liveBook · Manning

Chair-of-thought

As explained in The Architecture of Reasoning

A selectable reasoning architecture named in the book's taxonomy.

A team can review a decision carefully and still miss its central trade-off. If every question is asked from the same vantage point, thorough analysis may only reinforce that frame. Chair-of-Thought addresses this failure by holding the problem fixed while deliberately changing the perspective used to examine it. section 5.0

A sound review with a narrow lens

An architecture team approves platform consolidation after examining technical debt, operating costs, migration risk, and timelines. The review is coherent, but architectural integrity dominates. A temporary customer chair could ask whether standardization reduces client flexibility; a commercial chair could examine the revenue consequences. These questions need not reverse the decision. They make its costs explicit before commitment. section 5.1 section 5.6.2

A chair is an optimization frame, not a person or job title. Customer, finance, engineering, and compliance chairs foreground different success criteria and risks. Inviting those functions into a meeting does not itself ensure rotation: people may stay with their familiar positions, and the dominant lens may still govern the decision. section 5.2 section 5.3

Use the smallest rotation that reveals a material tension:

  • State one narrow problem and identify a few chairs with genuinely different optimization criteria. section 5.5
  • Visit each chair in sequence. Record what it values, what risks it sees, and which trade-offs it cannot easily accept; compare only after the passes are complete. section 5.5
  • Choose deliberately. Name which concerns the decision addresses and which perspectives it partly deprioritizes. section 5.5

Rotate when options have been explored but are still judged through one lens, especially where optimization criteria conflict. Do not turn a bounded question into a ceremony. Too many chairs raise cognitive load; shallow passes, defensiveness, or an unsafe setting can leave the dominant view intact. Treat each chair as a temporary pass, not a position to defend. section 5.7 section 4.8

Table 5.1 A compact reasoning scaffold for chair rotation
Element
Purpose
Guiding questions
Activate a chair
Shift the optimization frame
From the customer chair, what feels most fragile here?
From the long-term maintenance chair, what will this look like in two years?
If we were the regulator reviewing this, what would concern us?
Sequence rotation
Keep perspectives distinct
For the next few minutes, we’re reasoning only from the operations / customer / commercial seat.
Clarify each lens
Make the chair concrete
What does this seat optimise for?
What risks become visible from here?
What trade-offs feel unacceptable from this vantage point?
Compare across chairs
Surface tensions explicitly
Where do these chairs disagree most strongly?
Which risks appear across multiple lenses?
What tensions are unavoidable?
Converge deliberately
Choose trade-offs consciously
Which chair is least comfortable with this direction, and are we prepared to accept that?

Use the compact scaffold to activate and clarify each chair, keep passes separate, compare tensions, and then converge. A quick opening is: “Which chair are we currently reasoning from? Which chair would disagree most strongly?” section 5.6.1

How The Architecture of Reasoning covers it

How deep it goes
Defines chairs as optimization frames and gives a sequential facilitation procedure, including convergence and failure modes.
Examples
Uses a platform-consolidation decision and customer, commercial, and architectural perspectives to reveal otherwise implicit costs.
What it assumes
Assumes a team decision to examine but defines the chair as an optimization frame rather than requiring AI prompting expertise.
What you can do after
Practice/perform a brief chair rotation and state the trade-offs accepted before committing to a decision.
How it connects
Contrasts perspective rotation with Tree-of-Thought branching and ReAct evidence testing.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of The Architecture of Reasoning. The section links point to the parts of the book it draws on.