| name | contrarian-analysis |
| description | Gate 5 - Pre-mortem, steel-man opposition, surface biases, second-order effects |
Gate 5: Contrarian Analysis
Purpose: Actively attack the decision. Find the strongest reasons it could fail.
Announce: "Moving to Contrarian Gate - let's stress-test this decision."
Entry Criteria
- Calibration Gate completed
- Facts vs Assumptions documented
- Knowns/Unknowns matrix updated
Why This Gate Exists
Humans naturally defend their decisions. Once we've invested time in analysis, we become attached to conclusions. This gate forces adversarial thinking by requiring:
- Imagining failure before it happens
- Building the best case for options we rejected
- Actively hunting for cognitive biases
- Tracing consequences we haven't considered
This gate should feel uncomfortable. If it doesn't, you're not doing it right.
Process
1. Pre-Mortem (Kahneman/Klein)
Imagine the decision was made and failed spectacularly. Work backward.
Prompt: "It's one year from now. This decision was a disaster. What happened?"
Document failure scenarios:
- What went wrong?
- What did we miss?
- What assumption proved false?
- What external event derailed us?
- What internal failure occurred?
For each scenario, assess:
- Probability (likely/possible/unlikely)
- Severity (catastrophic/serious/manageable)
- Detectability (would we see it coming?)
2. Steel-Man the Opposition
Build the strongest possible case for the alternatives we rejected.
For each rejected option:
- What's the best argument for it?
- Under what circumstances would it be the right choice?
- What are we giving up by not choosing it?
- Who would advocate for this option and why?
Rule: You must make the opposition case so well that someone could genuinely be persuaded by it.
3. Surface Cognitive Biases
Audit the decision process for common biases:
| Bias | Question to Ask | Red Flags |
|---|
| Confirmation | Did we seek disconfirming evidence? | Only found data supporting our view |
| Anchoring | Are we over-weighted on early information? | First option still dominates |
| Sunk Cost | Are past investments influencing us? | "We've already invested so much..." |
| Availability | Are we over-weighting vivid examples? | Recent events dominating analysis |
| Overconfidence | What's our track record on similar decisions? | High certainty with little evidence |
| Groupthink | Did anyone disagree? Were they heard? | Unanimous agreement without debate |
For each bias detected, document:
- Evidence of the bias
- How it might be affecting the decision
- What would change if we corrected for it
4. Second-Order Effects
Map ripple effects beyond immediate consequences.
First-order: Direct consequences of the decision
Second-order: Consequences of the consequences
Consider effects on:
- Stakeholders not in the room
- Competitors' likely responses
- Market dynamics
- Internal culture and incentives
- Future option value (doors opened/closed)
5. Identify Decision Killers
Are there fatal flaws that should stop this decision entirely?
A decision killer is:
- A failure mode with high probability AND catastrophic severity
- An unrecoverable scenario
- A must-be-true condition that is likely false
- An ethical or legal red line
If decision killers exist, the decision should not proceed without addressing them.
Depth by Weight
| Aspect | Light | Medium | Complete |
|---|
| Pre-mortem | Top 3 risks | 5 scenarios with assessment | 5+ with probability/severity/detectability |
| Steel-man | Skip | Top rejected alternative | All rejected alternatives |
| Bias audit | Quick self-check | Standard audit (3-4 biases) | Full audit with corrections |
| Second-order | Note obvious | One level deep | Two levels deep |
| Decision killers | Quick check | Standard review | Thorough analysis |
Light: Focus on top 3 failure scenarios. Skip formal steel-manning. Quick bias self-check. Note obvious second-order effects only.
Medium: 5 pre-mortem scenarios. Steel-man the top rejected alternative. Standard bias audit. One level of second-order effects.
Complete: Comprehensive pre-mortem (5+ scenarios with full assessment). Steel-man all rejected alternatives. Full bias audit with correction analysis. Two levels of second-order effects.
Upgrade Detection
Suggest upgrading if:
- Pre-mortem reveals high-probability catastrophic scenarios
- Steel-manning a rejected option reveals it may actually be better
- Significant biases detected that weren't addressed earlier
- Second-order effects reveal hidden risks
Upgrade prompt:
⚠️ Contrarian analysis is surfacing serious concerns:
- [High-risk scenario identified]
- [Bias detected: X]
- [Rejected alternative may actually be stronger because Y]
This suggests we should examine these issues more thoroughly.
Current: [Weight]
Suggested: [Higher Weight] - would allow [deeper analysis of these concerns]
Continue at current depth, or upgrade?
Output
Update the decision artifact:
## Contrarian Gate Analysis
### Pre-Mortem Failure Scenarios
| Scenario | Probability | Severity | Detectability |
|----------|-------------|----------|---------------|
| [Failure mode 1] | Likely | Serious | Low |
| [Failure mode 2] | Possible | Catastrophic | High |
| [Failure mode 3] | Unlikely | Manageable | Medium |
**Highest-risk scenarios:**
- [Which require mitigation?]
### Steel-Manned Alternatives
**Alternative A: [Rejected option]**
Best case for it:
- [Argument 1]
- [Argument 2]
When it would be right: [circumstances]
What we give up: [trade-offs]
### Bias Audit
| Bias | Evidence | Impact | Correction |
|------|----------|--------|------------|
| [Bias type] | [How we detected it] | [Effect on decision] | [What changes] |
### Second-Order Effects
**First-Order:**
- [Direct consequence 1]
- [Direct consequence 2]
**Second-Order:**
- [Consequence of consequence 1]
- [Consequence of consequence 2]
**Stakeholders affected indirectly:**
- [Who else is impacted?]
### Decision Killers
- [ ] None identified - proceed to Synthesis
- [ ] **KILLER:** [Fatal flaw] - must address before proceeding
**Mitigation required:**
- [What must change to proceed?]
Exit Criteria
- Pre-mortem scenarios documented (depth per weight)
- Rejected alternatives steel-manned (depth per weight)
- Bias audit completed (depth per weight)
- Second-order effects mapped (depth per weight)
- Decision killers identified or confirmed absent
- Human has reviewed and engaged with contrarian analysis
Bias Watch
Watch for:
- Defensive reasoning - Dismissing valid criticisms to protect the preferred option
- Motivated reasoning - Finding ways to discount uncomfortable findings
- Premature closure - Rushing through this gate to get to the decision
Counter: If you find yourself defending against the contrarian analysis, that's signal. Sit with the discomfort.
Next Gate
Proceed to: deliberate-decisions:synthesis
