Can we match dependency types to coordination processes?

Malone and Crowston's 1994 ACM Computing Surveys paper, "The Interdisciplinary Study of Coordination," proposes a deliberately simple operational definition: "Coordination is managing dependencies between activities." They argue the definition's payoff is methodological: "further progress should be possible by characterizing different kinds of dependencies and identifying the coordination processes that can be used to manage them." The excerpt names several dependency types treated this way: "managing shared resources, producer/consumer relationships, simultaneity constraints, and task/subtask dependencies."

Their reasoning is that this definition acts as a translation layer across disciplines — organization theory, economics, computer science, linguistics, psychology — because "the concepts of coordination theory can help identify similarities among concepts and results in different disciplines," letting ideas "be transported back and forth across disciplinary boundaries." The definition is deliberately inclusive: political processes, cooperation, collaboration, and competition "each... involve[] managing dependencies between activities," and coordination occurs not only among people but in "human, computational, biological, and other" systems. Dependencies among non-activity objects — the authors give the example of components in a manufactured product — are folded in by treating them as dependencies among the activities that touch those objects (e.g., designing or redesigning a component).

This classic predates software agents by three decades, but the taxonomy move it proposes — classify the dependency first, then pick the coordination mechanism — is the same move made by Do single agents always hit organizational limits?, which argues system intelligence requires graph abstractions over agent relationships rather than per-agent capability gains. It is also the underlying structure behind How do humans and technology share control over organizational routines?, which sorts shared human-technology agency by who controls protocols versus actions — itself a dependency-management question in Malone and Crowston's sense, decades later and in a different vocabulary.

The excerpt is conceptual, not an empirical study of any real system: it names no AI agents and no LLMs, and the "coordination processes" it catalogs reflect the 1994 state of multidisciplinary research rather than a complete or current taxonomy. It sets out a framework and a research agenda, not a specific prescription for any architecture. Reading "dependency type → coordination process" as a ready-made checklist for governing today's agent systems would extend the paper beyond what it claims; it offers the logic of the move, not the inventory of agent-specific dependencies and mechanisms that logic would require.