inhibit innovation.

A generational divide also existed. ManyCHI researchers grew up in the 1960s and1970s, and did not appreciate the HF&E orien-tation toward military and government sys-tems, or the fondness of HF&E and IS for malegenerics (for example, man-machine inter-action). This reduced enthusiasm for buildingbridges and exploring literatures.



Table 1. Submission and acceptances rates (medians rounded to 10%).

Field Journals: Journals: Conferences:Annual % %

Submissions Accepted Accepted Human Factors 150 30 80 and Ergonomics Information 200 10 60 Systems ComputerHuman 50 30 20 Interaction

Efforts to find common groundThe Human Factors Society was deeply

involved with the first CHI conference, but asCHI leaders wrote of human factors secondclass status and embraced computer science,human factors professionals abandoned CHI.In recent interviews, some recalled feeling thatCHI researchers believed incorrectly that theyhad discovered the topic, ignored human fac-tors contributions, employed usability studymethods that were insufficiently rigorous, andseemed more interested in describing theirexperiences. Some CHI papers were indeeddescriptive, and the widely used thinking-aloud verbal protocols, introduced to interfacedesign by Clayton Lewis based on the theoriesof Allen Newell and Herb Simon, were notwidely accepted in experimental psychology.53

The Computer Supported Cooperative Workconference series tried to bridge IS and CHI andmet a similar fate. IS participation on the pro-gram committee and program, initially one-third,steadily declined. By 2002 no one on the pro-gram committee had a primary IS affiliation. Inthe early 1990s, IS papers were routinely reject-ed. In interviews, IS researchers said that CSCWreviewers were not interested in IS contribu-tions or expected unrealistic effort for confer-ence publications that count little in a field thatregards conference papers as work in progress.

IS participation in CSCW was dispropor-tionately represented by Scandinavian cooper-ative or participatory design, which appealedto many in CHI. This situation might seem oddon the surface. Participatory design shared thetraditional IS focus in internal developmentand operation, not product design. It was overt-ly political, whereas CHI was scrupulously apo-litical. However, both focused on discretion:The key Scandinavian goal was to let workerscontrol technology choices. In addition, mostScandinavian and CHI researchers were of thesame generation, influenced by the culture ofthe 1960s and 1970s. This alliance faded as dif-ferences became more apparent, albeit moreslowly than the management IS tie.

Today, AIS SIGHCI seeks a cognitive bridgebetween IS and CHI, but the cultural forcesmust be reckoned with. Although SIGHCI doesnot mention HFES, it has organized sessionsand special issues for the human factors-ori-ented HCII conference and journals Behaviourand Information Technology, International Journalof Human-Computer Studies, and InternationalJournal of Human-Computer Interaction. High-acceptance, work-in-progress conference ses-sions that yield human factors and IS journalspecial issues will draw few CHI researchers.

Cultural constraints can overpower apparentshared interests.

Other activities and perspectivesAnother thread of humancomputer inter-

action research is coalescing as information sci-ence, with conferences, journals, and societiesthat address database use, informationretrieval, and the digital evolution of library sci-ence. One component of information scienceresearch can be traced to office automationefforts that sprang up around minicomputersin the 1970s, between the mainframes thatspawned information systems and the PCs ofCHI. The Web-based shift to information repos-itories returned this thread to prominence.

More could be said about the telecommuni-cations industry. It had the most external cus-tomers and internal employees, and influencedevery facet of HCI research.54 Software engi-neering and artificial intelligence are relevantdisciplines passed over here. Finally, by empha-sizing tendencies around choice and mandate,this account bypasses research on interactivegraphics and other technical contributions.

TrajectoriesHuman-computer interaction has been a

particularly dynamic field, in large part due tothe steady increase in hardware capability.Understanding past and present trends mayprovide some help in anticipating directionsthe field could take.

DiscretionNow you see it, now you dontWe exercise choice more at home than at

work; a lot when buying online, none when con-fronted by a telephone answering system; con-siderable when young and healthy, less whenconstrained by injury or aging. Alternatives dis-appear: Software that was discretionary yesterdayis indispensable today, and the need to collabo-rate forces us to adopt common systems andconventions.

Consider a hypothetical team. In 1985, onemember still used a typewriter, others chose dif-ferent word processors. They exchanged print-ed documents. One emphasized phrases byunderlining, another by italicizing, a third bybolding. In 1995, in order to share documentsdigitally, group members had to adopt the sameword processor and conventions. Choice wascurtailed; it had to be exercised collectively.Technology can restore discretion: If it sufficesto share documents in PDF format, in 2005 theteam can use different word processors again,and one can envision a capability that allowsme to see in italics what you see as bold.


Shackel noted the progression under theheading From Systems Design to InterfaceUsability and Back Again.5 Early designersfocused at the system level and operators had tocope. When the PC merged the roles of opera-tor, output user, and program provider, the focusshifted to the human interface and choice. Thenindividual users again became components infully networked organizational systems. Whena technology becomes mission-critical, as emaildid for many in the 1990s, discretion is gone.

The converse also occurs. Discretion increas-es when employees download free software anddemand capabilities they have at home.Managers are less likely to mandate the use of atechnology that they use and find burdensome.For example, speech recognition systemsappealed to military officers who anticipatedthat subordinates would use them. When sen-ior officers become users, the situation changed:

Our military users generally flatly refuse to useany system that requires speech recognition. Over and over and over again, we were told If wehave to use speech, we will not take it. I dont evenwant to waste my time talking to you if it requiresspeech ... I have seen generals come out of using,trying to use one of the speech-enabled systemslooking really whipped. One really sad puppy, hesaid OK, whats your system like, do I have to usespeech? He looked at me plaintively. And when Isaid No, his face lit up, and he got so happy.55

As familiar applications become essential,and as security concerns curtail openness, onemight expect discretion to recede, but Mooreslaw, greater competition, and more efficient dis-tribution guarantee that a steady flow ofunproven technologies will find their way to us.

Looking aheadWill three HCI fields endure? Perhaps not,

perhaps HCI goals will be realized only when itceases to be a field of research altogether. In1988, Norman wrote of the invisible comput-er of the future.56 Like motors, he speculated,computers would be present everywhere andvisible nowhere. We interact with clocks, refrig-erators, and cars. Each has a motor, but there isno humanmotor interaction specialization. Adecade later, at the height of the Y2K crisis andthe Internet bubble, computers were more vis-ible than ever. We may always want a multi-purpose display or two, but part of Normansvision is materializing. With computers embed-ded everywhere, concern with our interactionwith them is everywhere. Today, as interactionwith digital technology is becoming part of

everyones research, the three HCI fields are los-ing participation.

Human factors and ergonomics. Imperatives toimprove training, expert performance, anderror handling have strong continued supportfrom government and the private sector. DavidMeister, author of The History of Human Factorsand Ergonomics, stresses the continuity of HF&Ein the face of technology change:

Outside of a few significant events, like theorganization of HFS in 1957 or the publicationof Proceedings of the annual meetings in 1972,there are no seminal occurrences no sharp dis-continuities that are memorable. A scientific dis-cipline like HF has only an intellectual history;one would hope to find major paradigm changesin orientation toward our human performancephenomena, but there are none, largely becausethe emergence of HF did not involve majorchanges from pre-World War II applied psychol-ogy. In an intellectual history, one has to look formajor changes in thinking, and I have not beenable to discover any in HF 57

Where among the 22 HFES technical groupsis HCI represented? Membership in theComputer Systems Technical Group hasdeclined sharply, but technology use is stressedin Cognitive Engineering and Decision Making,Communication, Human PerformanceModeling, Internet, System Development, andVirtual Environment technical groups. Nor canAging, Medical Systems, or others avoid invis-ible computers. HCI papers appear without anHCI label.58

Information systems. As IS thrived in the 1990s,other management school disciplinesfinance,marketing, operations research, organizationalbehaviorbecome more technically savvy. Whenthe bubble burst and enrollments declined, ISwas left with a less well-defined niche.

IS research issues, including HCI, remain sig-nificant, but this cuts two ways. With IT opera-tion standardization and outsourcing, Webportals and business-to-business ties get moreattention. Along with novel HCI issues, theybring in economic and marketing considera-tions, making it easier for HCI functions to beabsorbed by traditional management disci-plines. Some IS departments and individualsare aligning with computer science or infor-mation science, rather than management.

Computer-human interaction. This nomadicgroup started in psychology, obtained a place



at the edge of the table in computer science,and is increasingly drawn to information sci-ence. Lacking a well-defined academic niche,CHIs identity is tied to its conference, and CHIconference participation has dropped as spe-cialized conferences thrive.

The focus on discretionary use is under pres-sure as technologies appear and spread at anever-increasing pace. When an emerging tech-nology was slower to attract a critical mass ofusers, researchers on the topic first contributedto existing conferences and journals. Today,groups focused on new technologies can splitoff quickly. For example, soon after the Webemerged, annual WWW conferences drewpapers on HCI issues. High conference rejec-tion rates and a new generational divide couldaccelerate this dispersion of effort as successfulconferences are established for ubiquitous andpervasive computing, agents, design, and soon. HCI is invisibly present in each.

Conclusion

This review of humancomputer interactionexplores efforts that might benefit from closercoordination but have remained distinctifanything, moving apart over time. CHI focus-es on invention and design; some of its engi-neering and modeling components havemigrated to HF&E. A commercial software com-pany employee familiar with the human factorscommunity said (in a Sept. 2004 email com-munication to me from Edie Adams), After allthese years Ive concluded that we use the samemethods, we study the same things, but we doit to get new ideas, and they do it to improvewhat already exists.

CHI and IS could learn from one another.CHI discovered the limitations of laboratorystudies and surveys for understanding discre-tionary use. Many IS researchers now focusedon discretionary use still favor these tech-niques. IS has a more developed awareness ofeconomic, organizational, and marketing the-ory and practice than CHI. But strong culturalbarriers separate the two.

Moores law ensures that landscapes will shiftfor some time, providing new forms of interac-tion to explore, new practices to improve. Ayounger generation has grown up with gameconsoles and cell phones; communicating withIM and text messaging; developing skills atsearching, browsing, tagging, and synthesizing;and acquiring multimedia authoring talent viadigital cameras and blogs. However it comes tobe defined and wherever it is studied, humancomputer interaction is still in its early days.

AcknowledgmentsMore people than I can name assisted enthusi-astically in this research. I relied heavily on RonBaecker, Phil Barnard, Clayton Lewis, DonNorman, Gary Perlman, Brian Shackel, andGayna Williams. Edie Adams, Izak Benbasat,Ahmet Cakir, Jack Carroll, Bill Curtis, JayElkerton, Dennis Galletta, Wayne Gray, ThomasGreen, Pam Hinds, Jeff Johnson, Gary Klein,David Lenorovitz, Arnie Lund, David Meister,Tom Moran, Jane Mosier, William Newman,Judy Olson, Richard Pew, Jay Pittman, StevePoltrock, Lois Smith, John Thomas, JohnWhiteside, Patricia Wright, and Ping Zhangcontributed. Brian Gaines, Pam Hinds, JohnKarat, Chris Kemerer, Tom Stewart, and RonWeber provided publications data. My apologiesfor contributors not listed, and for errors ofomission, commission, and interpretation.

References and notes1. F.W. Taylor, The Principles of Scientific Management,

Harper, 1911.2. D. Meister, The History of Human Factors and

Ergonomics, Lawrence Erlbaum Assoc., 1999; S.N.Roscoe, The Adolescence of Engineering Psychology,Human Factors and Ergonomics Soc., 1997; F.Dyson, Disturbing the Universe, Harper and Row,1979; D. Meister, HFES History, HFES 2002-2003 Directory and Yearbook, Human Factors andErgonomic Soc., 2002, pp. 2-3.

3. This article focuses primarily on when changesbecame widespread. Thus, dates are approximate.For example, although prototype transistor-basedcomputers were available earlier, 1958 is a reason-able demarcation for eras of computer use.

4. G. Hopper, The Education of a Computer, Proc.ACM Conf., 1952, ACM, reprinted in Annals of theHistory of Computing, vol. 9, no. 3-4, 1987, pp.271-281; J. Sammet, Farewell to Grace HopperEnd of an Era! Comm. ACM, vol. 35, no. 4, 1992,pp. 128-131.

5. B. Shackel, HumanComputer Interaction:Whence and Whither? J. Am. Soc. for InformationScience, vol. 48, no. 11, 1997, pp. 970-986.

6. B. Shackel, Ergonomics for a Computer,Design, vol. 120, 1959, pp. 36-39; B. Shackel,Ergonomics in the Design of a Large DigitalComputer Console, Ergonomics, vol. 5, 1962,pp. 229-241; S.L. Smith, ManComputer Infor-mation Transfer, Electronic Information DisplaySystems, J.H. Howard, ed., Spartan Books, 1963,pp. 284-299.

7. V. Bush, As We May Think, The AtlanticMonthly, vol. 176, July 1945, pp. 101-108.

8. J. Licklider, ManComputer Symbiosis, IRE Trans.Human Factors in Electronics, vol. HFE-1, 1960, pp.4-11; J. Licklider and W. Clark, On-Line ManCom-


puter Communication, Proc. AFIPS Conf. Proc.,vol. 21, American Federation of Information Pro-cessing Societies (AFIPS), 1962, pp. 113-128; R.Fano and F. Corbato, Time-Sharing on Comput-ers, Scientific Am., vol. 214, no. 9, 1966, pp. 129-140; I. Sutherland, Sketchpad: A ManMachineGraphical Communication System, Proc. AFIPSConf. Proc., vol. 23, 1963, pp. 329-346; D. Engel-bart, A Conceptual Framework for the Augmen-tation of Mans Intellect, Vistas in InformationHandling, P. Howerton and D. Weeks, eds., vol. 1,Spartan Books, 1963, pp. 1-29; D. Engelbart andW. English, A Research Center for AugmentingHuman Intellect, Proc. AFIPS Conf., vol. 33, 1968,pp. 395-410; T. Nelson, A File Structure for theComplex, the Changing, and the Indeterminate,Proc. ACM Natl Conf., ACM, 1965, pp. 84-100; T.Nelson, A Conceptual Framework for ManMachine Everything, Proc. AFIPS Natl Joint Com-puter Conf., vol. 42, AFIPS, 1973, pp. M21-M26; A.Kay, The Reactive Engine, doctoral dissertation,Univ. of Utah, Dept. of Computer Science, 1969;A. Kay and A. Goldberg, Personal DynamicMedia, Computer, vol. 10, no. 3, Mar. 1977, pp.31-42. For more detail on the early history, see R.Baecker et al., Readings in HumanComputer Inter-action: Toward the Year 2000, Morgan Kaufmann,1995; B.A. Myers, A Brief History of Human Com-puter Interaction Technology, ACM Interactions,vol. 5, no. 2, 1988, pp. 44-54; R. Pew, Evolutionof HumanComputer Interaction: From MEMEX toBluetooth and Beyond, The HumanComputerInteraction Handbook, J.A. Jacko and A. Sears, eds.,Lawrence Erlbaum Assoc., 2003, pp. 1-17; B.Shackel, HumanComputer Interaction: Whenceand Whither? JASIS, vol. 48, no. 11, 1997, pp.970-986.

9. J.L. Bennett, The Commercial Impact of Usabilityin Interactive Systems, ManComputer Commu-nication, Infotech State-of-the-Art, vol. 2, B. Shack-el, ed., Pergamon-Infotech, 1979, pp. 1-17.

10. L. Bannon, From Human Factors to Human Actors:The Role of Psychology and HumanComputerInteraction Studies in System Design, Design atWork, J. Greenbaum and M. Kyng, eds., LawrenceErlbaum Assoc., 1991, pp. 25-44.

11. For example, S.L. Smith and N.C. Goodwin,Computer-Generated Speech and ManCom-puter Interaction, Human Factors, vol. 12, no. 2,1970, pp. 215-223.

12. J. Martin, Design of ManComputer Dialogues,Prentice-Hall, 1973; quotation from pp. 3-4.

13. A. Cakir, D.J. Hart, and T.F.M. Stewart, Visual Dis-play Terminals, Wiley, 1980; E. Grandjean and A.Vigliani, Ergonomics Aspects of Visual Display Ter-minals, Taylor and Francis, 1980; L. Damodaran,A. Simpson, and P. Wilson, Designing Systems forPeople, NCC Publications, 1980.

14. Data Processing (DP), Management InformationSystems (MIS), Information Systems (IS), and Infor-mation Technology (IT) have been used relativelyinterchangeably. MIS has dominated, but thefields focus has become less strictly management,so this article will use IS, and IT professionals todescribe corporate system support personnel.

15. P. Zhang, F.F.-H. Nah, and J. Preece, HCI Studiesin Management Information Systems, Behaviourand Information Technology, vol. 23, no. 3, 2004,pp. 147-151 (quotation, p. 147, from G.B. Davis,Management Information Systems: Conceptual Foun-dations, Structures, and Development, McGraw-Hill,1974); R.D. Banker and R.J. Kaufmann, The Evolu-tion of Research on Information Systems: A Fiftieth-Year Survey of the Literature in ManagementScience, Management Science, vol. 50, no. 3,2004, pp. 281-298; R.L. Ackoff, ManagementMisinformation Systems, Management Science,vol. 14, no. 4, 1967, pp. B147-B156.

16. J. Greenbaum, In the Name of Efficiency, TempleUniv., 1979; E. Mumford, A ComprehensiveMethod for Handling the Human Problems ofComputer Introduction, Proc. IFIP Congress 71,vol. 2, North-Holland, 1971, pp. 918-923.

17. N.A. Mylonopoulos, and V. Theoharakis, GlobalPerceptions of IS Journals, Comm. ACM, vol. 44,no. 9, 2001, pp. 29-32. The authors ranking of50 journals by IS researchers placed the Intl J.ManMachine StudiesHumanComputer Studies(IJMMS) 23rd when reanalyzed to merge figuresfor IJMMS and, reflecting a name change in1994, Intl J. HumanComputer Studies (IJHCS)(Mylonopoulos email to author, Aug. 2004).

18. R. Baecker et al., Readings in HumanComputerInteraction: Toward the Year 2000; G. Weinberg,The Psychology of Computer Programming, Van Nos-trand Reinhold, 1971; B. Shneiderman, SoftwarePsychology: Human Factors in Computer and Infor-mation Systems, Winthrop, 1980; B.A. Sheil, ThePsychological Study of Programming, ACM Com-puting Surveys, vol. 13, no. 1, 1981, pp. 101-120.

19. S.K. Card and T.P. Moran, User Technology:From Pointing to Pondering, Proc. Conf. on His-tory of Personal Workstations, ACM Press, 1986,pp. 183-198.

20. H.T. Smith, and T.R.G. Green, Human Interactionwith Computers, Academic Press, 1980.

21. R. Pew, Evolution of HumanComputer Interac-tion: From MEMEX to Bluetooth and Beyond,The HumanComputer Interaction Handbook, J.A.Jacko and A. Sears, eds., Lawrence ErlbaumAssoc., 2003.

22. S.K. Card, T.P. Moran, and A. Newell, Keystroke-Level Model for User Performance Time with Inter-active Systems, Comm. ACM, vol. 23, no. 7, 1980,pp. 396-410; GOMS was outlined in S.K. Card,T.P. Moran, and A. Newell, Computer Text-Edit-



ing: An Information-Processing Analysis of a Rou-tine Cognitive Skill, Cognitive Psychology, vol. 12,no. 1, 1980, pp. 396-410; S. Card, T.P. Moran,and A. Newell, The Psychology of HumanComputerInteraction, Lawrence Erlbaum Assoc., 1983.

23. Some consider a 1982 meeting in Gaithersburg tobe a CHI conference. SIGCHI was approved by theACM months later. The Gaithersburg proceedings,along with those of the 1981 SIGSOC workshop(Joint Conference on Easier and More Productive Useof Computer Systems) and a 1976 Siggraph work-shop (User-Oriented Design of Interactive GraphicsSystems) are all relevant and online at http://portal.acm.org/.

24. A. Newell and S.K. Card, The Prospects for Psy-chological Science in HumanComputer Interac-tion, HumanComputer Interaction, vol. 1, no. 3,1985, pp. 209-242.

25. D.A. Norman, Design Principles for HumanComputer Interfaces, Proc. ComputerHuman Inter-action Conf. (CHI 83), ACM Press, 1983, pp. 1-10.

26. S.K. Card, T.P. Moran, and A. Newell, Keystroke-Level Model for User Performance Time with Inter-active Systems, Comm. ACM, vol. 23, no. 7, 1980.

27. W.D. Gray et al., GOMS Meets the Phone Compa-ny: Analytic Modeling Applied to Real-World Prob-lems, Proc. IFIP Conf. Human-Computer Interaction(INTERACT 90), North-Holland, 1990, pp. 29-34.

28. J.D. Gould and C. Lewis, Designing for Usabili-tyKey Principles and What Designers Think,Proc. CHI 1983, ACM Press, 1983, pp. 50-53.

29. J. Grudin, The Computer Reaches Out: The His-torical Continuity of Interface Design, Proc. CHI1990, ACM Press, 1990, pp. 261-268.

30. J. Grudin and A. MacLean, Adapting a Psycho-physical Method to Measure Performance and Pref-erence Tradeoffs in HumanComputer Interaction,Proc. IFIP Conference on Human-Computer Interaction(INTERACT 84), North-Holland, 1984, pp. 338-342.

31. For histories of graphics and GUIs, see J. Johnsonet al., The Xerox Star: A Retrospective, Comput-er, vol. 22, no. 9, 1989, pp. 11-29; R. Baecker etal., Readings in HumanComputer Interaction:Toward the Year 2000; B.A. Myers, A Brief Historyof Human Computer Interaction Technology,ACM Interactions, vol. 5, no. 2, 1988, pp. 44-54;R. Pew, Evolution of HumanComputer Interac-tion: From MEMEX to Bluetooth and Beyond,The HumanComputer Interaction Handbook, J.A.Jacko and A. Sears, eds., Lawrence ErlbaumAssoc., 2003.

32. S.L. Smith and J.N. Mosier, Guidelines for DesigningUser Interface Software, ESD-TR-86-278, MITRE, 1986.

33. Natl Science Foundation, NSF 93-2: InteractiveSystems Program Description, 13 Jan. 1993.

34. Natl Science Foundation, NSF Committee of Visi-tors Report: Information and Intelligent SystemsDivision, 28 July 2003.

35. I. Benbasat and A.S. Dexter, An ExperimentalEvaluation of Graphical and Color-EnhancedInformation Presentation, Management Science,vol. 31, no. 11, 1985, pp. 1348-1364.

36. A. Friedman, Computer Systems Development: His-tory, Organization and Implementation, John Wiley& Sons, 1989.

37. G. Bjerknes, P. Ehn, and M. Kyng, eds., Computersand DemocracyA Scandinavian Challenge, Ave-bury, 1987.

38. A. Dennis et al., Information Technology to Sup-port Electronic Meetings, MIS Quarterly, vol. 12,no. 4, 1988, pp. 591-624; G. DeSanctis and R.B.Gallupe, A Foundation for the Study of GroupDecision Support Systems, Management Science,vol. 33, no. 5,1987, pp. 589-609.

39. J. Grudin, Managerial Use and Emerging Norms:Effects of Activity Patterns on Software Designand Deployment, Proc. Hawaii Intl Conf. SystemSciences (HICSS-37), IEEE CS Press, 2004; CD-ROM, 15 pp.

40. F.D. Davis, Perceived Usefulness, Perceived Easeof Use, and User Acceptance of Information Tech-nology, MIS Quarterly, vol. 13, no. 3, 1989, pp.319-339.

41. P. Zhang, F.F.-H. Nah, and J. Preece, HCI Studiesin Management Information Systems, Behaviourand Information Technology, vol. 23, no. 3, 2004,pp. 147-151; P. Zhang, AIS SIGHCI Three-YearReport, SIGHCI Newsletter, vol. 3, no. 1, 2004,pp. 2-6.

42. E.L. Hutchins, J.D. Hollan, and D.A. Norman,Direct Manipulation Interfaces, D.A. Normanand S.W. Draper, eds., User Centered SystemDesign, Lawrence Erlbaum Assoc., 1986.

43. A. Marcus, Branding 101, ACM Interactions,vol. 11, no. 5, 2004, pp. 14-21.

44. M.A. Blythe et al., eds., Funology: From Usability toUser Enjoyment, Kluwer Academic Publishers, 2003.

45. J. Nielsen, Usability Engineering at a Discount,Designing and Using HumanComputer Interfacesand Knowledge Based Systems, G. Salvendy andM.J. Smith, eds., Elsevier, 1989, pp. 394-401; H.Beyer and K. Holtzblatt, Contextual Design, Mor-gan Kaufmann, 1998; J. Pruitt and T. Adlin, ThePersona Lifecycle: Keeping People in Mind through-out Product Design, Morgan Kaufmann, 2005.

46. P. Barnard, Bridging between Basic Theories andthe Artifacts of HumanComputer Interaction,J.M. Carroll, ed., Designing Interaction: Psychology atthe HumanComputer Interface, Cambridge Univ.Press, 1991, pp. 103-127; J.M. Carroll and R.L.Campbell, Softening up Hard Science: Response toNewell and Card, HumanComputer Interaction,vol. 2, no. 3, 1986, pp. 227-249; J. Long, CognitiveErgonomics and HumanComputer Interaction, J.Long and A. Whitefield, eds., Cognitive Ergonomicsand HumanComputer Interaction, Cambridge Univ.


Press, 1989, pp. 4-34; A. Newell and S.K. Card,The Prospects for Psychological Science inHumanComputer Interaction, HumanComputerInteraction, vol. 1, no. 3, 1985, pp. 209-242.

47. J.M. Carroll, ed., HCI Models, Theories and Frame-works, Morgan Kaufmann, 2003.

48. D.A. Norman, Steps toward a Cognitive Engineer-ing: Design Rules Based on Analyses of HumanError, Proc. Conf. Human Factors in Computing Sys-tems, ACM Press, 1983, pp. 378-382; D.A. Norman,Cognitive Engineering, D.A. Norman and S.W.Draper, eds., User Centered System Design, LawrenceErlbaum Assoc., 1986; D.A. Norman, Design Prin-ciples for HumanComputer Interfaces, Proc. CHI1983, ACM Press, pp. 1-10; D.A. Norman, Psychol-ogy of Everyday Things, Basic Books, 1988; D.A.Norman, Emotional Design: Why We Love (or Hate)Everyday Things, Basic Books, 2004.

49. J. Grudin, Crossing the Divide, ACM Trans.ComputerHuman Interaction, vol. 11, no. 1,2004, pp. 1-25.

50. General journals such as Comm. ACM and HumanFactors were not considered. IJMMS/IJHCS andBehaviour and Information Technology (BIT) wereselected for Human Factors and Ergonomics(HF&E), Management Information Systems Quarterlyand Information Systems Research for IS, HCI andACM TOCHI for CHI. Data fluctuate from year toyear; medians were rounded to the nearest 10percent. Aggregate conference paper numbers arecomparable for the three groups. HF&E data arefrom the HumanComputer Interaction Internation-al conference and the computer systems track of theannual meeting. IS data are from the InternationalConference on Information Systems (its 20 percentacceptance rate is the greatest outlier), the Organi-zational Communication and Information Systemsand Technology and Innovation Managementtracks of the Academy of Management annualmeeting (60 percent), IS tracks at the Hawaii Inter-national Conference on System Sciences (50 per-cent), and IS HCI tracks at five conferences(median percent). The CHI, User Interface Softwareand Technology, and CSCW conference medianacceptances over the past three years were 16 per-cent, 21 percent, and 20 percent, respectively.

51. J. Grudin, Crossing the Divide, ACM Trans.ComputerHuman Interaction, vol. 11, no. 1,2004, pp. 1-25; J. Nunamaker, introductory com-ments presented at HICSS 2004 (IS data).

52. J. Grudin, Interface: An Evolving Concept,Comm. ACM, vol. 36, no. 4, 1993, pp. 110-119; J.Grudin, Human Factors, CHI and MIS, HumanComputer Interaction and Management Informa-tion Systems: Foundations, P. Zhang and D. Gallet-ta, eds., Advances in Management InformationSystems, vol. 4, M.E. Sharpe, in press.

53. C. Lewis, The Thinking Aloud Method in Inter-

face Evaluation, tutorial given at CHI 83, 1983;C. Lewis and R. Mack, Learning to Use a TextProcessing System: Evidence from ThinkingAloud Protocols, Proc. Conf. Human Factors inComputing Systems, ACM Press, 1982, pp. 387-392; A. Newell and H.A. Simon, Human ProblemSolving, Prentice-Hall, 1972.

54. E. Israelski and A.M. Lund, The Evolution ofHumanComputer Interaction during the Telecom-munications Revolution, The HumanComputerInteraction Handbook, J.A. Jacko and A. Sears, eds.,Lawrence Erlbaum Assoc., 2003, pp. 772-789.

55. K.D. Forbus, J. Usher, and V. Chapman, Qualita-tive Spatial Reasoning about Sketch Maps, Proc.Innovative Applications of AI, American Associationfor Artificial Intelligence (AAAI), 2003, pp. 85-92;K.D. Forbus, Sketching for Knowledge Capture,lecture, Microsoft Research, Redmond, Wash., 2May 2003.The system included speech recogni-tion capability, but the investigator concealedthis from the customers who once funded him tobuild it. A description of the system, its use withspeech turned off, and a warning against usingspeech is found in R. Rasch, A. Kott, and K.D. For-bus, Incorporating AI into Military DecisionMaking: An Experiment, IEEE Intelligent Systems,vol. 18, no. 4, 2003, pp. 18-26.

56. D.A. Norman, Psychology of Everyday Things,Basic Books, 1988.

57. D. Meister, History of Human Factors andErgonomics. Quotation is from email to author,September 2004.

58. For example, D. Peebles and P.C.H. Cheng,Modeling the Effect of Task and Graphical Rep-resentation on Response Latency in a GraphReading Task, Human Factors, vol. 45, no. 1,2003, pp. 28-46.

Jonathan Grudin is a seniorresearcher at Microsoft Research,focusing on digitally mediat-ed interaction. He was previ-ously professor of informationand computer science at theUniversity of California, Irvine.He was editor in chief of ACM

Transactions on ComputerHuman Interaction from 1997to 2003 and is associate editor for humancomputerinteraction for ACM Computing Surveys. After workingas a software developer in the 1970s, he obtained aPhD in cognitive psychology at UC San Diego in 1981.

Readers may contact Jonathan Grudin about thisarticle at [email protected].

For further information on this or any other com-puting topic, please visit our Digital Library at http://www.computer.org/publications/dlib.



HCI

Documents

Transcript of HCI