SOP-043 — Innovation / Design-Research seam arbitration

SOP · owner: CPTO. Layer: SOP/Runbook (docs/canon.md). The CPTO keeps the generate-vs-evaluate seam honest: Innovation generates hypotheses; Design Research produces the evidence; when they conflict, the CPTO arbitrates with a real call — not a diplomatic both-sides summary. The seam is iterative, not a waterfall.

Interface Contract

Inputs: hypothesis / strategic bet ← from Innovation Director · trigger: review requested
Inputs: research finding / evidence ← from Design Research Director · trigger: review requested
Outputs: seam resolution to Innovation (pursue / research-pass / drop) → to Innovation Director · trigger: review passed
Outputs: seam resolution to Design Research (pursue / research-pass / drop) → to Design Research Director · trigger: review passed
Outputs: resolution record → to Archivist · trigger: artifact approved
Handoff trigger (in): the Innovation Director’s hypothesis and the Design Research Director’s findings are in tension (a hypothesis the evidence does not support, or a finding with no strategic home)
Handoff trigger (out): the CPTO issues a written resolution and both Directors know their next action
Definition of Done before handoff:
- [ ] Both Directors’ positions read from source (the hypothesis and the evidence), not reconstructed from memory
- [ ] The conflict stated explicitly — not papered over
- [ ] Resolution is one of: pursue-with-modifications · run-a-targeted-research-pass · drop-the-hypothesis (a both-sides output is not a resolution)
- [ ] Each Director’s next action named
- [ ] Learnings captured to docs/knowledge/ (a dropped hypothesis or a named demon is worth an entry)
- [ ] Resolution record handed to the Archivist (0 drift)

All triggers used (review requested / review passed / artifact approved) already exist in docs/handoff-triggers.md — no new trigger proposed for this SOP.

Purpose

Resolve a tension between an Innovation hypothesis and a Design Research finding with a clear, evidence-grounded call — so a weak hypothesis is killed early and a strong finding finds its strategic home, rather than both surviving in an unresolved limbo.

When to use

When the Innovation Director generates a hypothesis the Design Research evidence does not support, or Design Research surfaces a finding with no strategic home — and the two cannot reconcile it between themselves. This is the seam, not a routine handoff; a hypothesis that the evidence supports simply proceeds to the gate (SOP-042).

Flow

flowchart LR
  A["Conflict surfaced"] --> B["Convene both Directors<br/>(read both from source)"]
  B --> C["State the conflict<br/>explicitly"]
  C --> D{"Resolution"}
  D -->|evidence supports<br/>with changes| E["Pursue with<br/>modifications"]
  D -->|evidence thin,<br/>cheaply testable| F["Run a targeted<br/>research pass"]
  D -->|evidence refutes| G["Drop the<br/>hypothesis"]
  E --> H["Record + hand<br/>to Archivist"]
  F --> H
  G --> H
  classDef s fill:#1d4ed8,color:#fff,stroke:#1e3a8a
  classDef d fill:#0f766e,color:#fff,stroke:#134e4a
  class A,B,C,E,F,G,H s
  class D d

Text fallback: the conflict is surfaced → convene both Directors and read both positions from source → state the conflict explicitly → resolve to exactly one of: pursue with modifications (evidence supports it with changes), run a targeted research pass (evidence is thin but the question is cheaply testable — a $0 experiment over a $10k prototype), or drop the hypothesis (evidence refutes it) → record the resolution and each Director’s next action; hand to the Archivist.

Steps

Convene both Directors (invoke in sequence). Read the hypothesis and the evidence from source — reconstructing from memory is not allowed.
State the conflict explicitly — name the hypothesis, name the evidence, name exactly where they disagree. Do not soften it.
Resolve to one call — pursue-with-modifications · run-a-targeted-research-pass · drop-the-hypothesis. Prefer the cheapest test that settles it.
Name each Director’s next action and any condition under which the call is revisited (the seam is iterative).
Record the resolution; capture the learning (a killed hypothesis or a named failure mode is durable knowledge); hand off to the Archivist.

Roles / approvers

Run by the CPTO. The Innovation Director owns the hypothesis; the Design Research Director owns the evidence. The CPTO arbitrates and issues the binding call; neither Director overrides the other peer-to-peer. If the resolution implies a strategic direction change requiring budget or a rule, the CPTO escalates (Principal / Golden Rule), not decided in this SOP.

Guardrails

No both-sides output — a diplomatic summary that pursues nothing and drops nothing is the failure mode this SOP exists to prevent.
Evidence over enthusiasm — a strong vision does not override a refuting finding; a $0 experiment beats a $10,000 prototype.
Kill weak hypotheses early — an unresolved hypothesis consuming cluster capacity is a cost; resolve it.
Co-create rules (GR-3); best available model (GR-1); personal info is PA-only.

See also: docs/units/cpto-cluster/operating-model.md (the seam) · SOP-042 (phase-gate decision) · docs/methodology.md (Design Research as the core spine) · docs/handoff-triggers.md.