Implement

Who Should Be in the Triangle?

For executive decisions:

• CEO/Executive Director
• Operations lead
• Finance/Strategy lead

For board decisions:

• Board chair
• One subject-matter expert
• One independent voice (not aligned with either)

For project decisions:

• Project lead
• Technical expert
• Stakeholder representative

What to Avoid

• Dyads — Two people collapse into conflict without a tiebreaker
• Large committees — Five+ people diffuse accountability
• Homogeneous triangles — Three people with identical expertise miss blind spots

Triangle Design Checklist

• Three people, distinct perspectives
• Clear decision domain (what this triangle decides)
• Meeting rhythm established (weekly? per-decision?)
• Decision log created (shared, accessible)
• AI advisory access configured (if applicable)

The Consensus-First Voting Rule

1. Facilitator aims for genuine consensus — extract information and disagreement from everyone. Not agreeing for the sake of it.
2. When consensus fails, 2/3 carries — A simple majority within the triangle
3. If facilitator is in the minority — they must represent the majority. Not override it.
4. Dissent is recorded — The minority view is logged with the decision

Facilitator Selection

1. Volunteer — Anyone who wants to be facilitator volunteers
2. Everyone votes — No self-voting allowed
3. Tiebreaker — External authority decides, or surrender to chance (coin flip)
4. Recorded — When tiebreaker used, it is logged publicly

The Decision Map

Decisions are not flat. Some depend on others. The Map makes the sequence visible.

1. Nodes — Each decision is a node on the Map
2. Arrows — Show dependencies between decisions
3. No loops — Forward only. Circular dependencies are flagged by AI
4. Parallel — Some decisions can happen simultaneously

Lock & Forward

1. Complete all four steps → decision is locked
2. No going back — The team moves to the next decision
3. Challenge path — Two non-originating triangles can reopen a lock
4. Authorized override — CEO or equivalent managing the system can reopen unilaterally, with full public logging

Decision Requests

When Team B needs a decision from Team A before they can proceed:

1. Team B sends request — becomes a node on Team A's Map
2. Timer starts — adjustable in dashboard
3. Team A completes four steps → locked → Team B notified
4. Team A doesn't respond → timeout → "No Decision" lock → Team B notified

The Challenge Rule

1. Two non-originating triangles must agree to challenge
2. Decider is excluded from the challenge vote
3. Biblical pattern — by the mouth of two or three witnesses
4. Fractal — same threshold at every scale

Dynamic Authority Protocol

Authority shifts based on:

1. Proximity to facts — Who is closest to the situation?
2. Track record — Whose prior judgments proved accurate?
3. Vindicated dissent — Whose past objections were proven right?
4. Facilitator assignment — The person chosen to run the four-step process for this decision

Example: In a safety crisis, the field coordinator leads the decision — not the CFO, not the CEO. The triangle decides who has authority per decision.

AI Dependency Tool Protocol

1. AI listens during meetings — Flags new decisions as they emerge
2. Suggests dependencies — What the new decision depends on, what depends on it
3. Auto-sequences — Suggests where the new decision fits in the Map
4. Flags conflicts — Identifies circular dependencies and cross-team blockers
5. Human confirmation required — AI advises, humans approve. Always.

Best Practices

• Record dissent verbatim — Don't summarize. Preserve the actual reasoning.
• Set follow-up dates — Every decision gets reviewed in 3-6 months.
• Track who was right — This builds the track record for dynamic authority.
• Make it accessible — New triangle members should read past decisions during onboarding.

Monthly Triangle Review

Once per month, the triangle reviews:

• Decisions made — What did we decide?
• Dissent patterns — Who dissented? Were they right?
• Authority shifts — Did the right person lead each decision?
• AI accuracy — When did AI recommend correctly? When did humans overrule wisely?

Quarterly Expansion

After one quarter of successful triangle operation:

• Document what worked — What decision types fit the triangle model?
• Identify gaps — What decisions are still solo? Why?
• Form new triangles — Expand to adjacent decision domains
• Train new members — Bring more people into the Triangle framework