This is Part 3 of the seven-part Logic and Reasoning field guide. Parts I and II established the foundations and limits of single-agent reasoning. This part takes the obvious next step: what happens when reasoning is spread across multiple cooperating agents? The shift is consequential — many problems that are intractable for a single agent become tractable when properly decomposed, and many problems that look identical from outside are qualitatively different depending on whether one agent or a team is solving them.

The natural companion is the AI Agent Teams monograph, which covers the engineering side of the same shift. This post is the theory side. The two read well together.

What it covers

About thirty minutes of careful reading.

Why decompose. The complexity argument. The expertise argument. The parallelism argument. When decomposition genuinely helps and when it just adds coordination overhead for no gain.

Architectural patterns. Blackboard systems (a shared workspace where agents post intermediate results). Contract net (auction-based work allocation). Market mechanisms (resource pricing as the coordination signal). Hierarchical patterns (manager-worker). The matrix of when each one is the right reach.

Coordination protocols. Synchronous versus asynchronous. The handoff disciplines. The deadlock-and-livelock failure modes and the textbook mitigations.

Modes of collective reasoning. Voting. Argumentation (Dung-style). Belief revision in groups. Bayesian aggregation. The handful of formal frameworks for “what does the team conclude?”

The communication problem. Why getting agents to agree on a vocabulary is harder than getting them to agree on conclusions. Ontologies, schemas, and the practical answer (just standardise on a shared schema and move on).

A worked example. A multi-agent reasoning problem (route planning with diverse constraints from different agents) solved three ways across three architectures, with the trade-offs laid bare.

Read it

Open Part III →

The series

This is Part 3 of 7:

  1. The Foundations and the Human Mind
  2. Limits, Machines, and Problem-Solving
  3. A Multi-Agent Problem-Solver — (this post)
  4. From a Mind to a Society of Minds
  5. The Health of a Thinking System
  6. The Dynamics of Thinking
  7. Understanding and Comprehension

← Back to Autonomy