Task Interdependence as a Design Constraint: Restoring Thompson’s Typology to Match Organizational Interventions to Work Structure

Peter Stefanyi, Ph.D., MCC.

Colaborix GmbH, January 2026

Abstract

Organizations invest heavily in interventions intended to improve performance—training, teamwork programs, collaboration platforms, process improvement systems, and more. Yet results are highly variable, and the same intervention can succeed dramatically in one context and fail in another. We argue that a major source of this variance is conceptual: research and practice frequently treat “task interdependence” as a single continuous construct (“low vs. high”) and thereby obscure a foundational structural distinction introduced by Thompson’s (1967) classic typology—pooled, sequential, and reciprocal interdependence. We develop a design-constrained theory of interventions: task interdependence structure determines (1) the dominant performance limiting mechanism, (2) the coordination mode required, and (3) which classes of interventions are likely to produce benefit versus pure overhead. We integrate organizational theory, team effectiveness research, coordination theory, and operations management into a single contingency framework. We advance propositions predicting differential intervention returns across interdependence structures, explain why “one-size-fits-all” collaboration and AI adoption programs generate mixed findings, and offer an agenda for cumulative research based on task-structure coding as a standard design variable in intervention studies.

Keywords: task interdependence; coordination; team interventions; organizational design; process improvement; human–AI collaboration

1. Introduction: the intervention variance problem

Organizations spend substantial resources trying to improve performance through people- and process-focused interventions: training, coaching, team development, collaboration tools, continuous improvement programs, and—more recently—enterprise AI rollouts. Yet observed outcomes are inconsistent: some initiatives produce clear gains, many plateau, and a nontrivial fraction create additional coordination load without commensurate benefit.

We propose that a major driver of inconsistent outcomes is that many interventions are applied without first identifying a key structural property of the work: the form of interdependence that links contributors to the output. Thompson (1967) distinguished pooled, sequential, and reciprocal interdependence and argued that each requires a distinct dominant coordination mechanism. This typology remains canonical in organization theory, but much empirical work and most managerial practice collapse interdependence into a single continuum or a high/low binary. That collapse weakens prescriptive power: “high interdependence” can describe an assembly line and a cross-functional innovation team, yet the most effective interventions for these contexts are often different in kind.

This article restores Thompson’s three-way distinction as a design constraint for selecting and evaluating interventions. We develop an integrative framework that links (a) task interdependence structure to (b) coordination requirements and performance-limiting mechanisms, and then to (c) classes of interventions with predictable payoffs. We contribute by (1) re-deriving why the three types imply different dominant constraints, (2) integrating literatures that rarely speak to each other (teams, coordination theory, operations/process improvement, and emerging human–AI work), and (3) offering falsifiable propositions and a research agenda that can reduce contradictory findings by systematically coding task structure.

2. Task interdependence: restoring a structural typology

2.1 Thompson’s three forms

Thompson (1967) proposed three ideal-typical forms of interdependence:

• Pooled interdependence: contributors work largely independently; outputs aggregate at a higher level.

• Sequential interdependence: work flows through an ordered chain; outputs of one role become inputs to the next.

• Reciprocal interdependence: participants’ outputs mutually shape each other through iterative adjustment.

Thompson’s core organizational design claim is that interdependence structure determines coordination mode: pooled work is coordinated primarily through standardization, sequential work through planning and scheduling, and reciprocal work through mutual adjustment (Thompson, 1967). This idea aligns with later information-processing accounts of organizational design: uncertainty and interdependence increase information-processing needs, demanding richer coordination mechanisms (Galbraith, 1973).

2.2 Why the distinction matters now

Across decades, research on interdependence increasingly emphasizes measured perceptions or continuous scales and often uses “task interdependence” as a moderator in team process models (e.g., meta-analytic integration of structural interdependence; Courtright et al., 2015 ). This has been useful for statistical modeling, but it can flatten the qualitative difference between sequential and reciprocal work—two forms that share “high dependence” yet differ in their dominant failure mode.

Simultaneously, organizations are aggressively scaling interventions—collaboration rituals, “team alignment” meetings, tool rollouts, and AI assistants—often with uniform designs across roles. That uniformity collides with structural heterogeneity: the same organization typically contains all three interdependence forms at once.

3. A design-constrained theory of interventions

3.1 Core claim

Interventions are not universally effective; they are conditionally effective because work structure changes what “performance improvement” means. 

Interventions can be thought of as acting on one (or more) of three generic levers:

1. Individual capability (skills, knowledge, tools, selection)

2. Flow capability (handoffs, bottlenecks, buffers, sequencing, throughput)

3. Coordination capability (shared understanding, communication patterns, role integration)

Our central argument is that each interdependence structure tends to be constrained primarily by a different lever, making mismatched interventions systematically low-return or even harmful.

3.2 Mechanism mapping

Pooled interdependence: additive output + coordination as overhead

In pooled work, output is mainly an aggregation of individual contributions; the dominant improvement lever is individual capability and tool enablement. Coordination primarily ensures compatibility (standards) and prevents interference (deconfliction). Excessive mutual adjustment mechanisms can create overhead without increasing output.

Proposition 1 (Pooled–coordination overhead):For pooled interdependence tasks, increases in rich coordination mechanisms (e.g., frequent cross-member synchronization) will exhibit diminishing or negative marginal returns to performance once basic standardization is met.

Sequential interdependence: throughput constrained by the weakest link

In sequential work, performance often depends on throughput and variability across steps; delays and rework propagate downstream. Operations research and practice emphasize constraint-focused improvement: elevating the constraint step changes system output, while improving non-constraints often does not (Goldratt, 1984 (Wikipedia)). Planning, buffers, and handoff design are dominant.

Proposition 2 (Sequential–constraint leverage):For sequential interdependence tasks, interventions targeted at the system constraint will produce significantly greater performance gains than interventions spread uniformly across non-constraint roles.

Reciprocal interdependence: integration quality as the dominant constraint

In reciprocal work, outputs are co-produced via iteration, negotiation, and integration. Performance hinges on the quality of mutual adjustment—shared understanding, timely information exchange, conflict handling, and integration decisions. Team intervention meta-analyses support that teamwork interventions can improve team performance, with effectiveness moderated by features such as team composition and stability (e.g., evidence of positive effects of team training; Salas et al., 2008 (PubMed); teamwork training meta-analysis; McEwan et al., 2017 (PLOS)). For reciprocal work, interventions that build coordination capability should outperform purely individual capability approaches once threshold competence is met.

Proposition 3 (Reciprocal–coordination dominance):For reciprocal interdependence tasks, interventions that strengthen coordination capability will yield larger performance improvements than interventions focused only on individual capability, controlling for baseline competence.

3.3 Coordination modes as “fit”

The framework implies a fit logic between structure and coordination mechanism:

• Pooled → Standardization (rules, interfaces, quality standards)

• Sequential → Planning (handoffs, schedules, buffers, constraint management)

• Reciprocal → Mutual adjustment (real-time communication, shared mental models, integration routines)

Proposition 4 (Fit):Interventions that reinforce the dominant coordination mode implied by a task’s interdependence structure will outperform “misfit” interventions that import a coordination mode from a different structure.

4. Implications for modern interventions, including AI rollouts

4.1 Why one-size collaboration programs create mixed results

Uniform “collaboration intensification” (more meetings, universal alignment, shared rituals) is often implicitly optimized for reciprocal work. When applied to pooled work, it becomes overhead; when applied to sequential work, it may distract from constraint and handoff design.

4.2 Human–AI work as a new test arena for the theory

The recent human–AI collaboration literature shows heterogeneity in outcomes. A preregistered systematic review and meta-analysis of experimental studies reports that, on average, human–AI combinations can underperform the better of human-only or AI-only baselines, with important moderation by task characteristics (Vaccaro et al., 2024 (PubMed)). This is precisely the pattern one would expect if studies mix interdependence structures but do not code them explicitly.

Rather than treating “augmentation vs. automation” as a universal strategic paradox, interdependence structure suggests differentiated design: pooled work often benefits from individualized assistance; sequential work benefits from constraint-focused insertion; reciprocal work benefits from integration routines and shared calibration.

This aligns conceptually with arguments that organizations face tensions between automation and augmentation strategies (Raisch & Krakowski, 2021 (journals.aom.org)), but we shift the locus of explanation: the tension may partly be an artifact of aggregating across heterogeneous task structures.

Proposition 5 (Human–AI heterogeneity):In human–AI collaboration settings, the effect of AI assistance on performance will vary systematically by task interdependence structure; pooling results across structures without coding structure will inflate heterogeneity and can produce near-zero average effects despite strong positive effects in some structures and negative effects in others.

5. Boundary conditions and extensions

5.1 Mixed work and decomposition

Many roles combine multiple structures across sub-activities. The theory predicts that intervention choice should be made at the activity level when feasible: e.g., design discussions are reciprocal, while approval workflows are sequential, while individual execution may be pooled.

Proposition 6 (Decomposition advantage):For mixed-structure roles, decomposed interventions matched to each sub-activity’s interdependence structure will outperform a single uniform intervention applied at the job or team level.

5.2 Virtuality, stability, and scale

Team stability and size moderate coordination feasibility; meta-analytic work on team training identifies moderators such as team size and membership stability (Salas et al., 2008 (PubMed)). Reciprocal coordination demands tend to increase nonlinearly with size, while pooled work scales more easily under standardization.

5.3 Measurement implication

A key research implication is methodological: intervention studies should code task interdependence structure explicitly rather than only measuring perceived interdependence on a single continuum. This would make meta-analyses more cumulatively informative.

6. Research agenda: making results cumulative

6.1 Minimum reporting standard for intervention studies

We propose a minimum standard for future studies:

1. Explicit classification of the focal work as pooled, sequential, reciprocal, or mixed (with operational criteria).

2. Description of coordination mechanism embedded in the intervention (standardization, planning, mutual adjustment).

3. A fit test: does structure × mechanism predict outcomes?

6.2 Suggested study designs

• Field quasi-experiment: matched units using different interventions, with pre/post outcomes and structure coding.

• Multi-site experiments: factorial designs that vary intervention class and structure.

• Re-analysis of prior human–AI studies: code interdependence structure and test whether it explains heterogeneity consistent with Vaccaro et al. (2024). (PubMed)

7. Discussion and contribution

This paper contributes a design-constrained theory that restores a foundational typology and makes it prescriptive again. The key move is shifting from “interdependence as a degree” to “interdependence as a form,” and then using form to predict dominant constraints, coordination modes, and intervention fit. The payoff is practical and scientific: organizations can stop applying uniform interventions across structurally different work, and researchers can reduce contradictory findings by coding structure as a primary design variable.

References :

Courtright, S. H., Thurgood, G. R., Stewart, G. L., & Pierotti, A. J. (2015). Structural interdependence in teams: An integrative framework and meta-analysis. Journal of Applied Psychology, 100(6), 1825–1846.

Galbraith, J. R. (1973). Designing complex organizations. Addison-Wesley.

Goldratt, E. M. (1984). The goal: A process of ongoing improvement. North River Press. (Wikipedia)

McEwan, D., Ruissen, G. R., Eys, M. A., Zumbo, B. D., & Beauchamp, M. R. (2017). The effectiveness of teamwork training on teamwork behaviors and team performance: A systematic review and meta-analysis of controlled interventions. PLOS ONE, 12(1), e0169604. (PLOS)

Raisch, S., & Krakowski, S. (2021). Artificial intelligence and management: The automation–augmentation paradox. Academy of Management Review, 46(1), 192–210. (journals.aom.org)

Salas, E., DiazGranados, D., Klein, C., Burke, C. S., Stagl, K. C., Goodwin, G. F., & Halpin, S. M. (2008). Does team training improve team performance? A meta-analysis. Human Factors, 50(6), 903–933. (PubMed)

Thompson, J. D. (1967). Organizations in action: Social science bases of administrative theory. McGraw-Hill.

Vaccaro, M., Almaatouq, A., & Malone, T. W. (2024). When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour, 8(12), 2293–2303. https://doi.org/10.1038/s41562-024-02024-1 (PubMed)