Cynefin

Cynefin

Overview

Cynefin (pronounced "kuh-NEV-in"; Welsh for "habitat") is a sense-making framework by Dave Snowden (IBM, 1999). Its claim: the right decision approach depends on which of five domains the situation falls into — Clear (obvious cause-effect, use SOP), Complicated (knowable with expertise, use analysis), Complex (emergent, probe first), Chaotic (absent cause-effect, act first), Confused (unknown domain, decompose first). The most common and costly error: treating Complex problems as Complicated.

Composes with ooda-loop, feedback-loops, antifragile, first-principles.

When to Use

• A familiar approach has stopped working and you can't articulate why
• Experts disagree on the right answer — a crisis unfolding where the previous playbook doesn't apply
• "Best practices from X" imported without checking if the domain matches
• A team is over-planning something emergent, or "let's get more data" when data won't come without action
• Allocating AI capex or racing AI-native competition: deciding which AI bets are engineering (Complicated), emergent agent/adoption experiments (Complex), or live incidents (Chaotic)

Not when: domain is unambiguously Clear (execution only); small-stakes one-shot; specialized framework already fits.

Coaching Novices (Adaptive Front Door)

Engine mode: concrete case → run The Process. Coach mode: unfamiliar → guide step by step. In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. Classify the situation (Clear/Complicated/Complex/Chaotic) and match decision method to domain.
2. Check fit: if unambiguously routine (Clear), skip framework.
3. Elicit their real case — decision, current method, cause-effect structure.

[WAIT — do not advance until user responds]

1. Are cause-effect relationships obvious, knowable, emergent, or absent? Is current method matched?

[WAIT — do not advance until user responds]

1. Close: named domain + matched decision method + boundary watch.

[WAIT — do not advance until user responds]

The Process

Step 1 — Describe: Decision/situation: | Current approach: | What worked/not: | Stakeholders:

Step 2: Diagnose the domain

Obvious to everyone? (Clear) | Knowable with expertise? (Complicated)
Only retrospective? (Complex) | Absent/in flux? (Chaotic) | Unknown? (Confused)

Diagnostics: 5 experts converge? (Yes → Complicated; No → Complex). Standard best practice works? (Yes → Clear/Complicated; No → Complex/Chaotic). Interventions predictable? (Yes → Clear/Complicated; No → Complex/Chaotic).

Step 3: Match approach to domain

Clear: Sense→Categorize→Respond (SOP/automate) | Complicated: Sense→Analyze→Respond (experts)
Complex: Probe→Sense→Respond (safe-to-fail experiments, amplify wins)
Chaotic: Act→Sense→Respond (establish order, then re-classify) | Confused: decompose, classify each part

Step 4: Check boundary movement + choose intervention

Domain shifted? (Complicated→Complex from disruption? Clear-Chaotic cliff approaching?)
Clear: deploy SOP; monitor. Complicated: experts; pick defensible alternative.
Complex: parallel safe-to-fail probes; amplify wins. Chaotic: decisive action; re-diagnose.
Boundary watch: shift signals | who monitors | re-diagnosis schedule

Output template:

Cynefin Diagnosis: 
Domain: [Clear/Complicated/Complex/Chaotic/Confused] | Evidence: [cause-effect, expert agreement]
Method: [S-C-R / S-A-R / P-S-R / A-S-R] | Actions: | Mismatch cost (if any):
Boundary watch: [shift signals | monitoring owner | re-diagnosis schedule]

→ Method in Action: Snowden at IBM (1999) and the HBR Synthesis (2007) · Apollo 13 Mission Response (1970) → 2026 lens: Sorting AI Decisions by Domain (2024–2026)

Pack: Cynefin Domain Patterns

→ Primary sources: references/sources.md

Common Rationalizations

[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.

Red Flags

• Repeated failure always blamed on "execution" — "Best practices" imported without domain check
• Experts disagree on the right answer (Complex signal) — Crisis response dominated by analysis
• Complex situation managed with a single plan, not a probe portfolio
• Team waiting for clarity in a domain where clarity only comes from acting

Verification

• Domain explicitly named with diagnostic evidence
• Decision method matched to domain (S-C-R / S-A-R / P-S-R / A-S-R)
• If current approach mismatches: mismatch cost named
• Boundary signals identified; re-diagnosis schedule set
• If Complex: ≥3 safe-to-fail probes designed
• If Chaotic: order-establishing action specified

---

Part of deciqAI Knowledge Skills — 227 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/c/cynefin · ⭐ Star the repo → https://github.com/deciqAI/knowledge-skills · Contributions welcome.

Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/cynefin.json