Jawn Lam

Socio-technical autonomous work groups

Organizational StructuresTeam-based and self-managing forms13 min read

Work groups own a whole task, set their own methods and allocations, and the social and technical systems are designed together.

What it is

A socio-technical design organizes production around semi-autonomous work groups, each owning a whole, bounded segment of the work: a coal-face cycle, a set of looms, a large part of a vehicle. The group sets its methods, divides its tasks, and controls its own work's variances. Management works at the group's boundary and does not supervise each job. The social and technical systems are designed together, so group boundaries follow the technology's natural breaks.

Socio-technical design gives each work group a whole segment and keeps management at its boundary

Select a unit or a letter to read it; the key above the drawing names the lines and letters used.

Reports toAdjust to each other directlyMMutual adjustmentNStandardization of norms

SpokespersonsManagementWork groupsManagement atthe boundaryWork groupWork groupWork groupEach owns a whole segmentMN
SpokespersonsManagementWork groupsManagersWorkgroupWorkgroupWorkgroupEach owns a whole segmentMN
Select a unit or a letter to read it. M, mutual adjustment: People coordinate their own work by informal communication with each other. N, standardization of norms: Shared beliefs and values, usually across the whole organization, guide the work.Management at the boundary. Layer: Management. Reports to no one. Works at the groups' boundary, regulating what crosses it, and does not supervise each job (Cummings 1978). At one manufacturing site, supervisory positions were eliminated when autonomous groups came in (Wall and colleagues 1986).Work group. Layer: Work groups. Reports to Management at the boundary. Each owns a whole segment. Owns a whole, bounded segment of the work, such as a coal-face cycle (Trist and Bamforth 1951), and sets its own methods and allocations.Work group. Layer: Work groups. Reports to Management at the boundary. Each owns a whole segment. Controls its work's variances as close to their source as possible (Cherns 1976); an example is a group of workers given a group of looms (Rice 1958).Work group. Layer: Work groups. Reports to Management at the boundary. Each owns a whole segment. For example, the groups that assembled whole vehicles or large parts in place of the paced line at Volvo's Uddevalla and Kalmar plants (Berggren 1992).M, mutual adjustment. People coordinate their own work by informal communication with each other.N, standardization of norms. Shared beliefs and values, usually across the whole organization, guide the work.
The structure in words
  1. Management
    • Management at the boundary, reports to no one. Works at the groups' boundary, regulating what crosses it, and does not supervise each job (Cummings 1978). At one manufacturing site, supervisory positions were eliminated when autonomous groups came in (Wall and colleagues 1986).
  2. Work groups
    • Work group, reports to Management at the boundary, grouped each owns a whole segment. Owns a whole, bounded segment of the work, such as a coal-face cycle (Trist and Bamforth 1951), and sets its own methods and allocations.
    • Work group, reports to Management at the boundary, grouped each owns a whole segment. Controls its work's variances as close to their source as possible (Cherns 1976); an example is a group of workers given a group of looms (Rice 1958).
    • Work group, reports to Management at the boundary, grouped each owns a whole segment. For example, the groups that assembled whole vehicles or large parts in place of the paced line at Volvo's Uddevalla and Kalmar plants (Berggren 1992).
  • Work group and Work group: adjust to each other directly (Spokespersons).
  • Each owns a whole segment: coordinated by mutual adjustment.
  • Each owns a whole segment: coordinated by standardization of norms.

Each group owns a whole, bounded segment of the work, its boundary drawn where the technology breaks, and sets its own methods and task allocations. Mutual adjustment and the group's own norms coordinate inside it, while management regulates what crosses its boundary and does not supervise each job.

Source: Cummings 1978; Cherns 1976; Trist and Bamforth 1951, an interview-based field study in one industry; Rice 1958; Berggren 1992, six case studies in the Swedish auto industry by one scholar; Wall and colleagues 1986, a long-term quasi-experiment in one company; Ingvaldsen and Rolfsen 2012, one case study.

Its profile, from Cummings (1978) and Cherns (1976) unless another work is named:

  • Grouping: product or purpose, our judgment: each group owns a whole segment.
  • Specialization: low, our judgment: members share the segment's tasks (with Rice 1958).
  • Formalization: low, theory: minimal critical specification.
  • Centralization: contested. Decentralized, theory: variances are controlled near their source. Mixed, on Kelly (1978), a critical reanalysis of several published case studies: the groups' autonomy is limited and subordinate to economic objectives.
  • Layers and span: flat, from evidence: Wall and colleagues (1986), a long-term quasi-experiment in one company, found supervisory positions eliminated.
  • Decision rights: contested, both our judgment: vertical and horizontal decentralization, Mintzberg's (1980) Type E held by groups; or selective vertical and horizontal decentralization, on Kelly.
  • Coordinating mechanism: spans mutual adjustment and standardization of norms, theory.
  • Lateral mechanisms: direct contact and liaison roles, from evidence: Ingvaldsen and Rolfsen (2012), one case study, explored rotating group spokespersons and shared leadership between groups.
  • Differentiation against integration: moderate, our judgment, with management joining the groups (Lawrence and Lorsch 1967).
  • Mode of exchange: spans bureaucracy and clan, our judgment: group norms inside an employment hierarchy (Ouchi 1980; Barker 1993, one ethnographic case study of self-managing teams).
  • Boundary permeability: selective, our judgment (Ashkenas and colleagues 1995).

Where socio-technical autonomous work groups sit on the eleven dimensions

The filled segments are the form's levels, and an outlined segment is the other view's level on a contested row; select a row for its reason and sources.

Select a dimension to read why the form sits there, and on what basis.Basis of grouping: Product or purpose. Basis: Judgment. Each group owns a whole, bounded segment of the work, with boundaries drawn by the technology's natural breaks; judgment for the basis. Sources: Cummings 1978; Cherns 1976.Specialization: Low. Basis: Judgment. Group members set their own allocations and share the tasks of a whole segment; judgment for the band. Sources: Cummings 1978; Rice 1958.Standardization and formalization: Low. Basis: Theory. Minimal critical specification: specify no more than is essential. Sources: Cherns 1976.Centralization: Decentralized. Basis: Theory. Variances are controlled as close to their source as possible, and the groups regulate themselves. Sources: Cherns 1976; Cummings 1978. View 2: Mixed. A review of the published cases found the autonomy granted to the groups limited and subordinate to economic objectives. Sources: Kelly 1978.Layers and span of control: Flat. Basis: Evidence. In one manufacturing site, productivity rose through the elimination of supervisory positions when autonomous groups were introduced. Sources: Wall and colleagues 1986.Decision rights and the locus of authority: Vertical and horizontal decentralization. Basis: Judgment. Power flows out to the operators of the operating core, the shape of Mintzberg's Type E, though held by groups rather than individual professionals; judgment. Sources: Mintzberg 1980; Cummings 1978; Kelly 1978. View 2: Selective vertical and horizontal decentralization. A review of the published cases found the autonomy granted to the groups limited and subordinate to economic objectives, so some decisions go to the groups and others stay with management, a reading that is judgment. Sources: Kelly 1978.Prime coordinating mechanism: Mutual adjustment and standardization of norms. Basis: Theory. Mutual adjustment within the group and group norms, with management at the boundary. Sources: Cummings 1978; Cherns 1976.Lateral mechanisms: Direct contact and liaison roles. Basis: Evidence. One study explored two designed structures for coordination between groups, rotating group spokespersons and shared leadership, as alternatives to hierarchical control, and found coordination between highly autonomous groups a major challenge, which may help explain why such groups are not widely used. Sources: Ingvaldsen and Rolfsen 2012.Differentiation against integration: Moderate differentiation, some integrating roles. Basis: Judgment. Groups own different segments and are joined at their boundaries by management; judgment. Sources: Cherns 1976; Lawrence and Lorsch 1967a.Mode of exchange and control: Bureaucracy and clan. Basis: Judgment. Group norms and peer pressure inside an employment hierarchy; judgment. Sources: Ouchi 1980; Barker 1993.Boundary permeability: Selective. Basis: Judgment. Management works at the group's boundary, regulating what crosses it; judgment for the band. Sources: Cummings 1978; Ashkenas and colleagues 1995.

Three levels rest on theory and two on evidence: flat layers, from supervisory posts eliminated at one site (Wall and colleagues 1986), and liaison between groups through rotating spokespersons (Ingvaldsen and Rolfsen 2012). Centralization and decision rights are contested, on Kelly's (1978) finding that the groups' autonomy was limited and subordinate to economic objectives.

Source: Cummings 1978; Cherns 1976; Rice 1958; Wall and colleagues 1986, a long-term quasi-experiment in one company; Mintzberg 1980, a synthesis of earlier research, offering propositions without a new test; Kelly 1978, a critical reanalysis of several published case studies; Ingvaldsen and Rolfsen 2012, one case study; Lawrence and Lorsch 1967a, a multi-firm comparative field study; Ouchi 1980, a conceptual article; Barker 1993, one ethnographic case study; Ashkenas and colleagues 1995.

Where it came from

Eric Trist and Ken Bamforth, of the Tavistock Institute, set out the problem in 1951, in an interview-based field study of British coal mining. Hand-got methods rested on small, self-selected groups taking the whole coal-getting task with responsible autonomy. Mechanized longwall working, three shifts of specialized roles, broke that pattern, with defensive reactions. They concluded that a social as well as a technological whole must exist, with responsible autonomy restored to primary groups that each have a satisfying sub-whole as their task.

The schools of organization design and the forms' origins, 1900 to now

Each row is a school's era; each square under them is a form at its first year. Select a row or a square to read it; on a form's page, its own square and span are marked, with the schools that treat it.

1900192019401960198020002020Bureaucracy and its criticsClassical administrationHuman relations and System 4Transaction cost economicsCarnegie schoolSocio-technical systemsLabor-managed firmModularity and mirroringStrategy and structureContingency theoryFrench bureaucratic schoolJapanese-firm literatureScandinavian and DutchConfiguration theoryInformation processingLabor processStrategic choiceGerman codeterminationOrganizational economicsInstitutional theoryPopulation ecologyResource dependenceBusiness groupsScandinavian institutionalSocietal effectNetwork perspectiveComputational designNew-forms debateAlgorithmic managementForms by first yearSocio-technical autonomous work groups, 1951 to 1965Now

The forms by first year

1900192019401960198020002020Socio-technical autonomous work groups, 1951 to 1965Now

Schools that treat this form: Socio-technical systems; the Scandinavian and Dutch socio-technical schools; and Labor process and critical theory.

Select a school's row or a form's square to read it.Select a form's square to read it; open a school below.Bureaucracy and its critics1900s to 1950s, revived from the 1990sWhat it claimsWeber held that rational-legal authority exercised through an office hierarchy, written rules, trained officials, and promotion by qualification is technically superior in precision, speed, and reliability. Critics studied its unintended consequences, Michels's iron law sharpest.What it has shownMerton (1940) argued, with no test, that rules can turn from means into ends; case studies followed. The Aston group concluded that the type's dimensions vary independently, and Wagner's meta-analysis (2021) found that many of their interrelationships have weakened.Who disputes itHuman relations and System 4 sets the costs of impersonal rules to motivation and cooperation against the technical superiority Weber stresses.Bureaucracy and its critics in The Schools of Organization DesignClassical administration1910s to 1940s, and in German to the 1970sWhat it claimsThere are general principles of administration, among them division of work, unity of command, a scalar chain of authority, and a limited span of control, and a structure built to them will be efficient in most settings. Taylor added that planners should set work methods.What it has shownThe school offered reasoning and experience, not tests. Woodward (1965) found that the spans and levels of the more successful firms in her sample differed by production system, and Simon (1946) showed that the principles come in pairs that point in opposite directions.Who disputes itStructural contingency theory found that the structure of successful firms varies with technology and environment, where classical writers treat their principles as general.Classical administration in The Schools of Organization DesignHuman relations and System 41920s to 1960sWhat it claimsOutput depends on the social relations of work, the norms of informal groups, and the style of supervision as much as on formal design. Likert's System 4, overlapping work groups joined by linking-pin managers under participative management, outperforms authoritative systems.What it has shownThe founding evidence is the Hawthorne studies, which later reanalyses dispute (Carey 1967; Levitt and List 2011). Likert's support came largely from his own institute's surveys. Wagner (1994) found participation's effects on performance and satisfaction significant but small.Who disputes itLabor process and critical theory reads human relations as a technique of managerial control rather than a change in the distribution of power.Human relations and System 4 in The Schools of Organization DesignTransaction cost economics and the M-form hypothesis1937, then the 1970s to the presentWhat it claimsFirms exist, and are organized as they are, to economize on transaction costs. Hierarchy replaces the market when assets are specific, uncertainty is high, and exchange is frequent, and the multidivisional form should outperform the functional form in large, diversified firms.What it has shownArmour and Teece (1978) found, in United States petroleum firms from 1955 to 1973, a positive relationship between the multidivisional form and profitability while the form was spreading, but not once an equilibrium in organizational form was reached.Who disputes itThe network perspective on organizations: Granovetter argued that the theory undersocializes economic action, which is embedded in networks of personal relations that also produce trust and order.Transaction cost economics and the M-form hypothesis in The Schools of Organization DesignThe Carnegie school and the attention-based view1940s to the presentWhat it claimsPeople in organizations are boundedly rational, so structure matters because it sets the premises of their decisions and channels their attention: who sees which problems, which goals they pursue, and which routines they follow. Ocasio makes structure distribute attention.What it has shownMarch and Simon (1958) drew propositions from the existing literature, and Cyert and March (1963) added process models and simulations. Two studies show hierarchy shaping what is sent up as well as what is chosen (Reitzig and Maciejovsky 2015; Keum and See 2017).Who disputes itOrganizational economics: decision rights, authority, and complementarities models a principal and agents with defined objectives; the Carnegie school, firms as coalitions with unresolved goal conflict.The Carnegie school and the attention-based view in The Schools of Organization DesignSocio-technical systems1950s to 1980sWhat it claimsAn organization is a social system and a technical system that must be designed together. The same technology admits more than one way of organizing work, and groups that own a whole task and control its variances at the source perform better and are more satisfied.What it has shownTrist and Bamforth (1951) found that longwall mechanization broke up self-regulating work groups. A review of 134 experiments found them extremely successful overall, and Stewart (2006) linked team autonomy to performance; studies of socialist brigades and Volkswagen qualify it.Who disputes itLabor process and critical theory: Kelly argued that labor intensification, not joint optimization, accounts for what socio-technical redesign achieved, and that the part of pay incentives was seriously underestimated.Socio-technical systems in The Schools of Organization DesignThe economics of the labor-managed firm and of enterprise ownership1958 to the presentWhat it claimsA firm run by its workers behaves differently from one run for investors: Ward modeled the Yugoslav-style firm as maximizing income per worker. Hansmann argued that an ownership form's success depends on the balance between the costs of market contracting and of ownership.What it has shownBurdín (2014) found a 29 percent lower hazard of dissolution for Uruguayan worker-managed firms than for conventional ones, and Fakhfakh, Pérotin, and Gago (2012) found French labor-managed firms as productive or more so in every industry studied.Who disputes itOrganizational economics (Jensen and Meckling) contrasts the labor-managed firm, in which tradable residual claims are legally prohibited, with cooperatives and partnerships that arise out of voluntary contracting.The economics of the labor-managed firm and of enterprise ownership in The Schools of Organization DesignModularity and the mirroring hypothesis1960s to the presentWhat it claimsComplex systems are built of nearly decomposable parts, and an organization's structure tends to mirror the architecture of what it makes: modular products allow loosely coupled organizations, and tightly coupled products need tightly coupled teams. Conway's law runs in reverse.What it has shownColfer and Baldwin (2016) review the empirical studies and report both support and principled exceptions. Direct tests point both ways: MacCormack, Baldwin, and Rusnak (2012) found open source products more modular than their commercial pairs; dynamic components break the mirror.Who disputes itNo school in the record disputes it head on; it answers information processing, which meets the coordination load inside the organization; mirroring writers locate it in the product's architecture.Modularity and the mirroring hypothesis in The Schools of Organization DesignStrategy and structure1960s to 1980sWhat it claimsStructure follows strategy: firms that diversify by product or geography outgrow the centralized functional structure and, after a lag and usually a crisis, adopt divisions. Chandler drew this from the histories of DuPont, General Motors, Standard Oil of New Jersey, and Sears.What it has shownChandler (1962) is comparative business history of four firms set against a wider survey. Rumelt (1974) related diversification and structure to performance, and Egelhoff (1988), with 34 multinationals, found some of Stopford and Wells's relationships supported and others not.Who disputes itInstitutional theory: in Fligstein's model divisions spread with a product-related or unrelated strategy, presidents from sales or finance, and other firms in the industry changing; strategy was one of three.Strategy and structure in The Schools of Organization DesignStructural contingency theory1960s to the presentWhat it claimsThere is no one best structure. Performance depends on fit between structure and contingencies such as environment, technology, size, and strategy, on an open-system view of the organization. Drazin and Van de Ven separate fit as selection, interaction, and systems.What it has shownBurns and Stalker, Woodward, and the Aston group related structure to context; Drazin and Van de Ven (1985) supported selection and systems fit only. Pennings (1975) found that fit did not explain effectiveness; Schoonhoven (1981) found traditional hypotheses unsupported.Who disputes itStrategic choice: Child argued that contingency theory explains structure by the constraints of context and leaves out the political process by which power-holders choose strategic action, structural forms included.Structural contingency theory in The Schools of Organization DesignThe French bureaucratic school (strategic analysis)1960s to 1980sWhat it claimsA bureaucracy is a system of power relations organized around zones of uncertainty. Impersonal rules and centralization protect members from depending on one another, those who control uncertainties hold power, and, unable to correct its errors, it changes only through crisis.What it has shownLe phénomène bureaucratique (1963; English 1964) rests on two case studies, a clerical agency and a state industrial monopoly. L'acteur et le système (1977; English 1980) is theoretical.Who disputes itSocietal effect and comparative business systems: D'Iribarne argues that national logics of conduct, such as a French logic of honor, explain differences that Crozier attributed to power games.The French bureaucratic school (strategic analysis) in The Schools of Organization DesignThe Japanese-firm literature1960s to the presentWhat it claimsThe large Japanese firm coordinates operating decisions through horizontal information sharing among semi-autonomous shops and uses its ranking hierarchy mainly as an incentive scheme (Aoki). It creates knowledge through middle-up-down management (Nonaka and Takeuchi).What it has shownAoki (1986) is a formal model. Lincoln, Hanada, and McBride (1986) compared 51 Japanese and 55 United States manufacturing plants, and Nonaka and Takeuchi (1995) rest on case studies, chiefly of Japanese firms. Fruin (1994) found the enterprise system defined at three levels.Who disputes itLabor process and critical theory reads shop-floor delegation as control: Sewell and Wilkinson argue that just-in-time and total quality regimes delegate tactical responsibility while centralizing strategic control.The Japanese-firm literature in The Schools of Organization DesignThe Scandinavian and Dutch socio-technical schools1960s to 2000sWhat it claimsThe Scandinavian program sought democracy at work through field experiments with semi-autonomous groups, then through democratic dialogue. The Dutch school's design theory turns complex organizations with simple jobs into simple organizations with complex jobs.What it has shownEmery and Thorsrud reported field experiments in Norwegian industry; Gustavsen (1992) records that they spread less than hoped, his reason for turning to dialogue. Berggren studied work organization in the Swedish auto industry. The Dutch work rests on design cases and argument.Who disputes itThe Japanese-firm literature: defenders of the Toyota system hold that standardized work with employee participation learns faster; Berggren reports Swedish long-cycle plants matching a line plant on productivity.The Scandinavian and Dutch socio-technical schools in The Schools of Organization DesignConfiguration theory1970s to the presentWhat it claimsStructures cluster into a small number of internally coherent configurations whose elements reinforce one another, so organizations move between a few types in large jumps, not small steps. Mintzberg's five configurations each rest on one coordinating mechanism.What it has shownDoty, Glick, and Huber (1993) found that fit to Miles and Snow's ideal types predicted effectiveness while fit to Mintzberg's did not. A meta-analysis of 40 tests (Ketchen and colleagues 1997) found that configuration partly explains performance, more so with broad definitions.Who disputes itStructural contingency theory: Donaldson counters that most organizations sit at intermediate points on a continuum of fits, not in a few configurations, and that studies over time find mostly incremental change.Configuration theory in The Schools of Organization DesignInformation processing1970s to the presentWhat it claimsThe more uncertain the task, the more information must be processed during its execution. An organization either reduces that need, through slack resources or self-contained tasks, or raises its capacity, through vertical information systems or lateral relations up to the matrix.What it has shownGalbraith (1973, 1974) built the framework from cases and from Simon's theory of bounded rationality, and Tushman and Nadler (1978) set it out as an integrating concept for design. Daft and Lengel (1986) argued that structure must reduce equivocality as well as supply data.Who disputes itThe Carnegie school and the attention-based view: Ocasio holds that firm behavior results from how the firm channels and distributes attention, so structure governs where attention goes, not only how much is processed.Information processing in The Schools of Organization DesignLabor process and critical theory1970s to the presentWhat it claimsStructure is above all a system of control over labor. Braverman argued that management deskills work by separating conception from execution; Edwards traced a sequence from simple control by the boss to technical control through machines to bureaucratic control through rules.What it has shownHistorical and theoretical (Braverman 1974; Edwards 1979). Delbridge (1995), observing a Japanese-owned plant in England, found that just-in-time and total quality regimes combined the control of labor with management's economic goals more completely than piece rates had.Who disputes itBureaucracy and its critics: Adler argues that labor process theory ignores the progressive socialization of the labor process, which drives long-term skill upgrading, and disputes deskilling as the trend.Labor process and critical theory in The Schools of Organization DesignStrategic choice1970s to the presentWhat it claimsStructure is the outcome of choices by those with power in the organization, who decide on strategic action, including the structural form, the manipulation of environmental features, and the performance standards to be met. Context constrains but does not determine structure.What it has shownLargely conceptual: Child (1972) built the argument from gaps in contingency findings, and the empirical case rests mostly on the modest share of structural variance contingencies explain. A 2026 review of 42 studies finds upper echelons theory their main theoretical anchor.Who disputes itStructural contingency theory: Donaldson replies that organizations mostly adopt the structures their contingencies dictate, to avoid the loss of performance misfit brings, so choice over structure is at most limited.Strategic choice in The Schools of Organization DesignThe German codetermination literature1970s to the presentWhat it claimsWorks councils and employee board seats form a structural layer through which coordination can become negotiation between management and employees. Defenders argue that it supports long-term employment, training, and cooperation; critics, that it weakens shareholder control.What it has shownJäger, Schoefer, and Heining (2021) found board-level codetermination left wages unchanged and, if anything, raised capital formation. Jirjahn (2011) reports studies pointing to positive works council effects; a meta-analysis found codetermination laws tied to lower productivity.Who disputes itOrganizational economics (Gorton and Schmid) found that firms with equal employee and shareholder representation did worse for their shareholders than firms with one-third employee representation.The German codetermination literature in The Schools of Organization DesignOrganizational economics: decision rights, authority, and complementarities1976 to the presentWhat it claimsDesign is the joint choice of who holds which decision rights, how performance is measured, and how it is rewarded. Rights should be placed where the specific knowledge lies, formal authority differs from real authority, and practices are complements that pay off in clusters.What it has shownIchniowski, Shaw, and Prennushi (1997) found, on 36 steel finishing lines, that lines using a set of innovative work practices achieved substantially higher productivity, consistent with complementarities. Bloom and colleagues found more decentralization where trust was higher.Who disputes itGhoshal argued that management research has shaped managers' worldview with ideologically inspired amoral theories, built on a scientific model that excludes human intentionality and so denies moral considerations.Organizational economics: decision rights, authority, and complementarities in The Schools of Organization DesignInstitutional theory1977 to the presentWhat it claimsFormal structures are often adopted because they are legitimate in the organization's field, as rationalized myths, and then decoupled from the work done. Organizations in a field grow alike through coercive, mimetic, and normative isomorphism, whether or not it is efficient.What it has shownTolbert and Zucker (1983) found that city characteristics predicted early but not later adoption of civil service reform, which they read as institutionalization. Palmer, Jennings, and Zhou (1993) found support for institutional and economic accounts of late M-form adoption.Who disputes itStructural contingency theory: Kraatz and Zajac found little support for neoinstitutional predictions in 631 liberal arts colleges, which grew less alike; the change they studied was curricular, not structural.Institutional theory in The Schools of Organization DesignPopulation ecology1977 to the presentWhat it claimsOrganizations face strong structural inertia, because selection favors the reliability and accountability of stable structures. Change in a population's forms comes mainly from founding and failure, not adaptation, and changing core features resets the liability of newness.What it has shownAmburgey, Kelly, and Barnett (1993) found that the effects of change on the hazard of failure vary by type of change and over time, and that rates of change depend on how often and how recently an organization has changed. Carroll and Hannan (2000) set out the program's methods.Who disputes itFor structural contingency theory, Donaldson argues that ecology understates managers' capacity to adapt structure.Population ecology in The Schools of Organization DesignResource dependence1978 to the presentWhat it claimsOrganizations depend on outside parties for critical resources, and structure follows the management of that dependence: units that cope with critical dependencies gain power, and boundaries are redrawn through cooptation, joint ventures, and mergers to absorb constraint.What it has shownPfeffer and Salancik (1978) assembled studies of boards, mergers, and joint ventures. A meta-analysis of 157 tests (Drees and Heugens 2013) corroborated the main prediction: organizations meet resource dependencies with interorganizational arrangements that raise their autonomy.Who disputes itTransaction cost economics and the M-form hypothesis explains boundaries by efficiency; resource dependence, by power and constraint.Resource dependence in The Schools of Organization DesignThe business-groups literature1978 to the presentWhat it claimsGroups of legally independent firms bound by ownership and social ties are a durable form of their own, neither one firm nor a market. Where institutions function poorly, groups can benefit member firms and can destroy value; they are sometimes paragons, sometimes parasites.What it has shownKhanna and Palepu (2000) found that affiliates of the most diversified Indian groups outperformed unaffiliated firms. A meta-analysis of 141 studies (Carney and colleagues 2011) found that affiliation lowers performance in general, less so where institutions are underdeveloped.Who disputes itOrganizational economics: agency writers read pyramids and cross-holdings as devices that separate control from ownership; Almeida and Wolfenzon offer a financing rationale for pyramids instead.The business-groups literature in The Schools of Organization DesignScandinavian institutionalism1980s to the presentWhat it claimsIdeas of organizing do not diffuse unchanged; they travel and are translated as they are adopted, so a form taken up in one place is not the same form elsewhere. Organizations can also be built of organizations (meta-organizations), which behave differently.What it has shownA discussion paper by Czarniawska and Joerges (1995) contrasts the diffusion of ideas with their translation, drawing on examples such as Swedish municipal reform; the work on meta-organizations (Ahrne and Brunsson 2008) is conceptual.Who disputes itInstitutional theory can read the same processes, seen from a translocal view, as institutional isomorphism, though translation holds that each adopter modifies an idea, so an institution re-created elsewhere is new.Scandinavian institutionalism in The Schools of Organization DesignSocietal effect and comparative business systems1980s to the presentWhat it claimsFirms matched on size, technology, and industry differ systematically by country, because how a society educates, trains, recruits, and promotes its workforce, and its industrial relations, shape how work is divided and supervised, so structure is partly a national product.What it has shownMaurice, Sorge, and Warner (1980) compared closely matched manufacturing units in France, West Germany, and Great Britain and found more pervasive differences in their organization than usually recognized. Whitley (1999) set out a comparative account of business systems.Who disputes itStructural contingency theory stresses culture-free elements of structure; matched comparisons find nationally different shapes of organization.Societal effect and comparative business systems in The Schools of Organization DesignThe network perspective on organizations1985 to the presentWhat it claimsOrganizations and their members are nodes joined by ties, and the pattern of ties shapes what they can do. Economic action is embedded in networks of personal relations (Granovetter 1985), and network forms are a distinct form, not a hybrid of market and hierarchy (Powell 1990).What it has shownPodolny and Page (1998) review the research and find that network forms foster learning, confer status and legitimacy, bring economic benefits, help manage resource dependencies, and give employees considerable autonomy, and they ask for attention to networks' dysfunctions.Who disputes itTransaction cost economics: Williamson treats the hybrid as a third generic form between market and hierarchy; for Powell, network forms are a distinct form, not a mix of the two.The network perspective on organizations in The Schools of Organization DesignComputational organization design and search1990s to the presentWhat it claimsDesign choices interact, so design space is a rugged landscape with many local peaks, and structure affects how well an organization searches it. Decentralizing interdependent decisions can trap units on poor peaks; temporary decentralization, then reintegration, can help.What it has shownMostly simulation, with one field test: Csaszar (2012), with more than 150,000 stock-picking decisions by 609 mutual funds, found that decision-making structure affects the number of initiatives pursued and the omission and commission errors made.Who disputes itNo school in the record disputes it head on; it disputes structural contingency theory, whose bivariate fit, its simulations show, cannot capture how the best design depends on interactions among choices.Computational organization design and search in The Schools of Organization DesignThe new-forms debate1990s to the presentWhat it claimsOrganizations must solve four universal problems (dividing tasks, allocating them, providing rewards, and providing information), and a form is new when it solves them in new ways. Self-managing organizations radically decentralize authority; skeptics hold hierarchy reappears.What it has shownFoss (2003) traced the rise and partial reversal of Oticon's spaghetti organization to the problem of selective intervention. A meta-analysis of 54 studies of 13,914 teams (Greer and colleagues 2018) found that hierarchy within teams on net slightly lowers team performance.Who disputes itBureaucracy and its critics: removing formal structure leaves informal elites whose power is unaccountable (Freeman), and democratic and network organizations are much less hierarchy-free than their labels suggest.The new-forms debate in The Schools of Organization DesignThe algorithmic-management literature2015 to the presentWhat it claimsSoftware does much of the directing, evaluating, and disciplining that managers and rules once did. Kellogg, Valentine, and Christin (2020) call algorithmic control a new contested terrain; Puranam (2021) treats dividing decisions between people and machines as a design problem.What it has shownMostly qualitative field studies of gig work. Wood and colleagues (2019), with 107 interviews in six countries and 679 survey respondents, found that algorithmic management tends to give remote gig workers flexibility and autonomy, but also low pay, isolation, and overwork.Who disputes itLabor process and critical theory reads control at work as contested terrain; Kellogg, Valentine, and Christin take up that view but argue that algorithmic control is distinct from earlier forms.The algorithmic-management literature in The Schools of Organization DesignFunctional structureOrigin: 1911 to 1937.Family: Classical hierarchies.Line and staffOrigin: 1916 to 1943.Family: Classical hierarchies.Multidivisional (M-form) by productOrigin: 1920s practice; named and analyzed 1962 to 1975.Family: Divisionalized forms.Machine bureaucracyOrigin: 1922 (Weber); 1979 (Mintzberg).Family: Classical hierarchies.Divisional by customer or market segmentOrigin: 1937; 2005.Family: Divisionalized forms.Matrix organizationOrigin: 1950s to 1960s practice; analyzed 1971 to 1977.Family: Lateral and multidimensional forms.Co-determined organizationOrigin: 1951 to 1976 (the laws); 1984 to 2022 (the main studies).Family: Forms from other traditions.Socio-technical autonomous work groupsOrigin: 1951 to 1965.Family: Team-based and self-managing forms.Stratified hierarchy (requisite organization)Origin: 1951 to 1989.Family: Classical hierarchies.Cooperative organizationOrigin: 1958 to 1999.Family: Forms from other traditions.Linking-pin organization (System 4)Origin: 1961 to 1967.Family: Team-based and self-managing forms.Divisional by regionOrigin: 1962 to 1972.Family: Divisionalized forms.Adhocracy (innovative organization)Origin: 1968 to 1979.Family: Professional, project, and knowledge forms.J-form (the Japanese firm)Origin: 1968 to 1990.Family: Forms from other traditions.Professional bureaucracyOrigin: 1968 to 1979.Family: Professional, project, and knowledge forms.Holding company (H-form)Origin: 1972 to 1975; 1987.Family: Divisionalized forms.Ambidextrous organizationOrigin: 1976 to 2004.Family: Professional, project, and knowledge forms.Simple structureOrigin: 1979.Family: Classical hierarchies.Hybrid structureOrigin: 1980 to the present.Family: Lateral and multidimensional forms.Missionary (ideological) organizationOrigin: 1980 to 1989.Family: Ideological forms.Network, modular, and virtual organizationOrigin: 1986 to 1990.Family: Network and market-like forms.Transnational and multidimensional organizationOrigin: 1987 to 1989.Family: Lateral and multidimensional forms.Business group (keiretsu, chaebol, family house)Origin: 1988 to 2007.Family: Forms from other traditions.Shamrock organizationOrigin: 1988 to 1989.Family: Network and market-like forms.Federal organizationOrigin: 1989 to 1992.Family: Network and market-like forms.Self-managing organization (less-hierarchical forms)Origin: 1989 to 2017.Family: Team-based and self-managing forms.Boundaryless organizationOrigin: 1990 to 1995.Family: Network and market-like forms.Front-back organizationOrigin: 1990s to 2007.Family: Lateral and multidimensional forms.Professional partnership (P2 form)Origin: 1990.Family: Professional, project, and knowledge forms.The corporate center and shared servicesOrigin: 1991 to 2000.Family: Divisionalized forms.Internal market organizationOrigin: 1993 to 1994.Family: Network and market-like forms.Process (horizontal) organizationOrigin: 1993 to 1999.Family: Lateral and multidimensional forms.Cellular organizationOrigin: 1994 to 1997.Family: Network and market-like forms.Hypertext organizationOrigin: 1994 to 1995.Family: Forms from other traditions.Team-based organizationOrigin: 1995.Family: Team-based and self-managing forms.Project-based organizationOrigin: 2000 to 2006.Family: Professional, project, and knowledge forms.Agile at scaleOrigin: 2001 to 2016.Family: Contemporary and emerging forms.Distributed (all-remote) organizationOrigin: 2002 to 2021.Family: Contemporary and emerging forms.Open-source and community-based organizationOrigin: 2002 to 2007.Family: Contemporary and emerging forms.Algorithmically managed organizationOrigin: 2015 to 2020.Family: Contemporary and emerging forms.Team of teamsOrigin: 2015.Family: Team-based and self-managing forms.

The schools by first year

  1. Bureaucracy and its critics1900s to 1950s, revived from the 1990sWhat it claimsWeber held that rational-legal authority exercised through an office hierarchy, written rules, trained officials, and promotion by qualification is technically superior in precision, speed, and reliability. Critics studied its unintended consequences, Michels's iron law sharpest.What it has shownMerton (1940) argued, with no test, that rules can turn from means into ends; case studies followed. The Aston group concluded that the type's dimensions vary independently, and Wagner's meta-analysis (2021) found that many of their interrelationships have weakened.Who disputes itHuman relations and System 4 sets the costs of impersonal rules to motivation and cooperation against the technical superiority Weber stresses.
  2. Classical administration1910s to 1940s, and in German to the 1970sWhat it claimsThere are general principles of administration, among them division of work, unity of command, a scalar chain of authority, and a limited span of control, and a structure built to them will be efficient in most settings. Taylor added that planners should set work methods.What it has shownThe school offered reasoning and experience, not tests. Woodward (1965) found that the spans and levels of the more successful firms in her sample differed by production system, and Simon (1946) showed that the principles come in pairs that point in opposite directions.Who disputes itStructural contingency theory found that the structure of successful firms varies with technology and environment, where classical writers treat their principles as general.
  3. Human relations and System 41920s to 1960sWhat it claimsOutput depends on the social relations of work, the norms of informal groups, and the style of supervision as much as on formal design. Likert's System 4, overlapping work groups joined by linking-pin managers under participative management, outperforms authoritative systems.What it has shownThe founding evidence is the Hawthorne studies, which later reanalyses dispute (Carey 1967; Levitt and List 2011). Likert's support came largely from his own institute's surveys. Wagner (1994) found participation's effects on performance and satisfaction significant but small.Who disputes itLabor process and critical theory reads human relations as a technique of managerial control rather than a change in the distribution of power.
  4. Transaction cost economics and the M-form hypothesis1937, then the 1970s to the presentWhat it claimsFirms exist, and are organized as they are, to economize on transaction costs. Hierarchy replaces the market when assets are specific, uncertainty is high, and exchange is frequent, and the multidivisional form should outperform the functional form in large, diversified firms.What it has shownArmour and Teece (1978) found, in United States petroleum firms from 1955 to 1973, a positive relationship between the multidivisional form and profitability while the form was spreading, but not once an equilibrium in organizational form was reached.Who disputes itThe network perspective on organizations: Granovetter argued that the theory undersocializes economic action, which is embedded in networks of personal relations that also produce trust and order.
  5. The Carnegie school and the attention-based view1940s to the presentWhat it claimsPeople in organizations are boundedly rational, so structure matters because it sets the premises of their decisions and channels their attention: who sees which problems, which goals they pursue, and which routines they follow. Ocasio makes structure distribute attention.What it has shownMarch and Simon (1958) drew propositions from the existing literature, and Cyert and March (1963) added process models and simulations. Two studies show hierarchy shaping what is sent up as well as what is chosen (Reitzig and Maciejovsky 2015; Keum and See 2017).Who disputes itOrganizational economics: decision rights, authority, and complementarities models a principal and agents with defined objectives; the Carnegie school, firms as coalitions with unresolved goal conflict.
  6. Socio-technical systems1950s to 1980sWhat it claimsAn organization is a social system and a technical system that must be designed together. The same technology admits more than one way of organizing work, and groups that own a whole task and control its variances at the source perform better and are more satisfied.What it has shownTrist and Bamforth (1951) found that longwall mechanization broke up self-regulating work groups. A review of 134 experiments found them extremely successful overall, and Stewart (2006) linked team autonomy to performance; studies of socialist brigades and Volkswagen qualify it.Who disputes itLabor process and critical theory: Kelly argued that labor intensification, not joint optimization, accounts for what socio-technical redesign achieved, and that the part of pay incentives was seriously underestimated.
  7. The economics of the labor-managed firm and of enterprise ownership1958 to the presentWhat it claimsA firm run by its workers behaves differently from one run for investors: Ward modeled the Yugoslav-style firm as maximizing income per worker. Hansmann argued that an ownership form's success depends on the balance between the costs of market contracting and of ownership.What it has shownBurdín (2014) found a 29 percent lower hazard of dissolution for Uruguayan worker-managed firms than for conventional ones, and Fakhfakh, Pérotin, and Gago (2012) found French labor-managed firms as productive or more so in every industry studied.Who disputes itOrganizational economics (Jensen and Meckling) contrasts the labor-managed firm, in which tradable residual claims are legally prohibited, with cooperatives and partnerships that arise out of voluntary contracting.
  8. Modularity and the mirroring hypothesis1960s to the presentWhat it claimsComplex systems are built of nearly decomposable parts, and an organization's structure tends to mirror the architecture of what it makes: modular products allow loosely coupled organizations, and tightly coupled products need tightly coupled teams. Conway's law runs in reverse.What it has shownColfer and Baldwin (2016) review the empirical studies and report both support and principled exceptions. Direct tests point both ways: MacCormack, Baldwin, and Rusnak (2012) found open source products more modular than their commercial pairs; dynamic components break the mirror.Who disputes itNo school in the record disputes it head on; it answers information processing, which meets the coordination load inside the organization; mirroring writers locate it in the product's architecture.
  9. Strategy and structure1960s to 1980sWhat it claimsStructure follows strategy: firms that diversify by product or geography outgrow the centralized functional structure and, after a lag and usually a crisis, adopt divisions. Chandler drew this from the histories of DuPont, General Motors, Standard Oil of New Jersey, and Sears.What it has shownChandler (1962) is comparative business history of four firms set against a wider survey. Rumelt (1974) related diversification and structure to performance, and Egelhoff (1988), with 34 multinationals, found some of Stopford and Wells's relationships supported and others not.Who disputes itInstitutional theory: in Fligstein's model divisions spread with a product-related or unrelated strategy, presidents from sales or finance, and other firms in the industry changing; strategy was one of three.
  10. Structural contingency theory1960s to the presentWhat it claimsThere is no one best structure. Performance depends on fit between structure and contingencies such as environment, technology, size, and strategy, on an open-system view of the organization. Drazin and Van de Ven separate fit as selection, interaction, and systems.What it has shownBurns and Stalker, Woodward, and the Aston group related structure to context; Drazin and Van de Ven (1985) supported selection and systems fit only. Pennings (1975) found that fit did not explain effectiveness; Schoonhoven (1981) found traditional hypotheses unsupported.Who disputes itStrategic choice: Child argued that contingency theory explains structure by the constraints of context and leaves out the political process by which power-holders choose strategic action, structural forms included.
  11. The French bureaucratic school (strategic analysis)1960s to 1980sWhat it claimsA bureaucracy is a system of power relations organized around zones of uncertainty. Impersonal rules and centralization protect members from depending on one another, those who control uncertainties hold power, and, unable to correct its errors, it changes only through crisis.What it has shownLe phénomène bureaucratique (1963; English 1964) rests on two case studies, a clerical agency and a state industrial monopoly. L'acteur et le système (1977; English 1980) is theoretical.Who disputes itSocietal effect and comparative business systems: D'Iribarne argues that national logics of conduct, such as a French logic of honor, explain differences that Crozier attributed to power games.
  12. The Japanese-firm literature1960s to the presentWhat it claimsThe large Japanese firm coordinates operating decisions through horizontal information sharing among semi-autonomous shops and uses its ranking hierarchy mainly as an incentive scheme (Aoki). It creates knowledge through middle-up-down management (Nonaka and Takeuchi).What it has shownAoki (1986) is a formal model. Lincoln, Hanada, and McBride (1986) compared 51 Japanese and 55 United States manufacturing plants, and Nonaka and Takeuchi (1995) rest on case studies, chiefly of Japanese firms. Fruin (1994) found the enterprise system defined at three levels.Who disputes itLabor process and critical theory reads shop-floor delegation as control: Sewell and Wilkinson argue that just-in-time and total quality regimes delegate tactical responsibility while centralizing strategic control.
  13. The Scandinavian and Dutch socio-technical schools1960s to 2000sWhat it claimsThe Scandinavian program sought democracy at work through field experiments with semi-autonomous groups, then through democratic dialogue. The Dutch school's design theory turns complex organizations with simple jobs into simple organizations with complex jobs.What it has shownEmery and Thorsrud reported field experiments in Norwegian industry; Gustavsen (1992) records that they spread less than hoped, his reason for turning to dialogue. Berggren studied work organization in the Swedish auto industry. The Dutch work rests on design cases and argument.Who disputes itThe Japanese-firm literature: defenders of the Toyota system hold that standardized work with employee participation learns faster; Berggren reports Swedish long-cycle plants matching a line plant on productivity.
  14. Configuration theory1970s to the presentWhat it claimsStructures cluster into a small number of internally coherent configurations whose elements reinforce one another, so organizations move between a few types in large jumps, not small steps. Mintzberg's five configurations each rest on one coordinating mechanism.What it has shownDoty, Glick, and Huber (1993) found that fit to Miles and Snow's ideal types predicted effectiveness while fit to Mintzberg's did not. A meta-analysis of 40 tests (Ketchen and colleagues 1997) found that configuration partly explains performance, more so with broad definitions.Who disputes itStructural contingency theory: Donaldson counters that most organizations sit at intermediate points on a continuum of fits, not in a few configurations, and that studies over time find mostly incremental change.
  15. Information processing1970s to the presentWhat it claimsThe more uncertain the task, the more information must be processed during its execution. An organization either reduces that need, through slack resources or self-contained tasks, or raises its capacity, through vertical information systems or lateral relations up to the matrix.What it has shownGalbraith (1973, 1974) built the framework from cases and from Simon's theory of bounded rationality, and Tushman and Nadler (1978) set it out as an integrating concept for design. Daft and Lengel (1986) argued that structure must reduce equivocality as well as supply data.Who disputes itThe Carnegie school and the attention-based view: Ocasio holds that firm behavior results from how the firm channels and distributes attention, so structure governs where attention goes, not only how much is processed.
  16. Labor process and critical theory1970s to the presentWhat it claimsStructure is above all a system of control over labor. Braverman argued that management deskills work by separating conception from execution; Edwards traced a sequence from simple control by the boss to technical control through machines to bureaucratic control through rules.What it has shownHistorical and theoretical (Braverman 1974; Edwards 1979). Delbridge (1995), observing a Japanese-owned plant in England, found that just-in-time and total quality regimes combined the control of labor with management's economic goals more completely than piece rates had.Who disputes itBureaucracy and its critics: Adler argues that labor process theory ignores the progressive socialization of the labor process, which drives long-term skill upgrading, and disputes deskilling as the trend.
  17. Strategic choice1970s to the presentWhat it claimsStructure is the outcome of choices by those with power in the organization, who decide on strategic action, including the structural form, the manipulation of environmental features, and the performance standards to be met. Context constrains but does not determine structure.What it has shownLargely conceptual: Child (1972) built the argument from gaps in contingency findings, and the empirical case rests mostly on the modest share of structural variance contingencies explain. A 2026 review of 42 studies finds upper echelons theory their main theoretical anchor.Who disputes itStructural contingency theory: Donaldson replies that organizations mostly adopt the structures their contingencies dictate, to avoid the loss of performance misfit brings, so choice over structure is at most limited.
  18. The German codetermination literature1970s to the presentWhat it claimsWorks councils and employee board seats form a structural layer through which coordination can become negotiation between management and employees. Defenders argue that it supports long-term employment, training, and cooperation; critics, that it weakens shareholder control.What it has shownJäger, Schoefer, and Heining (2021) found board-level codetermination left wages unchanged and, if anything, raised capital formation. Jirjahn (2011) reports studies pointing to positive works council effects; a meta-analysis found codetermination laws tied to lower productivity.Who disputes itOrganizational economics (Gorton and Schmid) found that firms with equal employee and shareholder representation did worse for their shareholders than firms with one-third employee representation.
  19. Organizational economics: decision rights, authority, and complementarities1976 to the presentWhat it claimsDesign is the joint choice of who holds which decision rights, how performance is measured, and how it is rewarded. Rights should be placed where the specific knowledge lies, formal authority differs from real authority, and practices are complements that pay off in clusters.What it has shownIchniowski, Shaw, and Prennushi (1997) found, on 36 steel finishing lines, that lines using a set of innovative work practices achieved substantially higher productivity, consistent with complementarities. Bloom and colleagues found more decentralization where trust was higher.Who disputes itGhoshal argued that management research has shaped managers' worldview with ideologically inspired amoral theories, built on a scientific model that excludes human intentionality and so denies moral considerations.
  20. Institutional theory1977 to the presentWhat it claimsFormal structures are often adopted because they are legitimate in the organization's field, as rationalized myths, and then decoupled from the work done. Organizations in a field grow alike through coercive, mimetic, and normative isomorphism, whether or not it is efficient.What it has shownTolbert and Zucker (1983) found that city characteristics predicted early but not later adoption of civil service reform, which they read as institutionalization. Palmer, Jennings, and Zhou (1993) found support for institutional and economic accounts of late M-form adoption.Who disputes itStructural contingency theory: Kraatz and Zajac found little support for neoinstitutional predictions in 631 liberal arts colleges, which grew less alike; the change they studied was curricular, not structural.
  21. Population ecology1977 to the presentWhat it claimsOrganizations face strong structural inertia, because selection favors the reliability and accountability of stable structures. Change in a population's forms comes mainly from founding and failure, not adaptation, and changing core features resets the liability of newness.What it has shownAmburgey, Kelly, and Barnett (1993) found that the effects of change on the hazard of failure vary by type of change and over time, and that rates of change depend on how often and how recently an organization has changed. Carroll and Hannan (2000) set out the program's methods.Who disputes itFor structural contingency theory, Donaldson argues that ecology understates managers' capacity to adapt structure.
  22. Resource dependence1978 to the presentWhat it claimsOrganizations depend on outside parties for critical resources, and structure follows the management of that dependence: units that cope with critical dependencies gain power, and boundaries are redrawn through cooptation, joint ventures, and mergers to absorb constraint.What it has shownPfeffer and Salancik (1978) assembled studies of boards, mergers, and joint ventures. A meta-analysis of 157 tests (Drees and Heugens 2013) corroborated the main prediction: organizations meet resource dependencies with interorganizational arrangements that raise their autonomy.Who disputes itTransaction cost economics and the M-form hypothesis explains boundaries by efficiency; resource dependence, by power and constraint.
  23. The business-groups literature1978 to the presentWhat it claimsGroups of legally independent firms bound by ownership and social ties are a durable form of their own, neither one firm nor a market. Where institutions function poorly, groups can benefit member firms and can destroy value; they are sometimes paragons, sometimes parasites.What it has shownKhanna and Palepu (2000) found that affiliates of the most diversified Indian groups outperformed unaffiliated firms. A meta-analysis of 141 studies (Carney and colleagues 2011) found that affiliation lowers performance in general, less so where institutions are underdeveloped.Who disputes itOrganizational economics: agency writers read pyramids and cross-holdings as devices that separate control from ownership; Almeida and Wolfenzon offer a financing rationale for pyramids instead.
  24. Scandinavian institutionalism1980s to the presentWhat it claimsIdeas of organizing do not diffuse unchanged; they travel and are translated as they are adopted, so a form taken up in one place is not the same form elsewhere. Organizations can also be built of organizations (meta-organizations), which behave differently.What it has shownA discussion paper by Czarniawska and Joerges (1995) contrasts the diffusion of ideas with their translation, drawing on examples such as Swedish municipal reform; the work on meta-organizations (Ahrne and Brunsson 2008) is conceptual.Who disputes itInstitutional theory can read the same processes, seen from a translocal view, as institutional isomorphism, though translation holds that each adopter modifies an idea, so an institution re-created elsewhere is new.
  25. Societal effect and comparative business systems1980s to the presentWhat it claimsFirms matched on size, technology, and industry differ systematically by country, because how a society educates, trains, recruits, and promotes its workforce, and its industrial relations, shape how work is divided and supervised, so structure is partly a national product.What it has shownMaurice, Sorge, and Warner (1980) compared closely matched manufacturing units in France, West Germany, and Great Britain and found more pervasive differences in their organization than usually recognized. Whitley (1999) set out a comparative account of business systems.Who disputes itStructural contingency theory stresses culture-free elements of structure; matched comparisons find nationally different shapes of organization.
  26. The network perspective on organizations1985 to the presentWhat it claimsOrganizations and their members are nodes joined by ties, and the pattern of ties shapes what they can do. Economic action is embedded in networks of personal relations (Granovetter 1985), and network forms are a distinct form, not a hybrid of market and hierarchy (Powell 1990).What it has shownPodolny and Page (1998) review the research and find that network forms foster learning, confer status and legitimacy, bring economic benefits, help manage resource dependencies, and give employees considerable autonomy, and they ask for attention to networks' dysfunctions.Who disputes itTransaction cost economics: Williamson treats the hybrid as a third generic form between market and hierarchy; for Powell, network forms are a distinct form, not a mix of the two.
  27. Computational organization design and search1990s to the presentWhat it claimsDesign choices interact, so design space is a rugged landscape with many local peaks, and structure affects how well an organization searches it. Decentralizing interdependent decisions can trap units on poor peaks; temporary decentralization, then reintegration, can help.What it has shownMostly simulation, with one field test: Csaszar (2012), with more than 150,000 stock-picking decisions by 609 mutual funds, found that decision-making structure affects the number of initiatives pursued and the omission and commission errors made.Who disputes itNo school in the record disputes it head on; it disputes structural contingency theory, whose bivariate fit, its simulations show, cannot capture how the best design depends on interactions among choices.
  28. The new-forms debate1990s to the presentWhat it claimsOrganizations must solve four universal problems (dividing tasks, allocating them, providing rewards, and providing information), and a form is new when it solves them in new ways. Self-managing organizations radically decentralize authority; skeptics hold hierarchy reappears.What it has shownFoss (2003) traced the rise and partial reversal of Oticon's spaghetti organization to the problem of selective intervention. A meta-analysis of 54 studies of 13,914 teams (Greer and colleagues 2018) found that hierarchy within teams on net slightly lowers team performance.Who disputes itBureaucracy and its critics: removing formal structure leaves informal elites whose power is unaccountable (Freeman), and democratic and network organizations are much less hierarchy-free than their labels suggest.
  29. The algorithmic-management literature2015 to the presentWhat it claimsSoftware does much of the directing, evaluating, and disciplining that managers and rules once did. Kellogg, Valentine, and Christin (2020) call algorithmic control a new contested terrain; Puranam (2021) treats dividing decisions between people and machines as a design problem.What it has shownMostly qualitative field studies of gig work. Wood and colleagues (2019), with 107 interviews in six countries and 679 survey respondents, found that algorithmic management tends to give remote gig workers flexibility and autonomy, but also low pay, isolation, and overwork.Who disputes itLabor process and critical theory reads control at work as contested terrain; Kellogg, Valentine, and Christin take up that view but argue that algorithmic control is distinct from earlier forms.

The forms by first year

  1. Functional structure: 1911 to 1937
  2. Line and staff: 1916 to 1943
  3. Multidivisional (M-form) by product: 1920s practice; named and analyzed 1962 to 1975
  4. Machine bureaucracy: 1922 (Weber); 1979 (Mintzberg)
  5. Divisional by customer or market segment: 1937; 2005
  6. Matrix organization: 1950s to 1960s practice; analyzed 1971 to 1977
  7. Co-determined organization: 1951 to 1976 (the laws); 1984 to 2022 (the main studies)
  8. Socio-technical autonomous work groups: 1951 to 1965
  9. Stratified hierarchy (requisite organization): 1951 to 1989
  10. Cooperative organization: 1958 to 1999
  11. Linking-pin organization (System 4): 1961 to 1967
  12. Divisional by region: 1962 to 1972
  13. Adhocracy (innovative organization): 1968 to 1979
  14. J-form (the Japanese firm): 1968 to 1990
  15. Professional bureaucracy: 1968 to 1979
  16. Holding company (H-form): 1972 to 1975; 1987
  17. Ambidextrous organization: 1976 to 2004
  18. Simple structure: 1979
  19. Hybrid structure: 1980 to the present
  20. Missionary (ideological) organization: 1980 to 1989
  21. Network, modular, and virtual organization: 1986 to 1990
  22. Transnational and multidimensional organization: 1987 to 1989
  23. Business group (keiretsu, chaebol, family house): 1988 to 2007
  24. Shamrock organization: 1988 to 1989
  25. Federal organization: 1989 to 1992
  26. Self-managing organization (less-hierarchical forms): 1989 to 2017
  27. Boundaryless organization: 1990 to 1995
  28. Front-back organization: 1990s to 2007
  29. Professional partnership (P2 form): 1990
  30. The corporate center and shared services: 1991 to 2000
  31. Internal market organization: 1993 to 1994
  32. Process (horizontal) organization: 1993 to 1999
  33. Cellular organization: 1994 to 1997
  34. Hypertext organization: 1994 to 1995
  35. Team-based organization: 1995
  36. Project-based organization: 2000 to 2006
  37. Agile at scale: 2001 to 2016
  38. Distributed (all-remote) organization: 2002 to 2021
  39. Open-source and community-based organization: 2002 to 2007
  40. Algorithmically managed organization: 2015 to 2020
  41. Team of teams: 2015

Twenty-three of the 29 schools run to the present, and 16 of those began in the 1970s or later. The forms arrive in waves: 14 of the 41 first appear between 1985 and 1994.

Source: Fayol 1949; Weber 1978; Mayo 1933; Trist and Bamforth 1951, an interview-based field study in one industry; Burns and Stalker 1961, a field study of twenty firms; Mintzberg 1979; Child 1972a; Coase 1937; Meyer and Rowan 1977; Hannan and Freeman 1977; Galbraith 1973; Jensen and Meckling 1976; Crozier 1963; Emery and Thorsrud 1969; Yoshino 1968; Streeck 1984a; Kellogg and colleagues 2020, a review of earlier research through one theoretical lens, with no new data; Puranam and colleagues 2014, a conceptual argument; Chandler 1962, several independent case histories by a historian; Pfeffer and Salancik 1978; Simon 1946; Maurice and colleagues 1980, a comparison of matched factories in three countries; Levinthal 1997; Simon 1962; Braverman 1974; Hamilton and Biggart 1988, a comparison of three economies by scholars; Ward 1958; Czarniawska and Joerges 1995; Granovetter 1985.

It answered why mechanized work lowered productivity and morale, and how to design social and technical systems jointly.

Trist and colleagues' Organizational Choice (1963) held that alternative work organizations can exist for the same technology. A. K. Rice (1958) set up semi-autonomous groups in Indian loom sheds from 1953. Fred Emery and Trist (1965) then typed organizational environments' causal textures, and Albert Cherns (1976, 1987) wrote design principles. Sector of origin: coal mining, then manufacturing.

Where it fits

We found no study testing this form against a varied contingency factor. Nine cells are our judgment, two theory, one evidence; three are contested and count in the explorer only where their views agree.

Socio-technical autonomous work groups against the twelve contingency factors

Each factor at its three levels, Fits, Partial, or Strain, or Contested where a contested cell's views disagree; select a row for its reason, sources, and grade.

Select a factor to read its reason, sources, basis, and grade.Size. Groups that own a whole task work in plants of any size, but the more autonomous groups there are, the harder coordination between them becomes. Basis: Judgment. Sources: Ingvaldsen and Rolfsen 2012; Beekun 1989.Age. Groups have been set up at a new site designed for them, and in an old mill they regressed where their boundary conditions were not maintained; that they are easiest to start at new sites is our judgment, since the one greenfield study had no older site to compare. Basis: Judgment. Sources: Miller 1975; Kemp and colleagues 1983.Environmental uncertainty (unpredictability, dynamism). Groups that control variances at the source handle shifting demands on the work; with very stable work the gain over well-designed individual jobs is smaller. Basis: Judgment. Sources: Pasmore 1988; Galbraith 1973; Schoonhoven 1981; Cherns 1976; Pennings 1975.Environmental complexity. Semi-autonomous groups answer the variances of their own production task; environmental complexity bears on them little. Basis: Judgment. Sources: Trist and colleagues 1963.Munificence and hostility. Groups that own a whole task can cut their own costs under pressure, as one study of groups policing their spending shows; nothing read ties the form to munificence as such. Basis: Judgment. Sources: Fuhrich 2019.Diversity of products and markets. Semi-autonomous groups are designed at the level of the production task, whatever the business mix. Basis: Judgment. Sources: Trist and colleagues 1963.Strategy type. Works disagree: Burton and Obel give a defender centralized, formalized control, the opposite of self-regulating groups, while reported sociotechnical experiments were largely successful on productivity and costs, the efficiency a defender competes on; no work read ties the form to a strategy type. Basis: Judgment. Sources: Pasmore and colleagues 1982; Burton and Obel 2004. View 1: Defender (exploitation) strain, Analyzer (exploitation and exploration) partial, Prospector (exploration) partial. Burton and Obel give a defender high formalization and centralization and a functional design; groups that regulate their own work, low in formalization and decentralized, meet neither. Sources: Burton and Obel 2004. View 2: Defender (exploitation) partial, Analyzer (exploitation and exploration) partial, Prospector (exploration) partial. A review of 134 reported sociotechnical experiments found them extremely successful overall on productivity, costs, quality, and satisfaction, an efficiency aim a defender shares, so the form is not ruled out for a defender; no work read ties it to a strategy type. Sources: Pasmore and colleagues 1982.Technology and routineness. The same technology admits more than one way of organizing work, and reported sociotechnical experiments were largely successful, though few involved technological change; whether that holds for routine work is disputed, and knowledge work with no known procedure is the team-based form's ground. Basis: Theory. Grade: conceptual only, no empirical test. Sources: Trist and colleagues 1963; Pasmore and colleagues 1982; Beekun 1989; Rice 1958; Mohrman and colleagues 1995; Kelly 1978; Wall and colleagues 1986; Miller 1975; Burton and Obel 2004; Hickson and colleagues 1969; Mohr 1971. View 1: Routine fits, Mixed fits, Nonroutine partial. The same technology admits more than one way of organizing work, and a review of 134 experiments found sociotechnical designs successful overall, though few involved technological change. Sources: Trist and colleagues 1963; Pasmore and colleagues 1982; Beekun 1989; Rice 1958; Mohrman and colleagues 1995. View 2: Routine partial, Mixed partial, Nonroutine partial. A reappraisal of the founding cases found organizational choice inadequate, group autonomy limited, and pay incentives underestimated; a long-term field experiment found no effect on motivation or performance; and group working persisted in a non-automatic loom shed but largely disappeared in automatic ones. Sources: Kelly 1978; Wall and colleagues 1986; Miller 1975. View 3: Routine strain, Mixed partial, Nonroutine partial. Burton and Obel prescribe high formalization and coordination through rules and planning for routine technology, which groups that regulate their own work by design do not supply. Sources: Burton and Obel 2004.Interdependence. Drawing group boundaries at the technology's natural breaks puts the back-and-forth work inside one group, so links between groups are only pooled or sequential; reciprocal work across groups is left to management at the boundary. Basis: Theory. Grade: conceptual only, no empirical test. Sources: Cherns 1976; Thompson 1967.Knowledge intensity and professionalization. Groups that own a whole task and share skills learned on the job are this form's origin in mining and weaving; skills formed by long outside training are less interchangeable among group members. Basis: Evidence. Grade: a multi-firm empirical study or several independent case studies by scholars. Sources: Rice 1958; Cherns 1976; Pasmore and colleagues 1982; Beekun 1989; Wall and colleagues 1986; Kemp and colleagues 1983; Kelly 1978; Osterman 1994. View 1: Low fits, Moderate fits, High partial. Groups that own a whole task and share skills learned on the job are this form's origin in mining and weaving; reported experiments were largely successful, and forming fully autonomous groups raised productivity more than semi-autonomous or nonautonomous ones. Sources: Rice 1958; Cherns 1976; Pasmore and colleagues 1982; Beekun 1989. View 2: Low partial, Moderate fits, High partial. Where autonomous groups did work learned on the job, one long-term field experiment found productivity rose only through removed supervisory positions, with no effect on motivation or performance, and a reanalysis of the founding cases argued the gains reflected intensified labor and underestimated pay incentives; teams and job rotation were adopted more where technology required high skill and heavy training. Sources: Wall and colleagues 1986; Kemp and colleagues 1983; Kelly 1978; Osterman 1994.Global integration pressure. No work read ties sociotechnical work groups to integration pressure across countries, so this factor gives no reason for or against. Basis: Judgment. Sources: Cherns 1976.Local responsiveness pressure. No work read ties sociotechnical work groups to local responsiveness across countries, so this factor gives no reason for or against. Basis: Judgment. Sources: Cherns 1976.

Counting the three contested cells, strategy, technology, and knowledge intensity, by their first views, the 36 settings read Fits at 12, Partial at 23, and Strain at 1, the one strain a defender strategy. Nine of the twelve cells rest on judgment.

Source: Ingvaldsen and Rolfsen 2012, one case study; Beekun 1989, a meta-analysis of 17 studies; Miller 1975, a 16-year follow-up of one mill; Kemp and colleagues 1983, a comparative study at one site; Pasmore 1988; Galbraith 1973; Schoonhoven 1981, a quantitative study of hospital operating room suites; Cherns 1976; Pennings 1975, a quantitative field test of the structural-contingency model; Trist and colleagues 1963, several field studies in one industry; Fuhrich 2019, one case study; Pasmore and colleagues 1982, a review of 134 reported experiments; Burton and Obel 2004; Rice 1958; Mohrman and colleagues 1995; Kelly 1978, a critical reanalysis of several published case studies; Wall and colleagues 1986, a long-term quasi-experiment in one company; Hickson and colleagues 1969, a multi-organization comparative survey; Mohr 1971, a quantitative field study of technology and structure; Thompson 1967; Osterman 1994, a survey of several hundred establishments.

Your own situation against socio-technical autonomous work groups

Set any factor to outline its cell in the strip above; the lines under the switches give the counts, the form's place among the 41, and the misfit rules that fire.

Size
Age
Environmental uncertainty (unpredictability, dynamism)
Environmental complexity
Munificence and hostility
Diversity of products and markets
Strategy type
Technology and routineness
Interdependence
Knowledge intensity and professionalization
Global integration pressure
Local responsiveness pressure

Set: 0 of 12 factors. Fits on 0, Partial on 0, Strain on 0, Contested on 0.

Its place in the explorer's rankingEvery factor is on Any: all 41 forms tie, and the explorer lists them alphabetically.1st of 411st of 41, tied with one other1st of 41, tied with 0 othersSunk by a misfit rule: 1st of 41Sunk by a misfit rule: 1st of 41, tied with one otherSunk by a misfit rule: 1st of 41, tied with 0 others

  • Flagged (its sources disagree): A defender without written standards. Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • Flagged (its sources disagree): A defender with authority spread wide. Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • Flagged (its sources disagree): An analyzer whose stable core has no standards. Disputed: Burton and Obel 2004; against it, Burton and Obel 2004; Mintzberg 1980.
  • Flagged (it meets the form only through part of its profile): An analyzer with authority spread wide. Source: Burton and Obel 2004.
  • Flagged (its sources disagree): A large organization with few written rules. Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • Flagged (its sources disagree): Routine work without written standards. Disputed: Burton and Obel 2004; against it, Trist and colleagues 1963; Pasmore and colleagues 1982; Beekun 1989; against it, Hickson and colleagues 1969; against it, Child and Mansfield 1972.
  • Flagged (its sources disagree): Few written rules in a stable environment. Disputed: Burton and Obel 2004; against it, Burton and Obel 2004; against it, Mintzberg 1980.
  • Flagged (its sources disagree): Reactor: a defender's strategy on a loose structure. Disputed: Miles and Snow 1978; Burton and Obel 2004; against it, Burton and colleagues 2015b; Snow and Hrebiniak 1980.

Ranking rule: by count of Fits on the factors the reader has set, then fewest Strains, then alphabetical; a fired misfit rule that no source disputes sinks the form to the bottom when every level of the form's profile meets it, a disputed one, or one that meets the form only through one end of a spanned level, one view of a contested profile cell, or one part of a two-part form, is flagged on the row (with both views where it is disputed), and a misfit between two of the reader's own settings is shown once above the list; a contested cell counts only where its views agree at the reader's level, and otherwise shows as Contested; factors on "Any" count for nothing; with every factor on "Any" the list is alphabetical and says so.

Open these settings in the explorer

The misfit rules that can fire for this form

  • A defender without written standards fires when Strategy type at Defender (exploitation): flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • A defender with authority spread wide fires when Strategy type at Defender (exploitation): flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • An analyzer whose stable core has no standards fires when Strategy type at Analyzer (exploitation and exploration): flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Burton and Obel 2004; Mintzberg 1980.
  • An analyzer with authority spread wide fires when Strategy type at Analyzer (exploitation and exploration): flagged (it meets the form only through part of its profile). Source: Burton and Obel 2004.
  • A large organization with few written rules fires when Size at Large: flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • Routine work without written standards fires when Technology and routineness at Routine: flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Trist and colleagues 1963; Pasmore and colleagues 1982; Beekun 1989; against it, Hickson and colleagues 1969; against it, Child and Mansfield 1972.
  • Few written rules in a stable environment fires when Environmental uncertainty (unpredictability, dynamism) at Stable: flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Burton and Obel 2004; against it, Mintzberg 1980.
  • Reactor: a defender's strategy on a loose structure fires when Strategy type at Defender (exploitation): flagged (its sources disagree). Disputed: Miles and Snow 1978; Burton and Obel 2004; against it, Burton and colleagues 2015b; Snow and Hrebiniak 1980.

Eight of the 39 contingency misfit rules can fire for this form, and none sinks it. Seven are disputed, and the eighth, authority spread wide in an analyzer, meets the form only through the decentralized view of its contested centralization, so it only flags.

Source: Burton and Obel 2004; Burton and colleagues 2020a.

  • Size. Judgment: fits small and medium, partial large, our reading that coordination gets harder as autonomous groups multiply, from one company (Ingvaldsen and Rolfsen 2012).
  • Age. Judgment: fits young and established, partial old, our reading that groups are easiest to start at new sites built for them and regress in older plants unless their boundary conditions are maintained. Kemp and colleagues (1983) studied one greenfield site; Miller (1975), a 16-year follow-up of one mill, found such regression, though time since an intervention is not age.
  • Environmental uncertainty. Judgment: partial stable, where the gain over well-designed individual jobs is smaller; fits shifting and unpredictable, on Pasmore's (1988) sociotechnical perspective, Cherns's (1976) variance control at the source, a premise about the work more than the environment, and Galbraith's (1973) information processing. Two limits, naming no form, set no level: Schoonhoven (1981), a quantitative study of hospital operating room suites, found traditional hypotheses drawn from Galbraith unsupported and more precise ones better supported, testing work units, not the environment; Pennings (1975), a quantitative field test of the structural-contingency model, found environment-structure fit not explaining effectiveness, a null Donaldson (2001) lays to method.
  • Complexity, diversity, munificence, global integration, and local responsiveness. Judgment: partial throughout. Neutral, since no work read ties the form to them: the groups answer their own task's variances (Trist and colleagues 1963), and cutting their own costs (Fuhrich 2019) is not munificence.
  • Strategy. Judgment, contested at defender: strain, since Burton and Obel's (2004) model, the most developed scholarly fit model, gives a defender high formalization and centralization; or partial, since Pasmore and colleagues (1982), a review of 134 reported experiments, found them extremely successful overall on productivity and costs. Analyzer and prospector: partial.
  • Technology. Theory, three views agreeing only at nonroutine (partial). One: fits routine and mixed, since the same technology admits more than one work organization (Trist and colleagues 1963, several field studies in one industry) and the designs were reported successful, though few changed the technology (Pasmore); nonroutine is our reading, on Mohrman, Cohen, and Mohrman (1995). Two: partial throughout, on Kelly (1978), Wall and colleagues (1986), and Miller, whose groups faded in automatic sheds (automatic as routine is our reading). Three: strain routine, on Burton and Obel. A limit on all: Hickson, Pugh, and Pheysey (1969), a multi-organization comparative survey, and Mohr (1971), a quantitative field study of technology and structure, found technology bearing weakly on structure, and Mohr found effectiveness not depending on their match, another null Donaldson lays to method.
  • Interdependence. Theory: fits pooled and sequential, partial reciprocal. Boundaries at the technology's breaks keep reciprocal work inside one group, as Thompson (1967) groups it first, with variances controlled at source (Cherns 1976); across groups it falls to management.
  • Knowledge intensity. Evidence, contested at low; both views: fits moderate, partial high. One: fits low, the form's origin in skills learned on the job, with Beekun (1989), a meta-analysis of 17 studies, finding fully autonomous groups raising productivity most. Two: partial low, since gains came through removed supervisors without motivation effects (Wall; Kemp, the same program) and, Kelly argued, reflected intensified labor and underrated pay. Moderate rests on the adoption finding of Osterman (1994), a survey of several hundred establishments, tying team practices to high-skill technology and training, not to results. Partial high is our reading.

What it does well

In theory, variances are controlled near their origin (Cherns 1976).

Pasmore and colleagues (1982) found the 134 reported experiments extremely successful overall on productivity, costs, quality, and satisfaction, as the reports state it. Beekun (1989) found a curvilinear duration effect, rejecting a temporary Hawthorne effect.

People like the work: Wall and colleagues (1986) found a lasting rise in intrinsic satisfaction, and Kemp and colleagues (1983), earlier in the same program, higher job satisfaction; Cordery, Mueller, and Smith (1991), a longitudinal study at two plants, better attitudes. At team level, Stewart (2006), a meta-analysis of 93 studies, found autonomy going with higher performance, varying by task type.

What it costs

The supervisors' jobs: Wall and colleagues' (1986) productivity gain came through eliminating supervisory positions, not greater motivation.

Turnover rose at Wall's site, and Cordery, Mueller, and Smith (1991) found turnover and absence both higher among group members. Beekun (1989) reports escape behavior dropping, less where technology also changed.

Pay, Kelly (1978) held, was seriously underestimated, and Beekun found added monetary incentives raising productivity more; pay design belongs to People & Talent.

Inside each group coordination runs on mutual adjustment, the costliest of Thompson's (1967) modes.

How it fails

The Coordination Failures links in parentheses are our judgment.

How socio-technical autonomous work groups fail, and the coordination failures behind them

The entry's failure modes at the left, each tied to the Coordination Failures patterns it names at the right; a pattern links its entry there.

Dormant-assumption reactivationLocal optimization against the global constraintGoal displacement and metric gaming
Dormant-assumption reactivationLocal optimization against the global constraintGoal displacement and metric gaming

How many ties the failure has. Highest: The groups regress when the conditions around them changeCoordination between highly autonomous groups gets hardThe group rationalizes against its own membersDormant-assumption reactivationLocal optimization against the global constraintGoal displacement and metric gaming

Select a failure to read its place in the network.The groups regress when the conditions around them change. Degree 1Coordination between highly autonomous groups gets hard. Degree 1The group rationalizes against its own members. Degree 1Dormant-assumption reactivation. Degree 1Local optimization against the global constraint. Degree 1Goal displacement and metric gaming. Degree 1
FailureDegree
The groups regress when the conditions around them change1
Coordination between highly autonomous groups gets hard1
The group rationalizes against its own members1
Dormant-assumption reactivation1
Local optimization against the global constraint1
Goal displacement and metric gaming1

Three failure modes name three patterns, one each. The entry names a further weakness, autonomy that stays partial, which no Coordination Failures pattern names.

Source: Miller 1975, a 16-year follow-up of one mill; Ingvaldsen and Rolfsen 2012, one case study; Fuhrich 2019, one case study.

The groups regress when the conditions around them change (dormant-assumption reactivation). Miller (1975) found group working largely gone in a newer automatic shed, explained by failure to maintain its boundary conditions as markets changed.

Coordination between highly autonomous groups gets hard (local optimization against the global constraint). Ingvaldsen and Rolfsen (2012), one case study, found it a major challenge.

The autonomy can also stay partial, which no Coordination Failures pattern names. Kelly (1978) found it subordinate to economic objectives. Héthy (1986), a comparative review of official reports and published surveys from several countries, found brigade leaders mostly appointed by management and some brigades of 50 to 200 members, too large to work as collectives.

The group rationalizes against its own members (goal displacement and metric gaming). Fuhrich (2019), one case study, found Volkswagen groups costing their output and pressing to remove weaker members; one group split into factions and bullied them. Barker (1993) found team members' own value-based rules controlling them more completely than the bureaucracy they replaced.

None of the misfit rules names the form; eight fire on its profile, and none sinks it. One is undisputed: authority spread wide for an analyzer, which flags it, met only on the decentralized view of its contested centralization. Seven are disputed and flagged: few written standards for a defender, and authority spread wide for one (on the decentralized view), against Mintzberg's (1980) informal, decentralized professional bureaucracy in complex but stable environments; few standards for an analyzer, against the same Burton and Obel book's simple configuration for a small, innovating one; a large organization with few rules (against Mintzberg); routine work without written standards (against the socio-technical evidence of Trist and colleagues, Pasmore, and Beekun, whose production work we read as routine; the Aston reappraisal; and Child and Mansfield 1972, a multi-firm comparative study, on size outweighing technology); few rules in a stable environment (against Mintzberg and another Burton and Obel proposition); and the reactor pairing, our construction from Miles and Snow (1978) and Burton and Obel, also needing the decentralized view (against Burton, Obel, and Håkonsson 2015, and Snow and Hrebiniak 1980, a study of top managers' perceptions in four industries, on data they call only suggestive). The propositions were tested only as an earlier set: in a quantitative study of 224 Danish small and medium-sized firms, misfits went with lower return on assets, added misfits with no further loss (Burton, Lauridsen, and Obel 2002).

What the research found

The reviews disagree. Pasmore and colleagues (1982) found success overall, in a literature favoring successes, but few technological changes, leaving much to learn about joint optimization. Kelly (1978) found two streams, not one school, put intensification of labor in place of joint optimization, and held organizational choice inadequate, the cases favoring one form. Beekun (1989) found interventions that also changed technology raised productivity less; his pay finding agrees with Kelly. Guzzo, Jette, and Katzell (1985), a meta-analysis of experiments on 11 types of psychologically based intervention, found a pooled productivity gain of nearly half a standard deviation; sociotechnical redesign is among the types only by secondary accounts.

One program qualifies the record: Kemp and colleagues (1983), a comparative study at one site, found no effect on motivation or commitment; Wall and colleagues (1986), following the same program longer, no effect on motivation or performance.

Héthy (1986) found the brigades driven by economic rationality and managerial control, with small average gains (1% to 2%, 10% to 15% in outstanding brigades) hard to separate from political support and pay changes; Kern and Schumann (1987), the authors' summary of their West German research, saw rationalization professionalizing a core and excluding others.

No study from 2010 on was found in the eight major journals. Weatherburn (2020), a historical archival study, found the Rowntree studies ranking financial incentives above social ones and called Hawthorne's unique status an invented tradition. The most recent, Banerjee (2026), one historical case reread from archives, revisits Rice and Miller, finding the Ahmedabad success achieved by erasing the lowest-caste spinners' needs and voices and silencing the union.

The schools dissent. Socio-technical writers reject Taylor's fragmentation of jobs and separation of planning from doing (Trist and Bamforth; Cherns), and against human relations hold better supervision not enough without changing the technical design. Woodward (1965), a survey of a hundred manufacturing firms with case studies, read structure as following technology, where Trist saw choice; the Aston reappraisal (Hickson and colleagues 1969) tied technology only to structure near the workflow. Scandinavian writers treat participation as a value in itself, lean production's defenders as one input to efficiency (Emery and Thorsrud 1976; Adler and Cole 1993). Berggren (1992), six case studies in the Swedish auto industry by one scholar, had set the Swedish groups against lean production on productivity and quality of work; Adler and Cole, an ungraded Sloan Management Review article, held NUMMI's standardized work better for organizational learning, though Uddevalla's groups were more promising for individual learning.

The evidence ledger holds 20 studies filed under this form.

In practice: one published case

Miller (1975), "Socio-technical systems in weaving, 1953 to 1970: a follow-up study," in Human Relations, is one case study with a 16-year follow-up, read in its abstract. Its subject, unnamed in the abstract and so here, is the Indian textile mill of Rice's experiments.

In 1953 to 1954, experimental changes in work organization were introduced in the mill's automatic and non-automatic loom sheds. In each, semi-autonomous work groups were made responsible for production and routine maintenance on a group of looms. The group system was later extended to other sheds.

A 1970 follow-up covered four locations, including the two original sites. In the non-automatic experimental shed, work organization and performance had stayed virtually unchanged over 16 years. In a newer automatic shed, group working had largely disappeared, the original automatic site had regressed considerably, and the remaining non-automatic site showed variation in modes of group working.

The study concluded that the persistence of the group system in at least one area substantially confirmed the assumptions of Rice's experiment, that the regression in the automatic sheds came from failure to maintain the boundary conditions for group working as markets and other environmental factors changed, and that Rice's innovations were more radical than recognized at the time.

Source: Miller 1975. Socio-technical systems in weaving, 1953 to 1970: a follow-up study, Human Relations 28(4), 349 to 386; a 16-year follow-up of one mill.

Variants and near relatives

Norwegian Industrial Democracy Program (Emery and Thorsrud 1976, four field experiments in Norwegian firms, reported by the program's researchers): 1960s experiments with semi-autonomous groups as the route to democracy at work. Democratic dialogue (Gustavsen) (1992) followed: broad change with broad participation, after the single-site experiments met problems of diffusion.

Dutch integral organizational renewal (de Sitter, den Hertog, and Dankbaar 1997): a design theory its authors present as a Dutch variant of sociotechnical systems design, turning complex organizations with simple jobs into simple organizations with complex jobs, as an expert approach or a route to self-design. Van Eijnatten and van der Zwaan (1998) find the national approaches sharing the ideal of participative democracy, though at the conceptual level they appear quite incompatible: local forms, they presume, of one sociotechnical paradigm.

Ahmedabad experiment (Rice 1958): weaving reorganized into groups of workers for groups of looms, at Calico Mills; Banerjee (2026) disputes the received account.

Swedish alternatives to lean production (Volvo Uddevalla and Kalmar) (Berggren 1992): groups assembling whole vehicles or large parts in place of the paced line.

Olivetti's production islands (Butera and De Witt 2011): 1970s integrated assembly units owning product quality and testing.

Socialist work brigades (Héthy 1986; Aage 1989, a conceptual model): teams inside one enterprise, given a whole phase of production and paid on collective results, distinct from the planned economy's M-form and U-form. Hungary's partnerships (VGMKs), formally separate undertakings, belong to the Internal market organization.

West German Humanization of Working Life program (Sauer 2011; Fuhrich 2019): the federal program begun in 1974 against health-damaging and monotonous work.

The Team-based organization is the nearest neighbor. This form starts from a production system and fits the group to the technology's breaks; the team-based form starts from teams and designs the whole organization around them for knowledge work.

Relatives: the Self-managing organization removes managers, where this form keeps them at the boundary. The Process (horizontal) organization makes the end-to-end process the unit, not a group owning one segment. The J-form (the Japanese firm) teaches the Toyota shop floor that Adler and Cole set against Uddevalla. Job design (Hackman and Oldham 1980) is out of scope and pointed to People & Talent.

Socio-technical work groups beside their nearest neighbor, the team-based organization

Select a number to fill what it marks in both drawings; the list under them gives the dimensions on which the two profiles differ.

It designs the social and technical systems of a production process together, so each group owns a whole segment bounded where the technology naturally breaks and controls that segment's variances at their source, with management placed at the group's boundary, not over each job.

Socio-technical autonomous work groups

SpokespersonsManagementWork groupsManagement atthe boundaryWork groupWork groupWork groupEach owns a whole segmentMN1234
SpokespersonsManagementWork groupsManagersWorkgroupWorkgroupWorkgroupEach owns a whole segmentMN1234

Reports toAdjust to each other directlyMMutual adjustmentNStandardization of norms

Team-based organization

Top managementUnits of teams(Moderate end)TeamsTop managementteamUnit managementteamUnit managementteamWork teamWork teamIntegratingteamWork teamImprovementteamBusiness units of teamsM122234
Top managementUnits of teams (Moderate end)TeamsTop teamUnitteamUnitteamWorkteamWorkteamIntegrateWorkteamImproveBusiness units of teamsM122234

Reports toAdjust to each other directlyMMutual adjustment

  1. Each socio-technical group owns a whole segment of a production process, bounded where the technology breaks; the team-based form groups its teams into business units, each built around the work it delivers.
  2. Management works at the socio-technical groups' boundary, regulating what crosses it, whereas the team-based form builds management itself as teams.
  3. Socio-technical groups coordinate by mutual adjustment and their own norms, the team-based form's teams by mutual adjustment.
  4. Between socio-technical groups, one case study explored rotating spokespersons and found coordination a major challenge; the team-based form adds integrating teams that join work teams across its units.

Differs on 6 of 11 dimensions

  • Specialization Low; Team-based organization: Low to medium
  • Standardization and formalization Low; Team-based organization: Medium
  • Layers and span of control Flat; Team-based organization: Flat to moderate
  • Decision rights and the locus of authority Vertical and horizontal decentralization; Team-based organization: Selective vertical and horizontal decentralization
  • Prime coordinating mechanism Mutual adjustment and standardization of norms; Team-based organization: Mutual adjustment
  • Lateral mechanisms Direct contact and liaison roles; Team-based organization: Task forces and teams

Both forms do the work in groups and keep managers. Socio-technical design draws each group around a whole segment of a production process and keeps management at its boundary, while the team-based form designs every level, management included, as teams.

Source: Cummings 1978; Cherns 1976; Ingvaldsen and Rolfsen 2012, one case study; Mohrman and colleagues 1995.

Moving into or out of it

Cherns (1987) binds the principles of the design process and design team to those of the design itself.

Supervisory posts may go or be questioned (Wall and colleagues 1986; Fuhrich 2019). Spreading is harder: Gustavsen (1992) records the Norwegian experiments' diffusion problems, and Fuhrich's workers feared their group work would be ended as too expensive. Moving out can happen by drift, as in Miller's (1975) automatic sheds, in one case.

Change & Transformation takes up how to run the reorganization.

The thinking behind it

Eric Trist and Ken Bamforth (1951): mechanized mining broke up self-regulating groups, so the two systems must be designed together. Trist (1963): organizational choice, the form's premise.

Fred Emery and Einar Thorsrud (1976): democracy at work, the Norwegian variant.

A. Kenneth Rice (1958): groups owning production and maintenance on a group of looms, the whole segment. Eric Miller (1975): boundary conditions need maintaining.

Albert Cherns (1976, 1987): minimal critical specification, the form's low formalization. Thomas Cummings (1978): the self-regulating work group and managerial functions for its design and supervision.

John Kelly (1978): limited autonomy and underrated pay, the contested centralization.

Bjørn Gustavsen (1992), action research: democratic dialogue, the Norwegian sequel. Ulbo de Sitter (1997): simple organizations with complex jobs, the Dutch variant. Christian Berggren (1992): an alternative to the paced line, the Swedish variant. Lajos Héthy (1986): why the brigades' autonomy stayed partial. Gina Fuhrich (2019): workers as rationalizers of their own groups.

James D. Thompson (1967): mutual adjustment for reciprocal interdependence, the group's own mechanism. Richard Burton and Børge Obel (2004): their fit model and misfit rules.

Also called

Socio-technical systems design (also sociotechnical). Self-regulating work groups (Cummings). Semi-autonomous work groups (Rice; Emery and Thorsrud). Autonomous work groups or workgroups (Wall; Cordery; Ingvaldsen). Group working (Miller). Production islands (isole di produzione) at Olivetti.

Works cited

On this page, “read” means read by this series: where a line says a work, or part of one, was not read, the series did not read it.

Back to the explorer