Skorl ana LjLori

UrcjaYLLzailonal

onous

Cbarles Perrow

rom the beginning, the forces of light and theforces of darkness have polarized the field oforganizational analysis, and the struggle hasbeen protracted and inconclusive. The forcesof darkness have been represented by the me-chanical school of organizational theory—those who treat the organization as a machine.This school characterizes organizations interms o£ such things as:

• centralized authority• clear lines of authority• specialization and expertise• marked division of labor• rules and regulations• clear separation of staff and UneThe forces of light, which by mid-

twentieth century came to be characterized asthe human relations school, emphasizes peoplerather than machines, accommodations ratherthan machine-like precision, and draws its in-spiration from biological systems rather thanengineering systems. It has emphasized suchthings as:

• delegation of authority• employee autonomy• trust and openness• concerns with the "whole person"• interpersonal dynamics

T H E RISE AI-O) FALL OF

SCIENTIFIC MANAGEMENT

The forces of darkness formulated their posi-tion first, starting in the early part of thiscentury. They have been characterized as thescientific management or classical manage-ment school. This school started by paradingsimple-minded injunctions to plan ahead,

keep records, write down policies, specialize,be decisive, and keep your span of control toabout six people. These injunctions wereneeded as firms grew in size and complexity,since there were few models around beyondthe railroads, the military, and the CatholicChurch to guide organizations. And their in-junctions worked. Executives began to dele-gate, reduce their span of control, keep records,and specialize. Planning ahead still is difficult,it seems, and the modern equivalent is Man-agement by Objectives.

But many things intruded to makethese simple-minded injunctions less relevant:

1. Labor became a more critical factorin the firm. As the technology increased in

sophistication it took longer to train people,and more varied and specialized skills wereneeded. Thus, labor turnover cost more andrecruitment became more selective. As a con-sequence, labor's power increased. Unions andstrikes appeared. Management adjusted bybeginning to speak of a cooperative system ofcapital, management, and labor. The machinemodel began to lose its relevancy.

2. The increasing complexity of mar-kets, variability of products, increasing num-ber of branch plants, and changes in technol-ogy all required more adaptive organization.The scientific management school was ill-equipped to deal with rapid change. It hadpresumed that once the proper structure wasachieved the firm could run forever withoutmuch tampering. By the late 1930s, people be-gan writing about adaptation and change inindustry from an organizational point of viewand had to abandon some of the principles ofscientific management.

3. Political, social, and culturalchanges meant new expectations regardingthe proper way to treat people. The dark,Satanic mills needed at the least a white-wash-ing. Child labor and the brutality of supervi-sion in many enterprises became no longerpermissible. Even managers could not be ex-pected to accept the authoritarian patterns ofleadership that prevailed in the small firm runby the founding father.

4. As mergers and growth proceededapace and the firm could no longer be viewedas the shadow of one man (the founding en-trepreneur), a search for methods o£ selectinggood leadership became a preoccupation. Agood, clear, mechanical structure would nolonger suffice. Instead, firms had to search forthe qualities of leadership that could fill thelarge footsteps of the entrepreneur. They tac-itly had to admit that something other thaneither "sound principles" or "dynamic leader-

4 ship" was needed. The search for leadership

traits implied that leaders were made, not justborn, that the matter was complex, and thatseveral skills were involved.

ENTER HUMAN RELATIONS

From the beginning, individual voices wereraised against the implications of the scientificmanagement school. "Bureaucracy" had al-ways been a dirty word, and the job designefforts of Frederick Taylor were even the sub-ject of a congressional investigation. But noeffective counterforce developed until 1938,when a business executive with academic tal-ents named Chester Barnard proposed the firstnew theory of organizations: Organizationsare cooperative systems, not the products ofmechanical engineering. He stressed naturalgroups within the organization, upward com-munication, authority from below rather thanfrom above, and leaders who functioned as acohesive force. With the spectre of labor un-rest and the Great Depression upon him,Barnard's emphasis on the cooperative natureof organizations was well-timed. The yearfollowing the publication of his Functions ofthe Executive (1938) saw the publication ofF. J. Roethlisberger and William Dickson'sManagement and the Worf^er, reporting onthe first large-scale empirical investigation ofproductivity and social relations. The re-search, most of it conducted in the Hawthorneplant of the Western Electric Company dur-ing a period in which the workforce was re-duced, highlighted the role of informalgroups, work restriction norms, the value ofdecent, humane leadership, and the role ofpsychological manipulation of employeesthrough the counseling system. World War 11intervened, but after the war the human re-lations movement, building on the insightsof Barnard and the Hawthorne studies, cameinto its own.

Whde Charles Perrow's primary affiliation hasbeen with sociology departments in a varietyof Universities {Michigan. Pittsburgh, Wis-consin, and now, the State University o/ NewYor){^ at Stony Broo\), he has tempered histheorizing by also teaching in professionalschools at these universities. "Business stu-dents want to i{riou> what difference theorymakes, and it is a salutory discipline jor thesociologists," he notes. This resulted in abook^ for business managers. OrganizationalAnalysis: A Sociological View, in 1970. Hehas published three other boo^s dealing withorganizations, as well as more than 20 schol-arly articles. The present article has grown outof his recent volume. Complex Organizations;A Critical Essay (1972).

He has consulted with business, got/-einment and voluntary agencies {but verymodestly), spent a year at the Industrial Re-lations Institute in Berkeley and the BusinessSchool there, in the year of the People's Parl^and the Cleaver demonstrations. Currently heIS spending this year at the London BusinessSchool and is running two-day training ses-sions on the structural analysis of organiza-tions for managers. He is also worthing withsome groups on the social responsibilities ofbusiness. Professor Perrow is on the editorialboard of Administrative Science Quarterlyand the American Sociological Review. Hisother hookas are The Radical Attack on Busi-ness, 1972, and Organization for Treatment,1966, with D. Street and R. Vinter.

The first step was a search for thetraits of good leadership. It went on furiouslyat university centers but at first failed to pro-duce more than a list of Boy Scout maxims: Agood leader was kind, courteous, loyal, coura-geous, etc. We suspected as much. However,the studies did turn up a distinction between"consideration," or employee-centered aspectsof leadership, and job-centered, technical as-pects labled "initiating structure." Both wereimportant, but the former received most ofthe attention and the latter went undeveloped.The former led directly to an examination ofgroup processes, an investigation that hasculminated in T-group programs and ismoving forward still with encounter groups.Meanwhile, in England, the Tavistock Insti-tute sensed the importance of the influence ofthe kind of task a group had to perform on thesocial relations within the group. The first im-portant study, conducted among coal miners,showed that job simplification and specializa-tion did not work under conditions of uncer-tainty and nonroutine tasks.

As this work flourished and spread,more adventurous theorists began to extend itbeyond work groups to organizations as awhole. We now knew that there were a num-ber o£ things that were bad for the moraleand loyalty of groups—routine tasks, submis-sion to authority, specialization of task, segre-gation of task sequence, ignorance of the goalsof the firm, centralized decision making, andso on. If these were bad for groups, they werelikely to be bad for groups of groups—i.e.,for organizations. So people like WarrenBennis began talking about innovative, rap-idly changing organizations that were madeup of temporary groups, temporary authoritysystems, temporary leadership and role as-signments, and democratic access to the goalsof the firm. If rapidly changing technologiesand unstable, turbulent environments were tocharacterize industry, then the structure of 5

firms should be temporary and decentralized.The forces of light, of freedom, autonomy,change, humanity, creativity, and democracywere winning. Scientific management sur-vived only in outdated text books. If the evan-gelizing of some of the human relations schooltheorists was excessive, and if Likert's System4 or MacGregor's Theory Y or Blake's 9 x 9evaded us, at least there was a rationale forconfusion, disorganization, scrambling, andstress: Systems should be temporary.

BUREAUCRACY'S COMEBACK

Meanwhile, in another part of the manage-ment forest, the mechanistic school wasgathering its forces and preparing to outflankthe forces of light. First came the numbersmen—the linear programmers, the budget ex-perts, and the financial analysts—with theirPERT systems and cost-benefit analyses.From another world, unburdened by most ofthe scientific management ideology and un-touched by the human relations school, theybegan to parcel things out and give somemeaning to those truisms, "plan ahead" and"keep records." Armed with emerging sys-tems concepts, they carried the "mechanistic"analogy to its fullest—and it was very produc-tive. Their work still goes on, largely un-troubled by organizational theory; the theory,it seems clear, will have to adjust to them,rather than the other way around.

Then the works of Max Weber, firsttranslated from the German in the 1940s—hewrote around 1910, incredibly—began to findtheir way into social science thought. At first,with his celebration of the efficiency of bu-reaucracy, he was received with only reluctantrespect, and even with hostility. All writerswere against bureaucracy. But it turned out,surprisingly, that managers were not. Whenasked, they acknowledged that they preferred

clear lines of communication, clear specifica-tions of authority and responsibility, and clearknowledge of whom they were responsible to.They were as wont to say "there ought to be arule about this," as to say "there are too manyrules around here," as wont to say "next weekwe've got to get organized," as to say "there istoo much red tape." Gradually, studies beganto show that bureaucratic organizations couldchange faster than nonbureaucratic ones, andthat morale could be higher where there wasclear evidence of bureaucracy.

What was this thing, thcn.^ Weberhad showed us, for example, that bureaucracywas the most effective way of ridding organ-izations of favoritism, arbitrary authority, dis-crimination, payola and kick-backs, and yes,even incompetence. His model stressed ex-pertise, and the favorite or the boss' nephewor the guy who burned up resources to makehis performance look good was not the onewith expertise. Rules could be changed; theycould be dropped in exceptional circum-

stances; job security promoted more innova-tion. The sins of bureaucracy began to looklike the sins of failing to follow its principles.

ENTER POWER, CONFLICT, AND DECISIONS

But another discipline began to intrude uponthe confident work and increasingly elaboratemodels of the human relations theorists(largely social psychologists) and the uneasytoying with bureaucracy of the "structional-ists" (largely sociologists). Both tended tostudy economic organizations. A few, likePhilip Selznick, were noting conflict and dif-ferences in goals (perhaps because he wasstudying a public agency, the Tennessee ValleyAuthority), but most ignored conflict ortreated it as a pathological manifestation ofbreakdowns in communication or the egotrips of unreconstructed managers.

But in the world of political parties,pressure groups, and legislative bodies, con-flict was not only rampant, but to be expected—it was even functional. This was the do-main of the political scientists. They kept talk-ing about power, making it a legitimateconcern for analysis. There was an open ac-knowledgement of "manipulation." Thesewere political scientists who were "behavior-ally" inclined—studying and recording behav-ior rather than constitutions and formal sys-tems of government—and they came to a muchmore complex view of organized activity. Itspilled over into the area of economic organ-izations, with the help of some economistslike R. A. Gordon and some sociologists whowere studying conflicting goals of treatmentand custody in prisons and mental hospitals.

The presence of legitimately conflict-ing goals and techniques of preserving andusing power did not, of course, sit well with acooperative systems view of organizations.But it also puzzled the bureaucratic school

(and what was left of the old scientific man-agement school), for the impressive Weberianprinciples were designed to settle questions ofpower through organizational design and tokeep conflict out through reliance on rational-legal authority and systems of careers, ex-pertise, and hierarchy. But power was beingovertly contested and exercised in covertways, and conflict was bursting out all over,and even being creative.

Gradually, in the second half of the1950s and in the next decade, the politicalscience view infiltrated both schools. Conflictcould be healthy, even in a cooperative sys-tem, said the human relationists; it was themode of resolution that counted, rather thanprevention. Power became reconceptualizedas "influence," and the distribution was lessimportant, said Arnold Tannenbaum, thanthe total amount. For the bureaucratic school—never a clearly defined group of people, andlargely without any clear ideology—it waseasier to just absorb the new data and theoriesas something else to be thrown into the pot.That is to say, they floundered, writing booksthat went from topic to topic, without a clearview of organizations, or better yet, produc-ing "readers" and leaving students to sort itall out.

Buried in the political science view-point was a sleeper that only gradually beganto undermine the dominant views. This wasthe idea, largely found in the work of Her-bert Simon and James March, that becauseman was so limited—in intelligence, reason-ing powers, information at his disposal, timeavailable, and means of ordering his prefer-ences clearly—he generally seized on the firstacceptable alternative when deciding, ratherthan looking for the best; that he rarelychanged things unless they really got bad, andeven then he continued to try what hadworked before; that he limited his search forsolutions to well-worn paths and traditional 7

sources of information and established ideas;that he was wont to remain preoccupied withroutine, thus preventing innovation. Theycalled these characteristics "cognitive limitson rationality" and spoke of "satisficing"rather than maximizing or optimizing. It isnow called the "decision making" school, andis concerned with the basic question of howpeople make decisions.

This view had some rather unusualimplications. It suggested that if managerswere so limited, then they could be easily con-trolled. What was necessary was not to givedirect orders (on the assumption that subor-dinates were idiots without expertise) or toleave them to their own devices (on the as-sumption that they were supermen whowould somehow know what was best for theorganization, how to coordinate with all theother supermen, how to anticipate marketchanges, etc.). It was necessary to control onlythe premises of their decisions. Left to them-selves, with those premises set, they could bepredicted to rely on precedent, keep thingsstable and smooth, and respond to signals thatreinforce the behavior desired of them.

To control the premises of decisionmaking, March and Simon outline a varietyof devices, all of which are familiar to you,but some of which you may not have seen be-fore in quite this light. For example, organ-izations develop vocabularies, and this meansthat certain kinds of information are high-lighted, and others are screened out—just asEskimos (and skiers) distinguish many va-rieties of snow, while Londoners see only one.This is a form of attention directing. Anotheris the reward system. Change the bonus forsalesmen and you can shift them from volumeselling to steady-account selling, or to sellingquality products or new products. If you wantto channel good people into a different func-tion (because, for example, sales should no

8 longer be the critical function as the mar-

ket changes, but engineering applicationsshould), you may have to promote mediocrepeople in the unrewarded function in orderto signal to the good people in the rewardedone that the game has changed. You cannotexpect most people to make such decision ontheir own because of the cognitive limits ontheir rationality, nor will you succeed by giv-ing direct orders, because you yourself proba-bly do not know whom to order where. Youpresume that once the signals are clear andthe new sets of alternatives are manifest,they have enough ability to make the decisionbut you have had to change the premises fortheir decisions about their career lines.

It would take too long to go throughthe dozen or so devices, covering a range ofdeciston areas (March and Simon are not thatclear or systematic about them, themselves, soI have summarized them in my own book),but I think the message is clear.

It was becoming clear to the humanrelations school, and to the bureaucraticschool. The human relationists had begun tospeak of changing stimuli rather than chang-ing personality. They had begun to see thatthe rewards that can change behavior can wellbe prestige, money, comfort, etc., rather thantrust, openness, self-insight, and so on. Thealternative to supportive relations need notbe punishment, since behavior can best bechanged by rewarding approved behaviorrather than by punishing disapproved behav-ior. They were finding that although leader- •ship may be centralized, it can function bestthrough indirect and unobtrusive means suchas changing the premises on which decisionsare made, thus giving the impression that thesubordinate is actually making a decisionwhen he has only been switched to a differ-ent set of alternatives. The implications ofthis work were also beginning to filter intothe human relations school through an em-phasis on behavioral psychology (the modern

version of the much maligned stimulus-response school) that was supplanting per-sonality theory (Freudian in its roots, anddrawing heavily, in the human relationsschool, on Maslow).

For the bureaucratic school, this newline of thought reduced the heavy weightplaced upon the bony structure of bureaucracyby highlighting the muscle and flesh thatmake these bones move. A single chain ofcommand, precise division of labor, and clearlines of communication are simply notenough in themselves. Control can beachieved by using alternative communicationchannels, depending on the situation; by in-creasing or decreasing the static or "noise" inthe system; by creating organizational mythsand organizational vocabularies that allowonly selective bits of information to enter thesystem; and through monitoring performancethrough indirect means rather than direct sur-veillance. Weber was all right for a starter,but organizations had changed vastly, andthe leaders needed many more means of con-trol and more subtle means of manipulationthan they did at the turn of the century.

THE TECHNOLOGICAL QUALIFICATION

By now the forces of darkness and forces oflight had moved respectively from midnight

and noon to about 4 AJ . and 8 A.M. But anyconvergence or resolution would have to beon yet new terms, for soon after the politicalscience tradition had begun to infiltrate theestablished schools, another blow struck bothof the major positions. Working quite inde-pendently of the Tavistock Group, with itsemphasis on sociotechnical systems, and be-fore the work of Burns and Stalker on mech-anistic and organic firms, Joan Woodwardwas trying to see whether the classical scien-tific principles of organization made any sensein her survey of 100 firms in South Essex. Shetripped and stumbled over a piece of gold inthe process. She picked up the gold, labeled it"technology," and made sense out of herotherwise hopeless data. Job-shop firms, mass-production firms, and continuous-processfirms all had quite different structures becausethe type of tasks, or the "technology," was dif-ferent. Somewhat later, researchers in Amer-ica were coming to very similar conclusionsbased on studies of hospitals, juvenile correc-tional institutions, and industrial firms. Bu-reaucracy appeared to be the best form of or-ganization for routine operations; temporarywork groups, decentralization, and emphasison interpersonal processes appeared to workbest for nonroutine operations. A raft of stud-ies appeared and are still appearing, all tryingto show how the nature of the task aflects thestructure of the organization.

alternative to supportive relations need

not be punishment, since behavior can best be

changed by rewarding approved behavior rather

than by punishing disapproved behavior/'

10

This severely complicated tbings forthe human relations school, since it suggestedthat openness and trust, while good things inthemselves, did not have much impact, or per-haps were not even possible in some kinds ofwork situations. The prescriptions tbat werebeing handed out would have to be drasticallyqualified. What might work for nonroutine,high-status, interesting, and challenging jobsperformed by highly educated people mightnot be relevant or even beneficial for the vastmajority of jobs and people.

It also forced the upholders of therevised bureaucratic theory to qualify theirrecommendations, since research and develop-ment units should obviously be run differ-ently from mass-production units, and thedifference between both of these and highlyprogrammed and highly sophisticated con-tinuous-process firms was obscure in terms ofbureaucratic theory. But the bureaucraticschool perhaps came out on top, because tbeforces o£ evil—authority, structure, division oflabor, etc.—no longer looked evil, even i£they were not applicable to a minority of in-dustrial units.

The emphasis on tecbnology raisedotber questions, however. A can companymight be quite routine, and a plastics divisionnonroutine, but there were both routine andnonroutine units within each. How shouldthey be integrated if the prescription werefollowed tbat, say, production sbould be bu-reaucratized and R&D not? James Thompsonbegan spelling out different forms of interde-pendence among units in organizations, andPaul Lawrence and Jay Lorsch looked closelyat the nature of integrating mechanisms. Law-rence and Lorsch found that firms performedbest when the differences between units weremaximized (in contrast to both the humanrelations and the bureaucratic school), as longas the integrating mechanisms stood half-waybetween the two—being neither strongly bu-

reaucratic nor nonroutine. They also notedtbat attempts at participative management inroutine situations were counterproductive,that the environments of some kinds of or-ganizations were far from turbulent and cus-tomers did not want innovations and changes,that cost reduction, price, and efficiency weretrivial considerations in some firms, and so on.The technological insight was demolishingour comfortable truths right and left. Theywere also being questioned from anotherquarter.

ENTER GOALS, ENVIRONMENTS, AND SYSTEMS

The final seam was being mined by the so-ciologists while all this went on. This was theconcern with organizational goals and the en-vironment. Borrowing from the politicalscientists to some extent, but pushing aheadon their own, this "institutional school" cameto sec that goals were not fixed; conflicting

goals could be pursued simultaneously, ifthere were enough slack resources, or sequen-tially (growth for the next four years, thencost-cutting and profit-taking for the nextfour); that goals were up for grabs in organ-izations, and units fought over them. Goalswere, of course, not what they seemed to be,the important ones were quite unofficial; his-tory played a big role; and assuming profit asthe pre-eminent goal explained almost nothingabout a firm's behavior.

They also did case studies that linkedthe organization to the web of influence ofthe environment; that showed how unique or-ganizations were in many respects (so that,once again, there was no one best way to dothings for all organizations); how organiza-tions were embedded in their own history,making change difficult. Most striking of all,perhaps, the case studies revealed that thestated goals usually were not the real ones;the official leaders usually were not the power-ful ones; claims of effectiveness and efficiencywere deceptive or even untrue; the public in-terest was not being served; political influ-ences were pervasive; favoritism, discrimina-tion, and sheer corruption were commonplace.The accumulation of these studies presentedquite a pill for either the forces of light ordarkness to swallow, since it was hard to seehow training sessions or interpersonal skillswere relevant to these problems, and it wasalso clear that the vaunted efficiency of bu-reaucracy was hardly in evidence. What couldthey make of this wad of case studies.'

We are still sorting it out. In onesense, the Weberian model is upheld becauseorganizations are not, by nature, cooperativesystems; top managers must exercise a greatdeal of effort to control them. But if organiza-tions are tools in the hands of leaders, theymay be very recalcitrant ones. Like the broomin the story of the sorcerer's apprentice, theyoccasionally get out of hand. If conflicting

goals, bargaining, and unofficial leadershipexists, where is the structure of Weberianbones and Simonian muscle? To what extentare organizations tools, and to what extent arethey products of the varied interests andgroup strivings of their members.' Does itvary by organization, in terms of some typo-logical alchemy we have not discovered.' Wedon't know. But at any rate, the bureaucraticmodel suffers again; it simply has not reck-oned on the role of the environment. Thereare enormous sources of variations that theneat, though by now quite complex, neo-Weberian model could not account for.

Tbe human relations model has alsobeen badly shaken by the findings of the in-stitutional school, for it was wont to assumethat goals were given and unproblematical,and that anything that promoted harmonyand efficiency for an organization also wasgood for society. Human relationists as-sumed that the problems created by organiza-tions were largely limited to the psychologicalconsequences of poor interpersonal rela-tions within them, rather than their impacton the environment. Could the organizationreally promote the psychological health of itsmembers when by necessity it had to definepsychological health in terms of the goals ofthe organization itself.' The neo-Weberianmodel at least called manipulation "manipu-lation" and was skeptical of claims about au-tonomy and self-realization.

But on one thing all the variedschools of organizational analysis now seemedto be agreed: organizations are systems—in-deed, they are open systems. As the growthof the field has forced ever more variablesinto our consciousness, flat claims of predic-tive power are beginning to decrease and re-search has become bewilderingly complex.Even consulting groups need more than oneor two tools in their kit-bag as the softwaremultiplies.

12

The systems view is intuitively sim-ple. Everything is related to everything else,though in uneven degrees of tension andreciprocity. Every unit, organization, depart-ment, or work group takes in resources,transforms them, and sends them out, andthus interacts with the larger system. The psy-chological, sociological, and cultural aspects ofunits interact. The systems view was explicitin the institutional work, since they triedto study whole organizations; it became ex-plicit in the human relations school, becausethey were so concerned with the interactionsof people. The political science and tech-nology viewpoints also had to come to thisrealization, since they dealt with parts affect-ing each other (sales affecting production;technology affecting structure).

But as intuitively simple as it is, thesystems view has been difficult to put intopractical use. We still Hnd ourselves ignoringthe tenets of the open systems view, possiblybecause of the cognitive limits on our rational-ity. General systems theory itself has not livedup to its heady predictions; it remains rather

nebulous. But at least there is a model for call-ing us to account and for stretching ourminds, our research tools, and our troublednostrums.

SOME CONCLUSIONS

Where does all this leave us} We might sum-marize the prescriptions and proscriptions formanagement very roughly as follows:

1. A great deal of the "variance" in afirm's behavior depends on the environment.We have become more realistic about the lim-ited range of change that can be inducedthrough internal efforts. The goals of organi-zations, including those of profit and effi-ciency, vary greatly among industries andvary systematically by industries. This sug-gests that the impact of better managementby itself will be limited, since so much willdepend on market forces, competition, legisla-tion, nature of the work force, available tech-nologies and innovations, and so on. An-other source of variation is, obviously, thehistory of the firm and its industry and itstraditions.

2. A fair amount of variation in bothfirms and industries is due to the type of workdone in the organization—the technology.We are now fairly confident in recommend-ing that if work is predictable and routine,the necessary arrangement for getting thework done can be highly structured, and onecan use a good deal of bureaucratic theory inaccomplishing this. If it is not predictable, ifit is nonroutinc and there is a good deal of un-certainty as to how to do a job, then one hadbetter utilize the theories that emphasizeautonomy, temporary groups, multiple linesof authority and communications, and so on.We also know that this distinction is impor-tant when organizing different parts of an or-ganization.

We are also getting a grasp on thequestion of what is the most critical functionin different types of organizations. For someorganizations it is production; for others,marketing; for still others, development. Fur-thermore, firms go through phases wherebythe initial development of a market or a prod-uct or manufacturing process or accountingscheme may require a non-bureaucratic struc-ture, but once it comes on stream, the struc-ture should change to reflect the changedcharacter of the work.

3. In keeping with this, managementshould be advised that the attempt to producechange in an organization through manage-rial grids, sensitivity training, and even job en-richment and job enlargement is likely to befairly ineffective for all but a few organiza-tions. The critical reviews of research in allthese fields show that there is no scientificevidence to support the claims of the propo-nents of these various methods; that researchhas told us a great deal about social psychol-ogy, but little about how to apply the highlycomplex findings to actual situations. The keyword is selectivity: We have no broad-spec-trum antibiotics for interpersonal relations. Ofcourse, managers should be sensitive, decent,kind, courteous, and courageous, but we have

known that for some time now, and beyond aminimal threshold level, the payoff is hard tomeasure. The various attempts to make workand interpersonal relations more humane andstimulating should be applauded, but weshould not confuse this with solving problemsof structure, or as the equivalent of decentrali-zation or participatory democracy.

4. The burning cry in all organiza-tions is for "good leadership," but we havelearned that beyond a threshold level of ade-quacy it is extremely difficult to know whatgood leadership is. The hundreds of scientificstudies of this phenomenon come to one gen-eral conclusion: Leadership is highly variableor "contingent" upon a large variety of im-portant variables such as nature of task, sizeof the group, length of time the group hasexisted, type of personnel within the groupand their relationships with each other, andamount of pressure the group is under. It docsnot seem likely that we'll be able to devise away to select the best leader for a particularsituation. Even if we could, that situationwould probably change in a short time andthus would require a somewhat different typeof leader.

Furthermore, we are beginning torealize that leadership involves more than

'The burning cry in all organizations is for

'good leadership/ but we have learned that

beyond a threshold level of adequacy it is

extremely difficult to know what good

leadership is/'13

smoothing the paths of human interaction.What has rarely been studied in this area isthe wisdom or even the technical adequacy ofa leader's decision. A leader does more thanlead people; he also makes decisions about theallocation of resources, type of technology tobe used, the nature of the market, and so on.This aspect of leadership remains very ob-scure, but it is obviously crucial.

5. If we cannot solve our problemsthrough good human relations or throughgood leadership, what are we then left with?The literature suggests that changing thestructures of organizations might be the mosteffective and certainly the quickest and cheap-est method. However, we are now sophisti-cated enough to know that changing the for-mal structure by itself is not likely to producethe desired changes. In addition, one must beaware of a large range of subtle, unobtrusive,and even covert processes and change de-vices that exist. If inspection procedures arenot working, we are now unlikely to rush inwith sensitivity training, nor would we senddown authoritative communications tellingpeople to do a better job. We are more likelyto find out where the authority really lies,whether the degree of specialization is ade-quate, what the rules and regulations are, andso on, but even this very likely will not beenough.

According to the neo-Weberian bu-reaucratic model, as it has been influenced bywork on decision making and behavioral psy-chology, we should find out how to ma-nipulate the reward structure, change thepremises of the decision-makers through finercontrols on the information received and theexpectations generated, search for interdepart-mental conflicts that prevent better inspectionprocedures from being followed, and aftermanipulating these variables, sit back andwait for two or three months for them to take

14 hold. This is complicated and hardly as

dramatic as many of the solutions currentlybeing peddled, but I think the weight of or-ganizational theory is in its favor.

We have probably learned more,over several decades of research and theory,about the things that do not work (eventhough some of them obviously should haveworked), than we have about things that dowork. On balance, this is an important gainand should not discourage us. As you know,organizations are extremely complicated. Tohave as much knowledge as we do have in afledgling discipline that has had to borrowfrom the diverse tools and concepts of psy-chology, sociology, economics, engineering,biology, history, and even anthropology is notreally so bad.

SELECTED BIBLIOGRAPHY

This paper is an adaptation of the discussion tobe found in Charles Perrow, Complex Organiza-tions: A Critical Essay, Scott, Forcsman & Co.,GlcnvJlle, Illinois, 1972. AU the points made inthis paper are discussed thoroughly in that vol-ume.

The best overview and discussion of clas-sical management theory, and its changes overtime is by Joseph Massie—"Management Theory"in the Handbook^ of Organizations edited byJames March, Rand McNally & Co., Chicago,1965, pp.387-422.

The best discussion of the changingjustifications for managerial rule and workerobedience as they are related to changes in tech-nology, etc., can be found in Reinhard Bendix'sH or and Authority in Industry, John Wiley &Sons, Inc., New York, 1956. See especially thechapter on the American experience.

Some of the leading tights of the classi-

cal view—F. W. Taylor, Col. Urwick, and HenryFayol—arc briefly discussed in Writers on Organi-zations by D. S. Pugh, D. J. Hickson and C. R.Hinings, Penguin, 1971. This brief, readable,and useful book also contains selections frommany other schools that I discuss, includingWeber, Woodward, Cyert and March, Simon,the Hawthorne Investigations, and the HumanRelations Movement as represented by Argyris,Hcrzberg, Likcrt, McGregor, and Blake andMouton.

As good a place as any to start examin-ing the human relations tradition is Rensis Likert,The Human Organization, McGraw-Hill, NewYork, 1967. See also bis New Patterns of Man-agement, McGraw-Hill Book Company, NewYork, 1961.

The Buck Rogers scbool of organiza-tional tbeory is best represented by Warren Bcn-nis. See bis Changing Organizations, McGraw-Hill Book Company, New York, 1966, and bisbook with Pbillp Slater, The Temporary Society,Harper & Row, Inc., New York, 1968. Mucbo£ tbis work is linked into more general studies,e.g., Alvin TofHer's very popular paperbackFuture Shoc\, Random House, 1970, and Ban-tam Paperbacks, or Zibigniew Brzezinsky's Be-tween Two Ages: America's Role in the Techni-tronic Era, the Viking Press, New York, 1970.One of the first intimations of tbe new type ofenvironment and firm and still perhaps tbemost perceptive is to be found in tbe volumeby Tom Burns and G. Stalker, The Managementof Innovation, Tavistock, London, 1961, wberethey distinguished between "organic" and "mech-anistic" systems. Tbe introduction, wbicb is notvery long, is an excellent and very tight summaryof tbe book.

Tbe political science tradition came intbrougb tbree important works. First, HerbertSimon's Administrative Behavior. Tbe MacMil-lan Co., New York, 1948, followed by tbe secondbalf of James Marcb and Herbert Simon's Organi-zations, John Wiley 6t Sons, Inc., New York,1958, tben Ricbard M. Cyert and James Marcb'sA Behavioral Theory of the Firm, Prentice-Hall,

Inc. Englewood Cliffs, N.J., 1963. All three oftbese books arc fairly rougb going, tbougbcbapters 1, 2, 3, and 6 of tbe last volume arefairly sbort and accessible. A quite interestingbook in this tradition, tbougb somtwbat beavy-going, is Micbael Crozier's The BureaucraticPhenomenon, University of Cbicago, and Tavi-stock Publications, 1964. Tbis is a striking de-scription of power in organizations, tbougbtbere is a somewbat dubious attempt to linkorganization processes in France to tbe culturaltraits of tbe Frcncb people.

Tbe book by Joan Woodward IndustrialOrganisation: Theory and Practice, Oxford Uni-versity Press, London, 1965, is still very mucbwortb reading. A fairly popular attempt to dis-cuss the implications for tbis for managementcan be found in my own book. OrganizationalAnalysis: A Sociological View, Tavistock, 1970,Chapters 2 and 3. Tbe impact of tecbnology onstructure is still fairly controversial. A number oftecbnical studies bave found botb support andnonsupport, largely because tbe concept is de-fined so differently, but tbere is general agree-ment tbat different structures and leadershiptecbniques arc needed for different situations.For studies tbat support and document this view-point sec James Tbompson, Organizations inAction, McGraw-Hill Book Company, NewYork, 1967, and Paul Lawrence and Jay Lorscb,Organizations and Environment, Harvard Uni-versity Press, Cambridge, Mass., 1967.

Tbe best single work on tbe relation be-tween tbe organization and tbe environment andone of tbe most readable books in the field isPbilip Selznick's sbort volume Leadership in Ad-ministration, Row, Peterson, Evanston, Illinois,1957. But the large number of tbese studies arescattered about. I have summarized several in myComplex Organizations: A Critical Essay.

Lasdy, tbe most elaborate and persuasiveargument for a systems view of organizations isfound in tbe first 100 pages of tbe book byDaniel Katz and Robert Kabn, The Social Psy-chology of Organizations, Jobn Wiley and Co.,1966. It is not easy reading, however.

15