| name | using-deliberate-decisions |
| description | Introduction to the Deliberate Decisions framework - use when helping humans make high-stakes decisions |
Deliberate Decisions
A framework for systematic decision-making. Helps surface hidden assumptions, separate facts from speculation, and make rigorous choices.
When to Use
Use this framework when a human faces a high-stakes decision:
- Strategic business decisions (M&A, market entry, major investments)
- Career decisions (job changes, pivots)
- Technical decisions (architecture, vendor selection)
- Any decision where being wrong is costly
Decision Weight
Not all decisions need the same rigor. After understanding the decision in Thesis Gate, suggest appropriate depth:
| Weight | Time | Best For |
|---|
| Light | <10 min | Reversible decisions, domain expertise, lower stakes |
| Medium | 30-40 min | Moderate stakes, some uncertainty |
| Complete | 60-90 min | High stakes, irreversible, significant uncertainty |
Weight affects depth, not gates. All 7 gates always run.
See deliberate-decisions:weight-classification for detailed guidance.
The Seven Gates
Every decision passes through seven mandatory gates. You can exit early, but you cannot skip gates.
| Gate | Name | Purpose |
|---|
| 1 | Thesis | Articulate what we're deciding and why |
| 2 | Landscape | Map Decision Points, Intersections, alternatives |
| 3 | Research | Gather evidence, find reference class |
| 4 | Calibration | Separate Facts from Assumptions, update Knowns/Unknowns |
| 5 | Contrarian | Pre-mortem, steel-man opposition, surface biases |
| 6 | Synthesis | Must-Be-True Conditions, Exit Criteria, recommendation |
| 7 | Decision | Human chooses with full picture |
Starting a Decision
Use /deliberate-decisions:decide to begin. Options:
/deliberate-decisions:decide - AI suggests weight after Thesis gate/deliberate-decisions:decide --light - Start at Light depth/deliberate-decisions:decide --medium - Start at Medium depth/deliberate-decisions:decide --complete - Start at Complete depth
Weight can be upgraded mid-process but not downgraded.
Core Principles
- Decisions fail in small assumptions - Surface the hidden structure
- Facts vs Assumptions - Rigorous separation of verifiable from speculative
- System 2 thinking - Gates enforce deliberate, not intuitive, analysis
- Adversarial by default - Pre-mortem and steel-manning are required
- One question at a time - AI builds the model incrementally
- Right-sized rigor - Match depth to stakes with weight classification
Key Concepts
- Decision Points: Individual choices or assumptions that influence outcome
- Decision Intersections: How Decision Points depend on or conflict with each other
- Must-Be-True Conditions: Assumptions that must hold for the decision to succeed
- Exit Criteria: Pre-committed tripwires to reverse the decision if needed
- Weight: Light/Medium/Complete depth classification
Upgrading Mid-Process
User can upgrade weight anytime by saying "let's go deeper."
AI should suggest upgrading when detecting:
- Emerging complexity beyond current weight
- Low confidence on critical assumptions
- Surprising research findings
- Stakes higher than initially framed
Already-completed gates stay as-is. Only remaining gates run at higher depth.
Commands
| Command | Action |
|---|
/deliberate-decisions:decide | Start new decision from Thesis gate |
/deliberate-decisions:decide --light | Start at Light depth |
/deliberate-decisions:decide --medium | Start at Medium depth |
/deliberate-decisions:decide --complete | Start at Complete depth |
/deliberate-decisions:advocate | Generate compelling case FOR |
/deliberate-decisions:detract | Generate compelling case AGAINST |
/deliberate-decisions:review-decision | Review past decision against outcomes |
/deliberate-decisions:resume-decision | Continue paused decision |
