Skip to main content

​
Overview: Structure → Unstructure → Restructure

The dialectical process has seven phases, but they follow John Boyd’s fundamental cycle:
  1. Structure — You have a position, tension, or contradiction
  2. Unstructure — The Monks shatter both positions into atomic parts, exposing the hidden question
  3. Restructure — Synthesis recombines the parts with cross-domain material into something neither side could have conceived
Then the cycle repeats. Each synthesis becomes the next round’s starting position.
The first round is calibration — the least insightful output. By Round 2–3, the dialectic has dug past the obvious framing into territory that neither you nor the Monks could have reached from the starting question.

​
The Seven Phases

You (Orchestrator)
├── Phase 1: Elenctic Interview + Research (you, with the user)
│   ├── Explain the process — set expectations
│   ├── Identify the user's belief burden and calibrate monk roles
│   ├── Ground the monks (research or deep interview, domain-dependent)
│   ├── Write context briefing document to file
│   └── Confirm framing with user — ask about gaps in coverage
├── Phase 2: Generate Electric Monk prompts (you) — reference briefing file
├── Phase 3: Spawn the Electric Monks (subagents, read briefing, BELIEVE fully)
│   ├── Decorrelation check: did monks genuinely diverge?
│   └── User checkpoint: gaps in evidence or comparison classes?
├── Phase 4: Determinate Negation (you — structural analysis, saved to file)
│   ├── Internal tensions — where does each monk undermine itself?
│   └── Boydian decomposition — shatter, find cross-domain connections
├── Phase 5: Sublation / Aufhebung (you — synthesis, saved to file)
│   └── Abduction test: does synthesis make contradiction predictable?
├── Phase 6: Validation (Monks A & B evaluate — elevated or defeated?)
│   ├── Adversarial check: would the hardest-hit monk accept this?
│   ├── Hostile Auditor: fresh agent attacks the synthesis
│   └── Refine: present improvements individually, incorporate accepted ones
└── Phase 7: Recursion — propose 2-4 directions, user chooses
    ├── Queue unexplored contradictions as orientation library
    └── Repeat from Phase 2 (or Phase 1 if new research needed)

​
Phase 1: Elenctic Interview + Research

Surface the real contradiction. The orchestrator interviews you Socratically — surfacing hidden assumptions, finding the deepest version of the contradiction, and identifying your belief burden. Then it researches the domain to ground both sides in specifics.
1

Explain the Process

Before anything else, the orchestrator tells you what’s about to happen and why. You need to understand the shape of what’s coming so you can be an active co-pilot, not a passive consumer.
2

Understand What You Want

Are you stress-testing one idea? Or do you have two positions in tension? This determines how the Monks are calibrated.
3

Socratic Probing

The orchestrator asks questions to surface: hidden assumptions, the deepest version of the contradiction, what domain type this is, what specific parameters of your mental model you want updated.
4

Identify Your Belief Burden

What are you stuck believing? Different cognitive styles produce different belief burdens. The Monks need to be calibrated to what you specifically can’t hold at full conviction.
5

Ground the Monks

For external domains: deep parallel research. For personal domains: deep interview (you are the source). For mixed domains: both.
6

Write Context Briefing

Everything gets synthesized into a neutral briefing document. Both Monks read this before writing.
7

Confirm with You

“Are there companies, thinkers, comparison classes, or evidence we’re missing?” This question consistently produces the highest-leverage interventions.
The interview surfaces what you’re actually wrestling with. The research ensures the downstream arguments are grounded in specifics, not generics.

​
Phase 2: Generate Electric Monk Prompts

Calibrate the belief assignments. The orchestrator crafts two prompts — one per Monk — calibrated to your specific belief burden. Each prompt includes:
  • Framing corrections that prevent the Monk from falling into the obvious, boring version of the argument
  • Anti-hedging instructions — the Monk’s one job is to believe fully
  • Targeted research directives for position-specific evidence
  • Argument structure requirements — ontological claim, opponent’s strongest case, diagnosis of failure, deeper principle, push to the extreme
Anti-hedging is a functional requirement, not a stylistic preference. A hedging Monk has failed its core function.

​
Phase 3: Spawn the Electric Monks

Two fully committed position essays. Two separate AI agents — each in a fresh, isolated context — write fully committed position essays. They don’t hedge. They don’t try to be balanced. Each one inhabits its position and makes the absolute strongest case.
Spawning them in separate sessions with no shared context produces structural decorrelation — genuinely different reasoning paths, not the same analysis with different conclusions bolted on.
After both complete, verify the monks actually diverged:
  • Do they cite different evidence?
  • Do they frame the problem using different conceptual vocabularies?
  • Do their unstated assumptions diverge?
  • Would a reader recognize these as genuinely different perspectives?
Before proceeding: “Is there a claim either monk makes that should be tested against evidence neither has considered?” Catching untested assumptions before synthesis prevents building on shaky foundations.

​
Phase 4: Determinate Negation

Find where each argument undermines itself. The orchestrator analyzes both essays to find:
  • Internal tensions: Where does each position’s own logic undermine itself? (self-sublation)
  • Shared assumptions: What both sides implicitly agree on without realizing it
  • Determinate negation: The specific way each position fails — not “it’s wrong” but “it fails in THIS way, which points toward THIS thing that’s missing”
  • The hidden question: The deeper question the contradiction is actually about — the question neither agent asked because they were both too committed to their answers

​
Boydian Decomposition (The Destructive Step)

Before attempting synthesis, shatter both positions into atomic parts:
1

Identify the Generic Space

The abstract relational structure both positions share. Both assume a particular unit of analysis, causal model, temporal frame.
2

List Atomic Components

Individual claims, mechanisms, evidence, assumptions, metaphors, principles — stripped of which agent said them. Create an unstructured collection.
3

Find Surprising Connections

What mechanisms from A illuminate evidence from B? What assumptions from B reframe principles from A?
4

Ask: What Material from Adjacent Domains?

What analogies, frameworks, or mechanisms from outside the original debate space could bind these parts into something neither agent could have conceived?
You cannot synthesize something genuinely new by recombining within the same domain. You must first destroy the existing wholes, scatter the parts, and look for cross-domain connections.

​
Phase 5: Sublation (Aufhebung)

Synthesize something neither side could reach. The orchestrator generates a synthesis that simultaneously:
  1. CANCELS both original positions as complete truths
  2. PRESERVES the genuine insight in each position
  3. ELEVATES to a new concept that transforms the question itself

​
What Sublation is NOT

❌ Division of Labor

“Use A for some cases and B for others”

❌ Compromise

“Build something that combines the best of A and B”

❌ Surrender

“It depends on the context”

❌ Policy

“A should do X more” — not reconceptualization

​
What Sublation IS

A reconceptualization of what the thing IS — something that, once stated, makes it hard to go back to thinking in the old terms

​
Abduction Test

The synthesis is an abductive hypothesis: What would make the contradiction between the Monks unsurprising? If someone heard your synthesis first, would they predict the approximate shape of both monks’ positions? If yes, you’ve found a genuine reframing.

​
Phase 6: Validation

Did the Monks feel elevated or defeated? Both Monks evaluate the synthesis. The critical question for each:
1

Does it preserve your core insight?

What specifically does it get right about what you were arguing?
2

Does it reveal a genuine limitation?

What were you missing? What assumption were you trapped in?
3

Do you feel ELEVATED or DEFEATED?

Elevated: “I see my position as partial truth within a larger truth.” Defeated: “My position was just dismissed or diluted.”
4

What's wrong with this synthesis?

Where is it weak, evasive, or just splitting the difference?

​
The Hostile Auditor

A separate agent whose sole mandate is to find what’s wrong with the synthesis. Not another monk with a position — it has no position. Its job is to be correct, not fair. The auditor looks for:
  • Hidden shared assumptions
  • Undercutting defeaters (broken inferential links)
  • Self-defeating structure
  • Compromise disguised as transcendence
  • The harder contradiction the synthesis misses
The auditor reads only the monks’ essays and the synthesis — NOT the orchestrator’s structural analysis. Fresh eyes, no inherited framing.

​
Refinement

After validation, the orchestrator:
  1. Digests the feedback
  2. Presents improvements one at a time (not all at once)
  3. Discusses each with the user
  4. Revises the synthesis with accepted improvements
  5. Identifies remaining tensions as recursion targets

​
Phase 7: Recursion

Where the real value lives.
The first round is calibration — the least insightful output. Each subsequent round gets sharper, more specific to your situation, and more likely to surface something genuinely new.
Each synthesis generates new contradictions. The orchestrator proposes 2–4 directions; you choose which to pursue. The process repeats — and each round gets sharper.

​
Sources for Recursive Contradictions

  • New contradictions identified in the synthesis itself
  • Convergent critiques from validation agents
  • Your interventions (often the most powerful source)
  • Unresolved tensions the synthesis names but doesn’t engage
  • The hostile auditor’s “harder contradiction”

​
The Dialectic Queue

A running list of proposed contradictions with their source round and status (explored, queued, deferred). This becomes a map of the dialectical territory — where you’ve been, where you could go, and what’s still open.

​
Why Recursion Matters

In test runs:
  • A React/Vue dialectic evolved from “corporate lab vs. auteur” → “the Layer Thesis” → “co-evolutionary arms race”
  • An institutional identity dialectic went through seven cycles, pulling in Gödel’s incompleteness theorem, Coasean transaction costs, and jurisprudential concepts that had nothing to do with the original question — but were essential by the time the dialectic reached them
Each round compresses understanding upward. The recursive rounds produce the most valuable insights.

​
The Full Cycle

Each recursive cycle is Boyd’s full cycle:
  1. Structure — The previous synthesis is a position
  2. Unstructure — A new contradiction shatters it
  3. Restructure — New material enters, fresh Monks believe new positions, synthesis elevates again
Seven cycles in an hour = seven reorientations with zero belief inertia. This is the engine of the skill.

Build docs developers (and LLMs) love

Get started for freeTalk to us