thinking-model-combination

# Model Combination **Core rule:** Combine only when each model answers a different named question. Cap at three, name the conflict rule before applying, then synthesize once. ## When to Use - One model already applied (or clearly primary) still leaves a material blind spot that another mechanism covers. - Problem spans domains (e.g. risk + choice + system structure) and stakes justify multi-lens work. - You need independent checks, not confirmation of the same conclusion. - You can name a distinct role per model before running them. ## When NOT to Use - A single catalog skill fully answers the unknown — apply that skill alone. - Routine, local, or fully reversible work where multi-lens cost exceeds upside. - You cannot state what unique question each extra model answers (checkbox / model soup). - Near-duplicate mechanisms (two diagnosis skills that ask the same causal question). - Time budget cannot support genuine synthesis — prefer one honest model over contradictory partials. ## Procedure 1. **State the unknown and the gap.** Write the decision question. If one model already covers it, stop and use that model alone. Otherwise name the specific blind spot (e.g. "failure modes unexamined", "displaced alternative unknown"). 2. **Pick 2–3 models with distinct roles.** For each, record: model id, role (narrow / decide / stress / cost / …), and the unique question it answers. Drop any model that only rephrases another. Prefer sequential pipeline (narrow → stress → decide) over parallel unless independent concurrent checks are required. 3. **Lock the relation and conflict rule before applying.** Choose pattern: sequential, parallel, nested (macro→meso→micro), or adversarial (for/against). Predeclare the tiebreaker (e.g. reversibility class, evidence strength, ruin constraint, primary decision owner). Incompatible worldviews run sequential or adversarial — never blended. 4. **Apply each model fully for its role only.** Capture one key insight per model plus what only that model revealed. Do not re-run a model that adds no new insight. 5. **Synthesize once.** Record convergence, divergence, how the conflict rule resolves divergence, and a single combined recommendation with residual uncertainty. Stop when the recommendation is decision-ready or when further models would only reconfirm. ## Output ```text problem: <decision question> gap: <named blind spot justifying combination> pattern: sequential | parallel | nested | adversarial models: - id: <skill> role: <named job> unique_question: <what only this answers> insight: <key finding> conflict_rule: <predeclared tiebreaker> convergence: <where models agree> divergence: <where they conflict + resolution> recommendation: <single decision-ready answer> stop_reason: gap_closed | single_model_suffices | budget ``` ## Verification - **Falsify / stop:** Remove a model only when it changes none of the recommendation, supporting evidence, confidence, residual risks, or mitigations; then re-synthesize with fewer. If no predeclared conflict rule exists and models disagree, do not average — pick one primary model or stop and re-route. - **Over-application guard:** Never exceed three models. Never add a model for thoroughness theater. If the first adequate single model already closes the gap, combination is wrong for this task.