Thinkers whose frameworks illuminate this section.
Brooks coined 'Conway's Law' in The Mythical Man-Month and introduced conceptual integrity — the one-mind property that Conway's Law implies is architecturally optimal. Conway's entire AI argument is the return of Brooks's ideal at industrial scale.
Simon's bounded rationality and satisficing are the cognitive foundations of why source noise exists — humans approximate rather than optimize, and AI faithfully reproduces those approximations. Simon's architecture of complexity is the theoretical framework behind Conway's empirical law.
Shannon's information theory provides the precise mathematical framework for the signal/noise distinction that structures ch1 and ch3. Transmission noise, source noise, and signal fidelity are Shannon's vocabulary applied to organizational communication.
Weick's sensemaking theory — how people impose retrospective meaning on ambiguous situations — explains why cognitive diversity in committees catches errors that one-mind systems cannot: different sensemaking frames produce different error-detection sensitivity.
Argyris's double-loop learning is the mechanism that ch5 invokes when arguing that committees provide genuine external accountability — the ability to question assumptions, not just correct errors within existing assumptions. The Inverse Cognitive Maneuver is a protocol for individual double-loop learning.
Senge's learning organization and systems thinking provide the organizational counterpart to Conway's structural claims — the argument that organizations must develop the capacity to see system-wide patterns, not just optimize local components.
Edmondson's psychological safety research shows empirically what ch5 argues theoretically: that teams produce better decisions when members can surface concerns without fear of reprisal. The committee's function is to institutionalize that condition.
Perrow's normal accidents theory — that complex, tightly coupled systems will inevitably produce catastrophic failures — is the safety-systems counterpart to Conway's architectural argument. Challenger is exactly the kind of accident Perrow predicted.
Vaughan's analysis of the Challenger disaster as normalization of deviance — the gradual organizational acceptance of risk that should have remained unacceptable — is the specific mechanism ch5 invokes. Her work turns Challenger from a story about failed committees into a story about successful ones.
Beer's viable system model provides the cybernetic architecture of exactly the governance structures Conway's law implies are optimal — recursive, autonomous subsystems communicating through defined channels, which is what both two-pizza teams and one-mind AI-augmented builders aspire to.
Mintzberg's organizational configurations and the distinction between adhocracy and bureaucracy map directly onto Conway's Law's predictions: adhocracies produce flexible architectures, bureaucracies produce rigid ones. The AI moment is the emergence of the ultimate adhocracy.
Arthur's increasing returns and path dependence explain why Conway's Law's architectural consequences persist long after the organizational conditions that caused them have changed — the fossil-record metaphor in ch2 is Arthur's path dependence applied to codebases.
