Initial commit

2025-11-30 08:38:26 +08:00
commit 41d9f6b189
304 changed files with 98322 additions and 0 deletions
--- a/skills/forecast-premortem/SKILL.md
+++ b/skills/forecast-premortem/SKILL.md
@@ -0,0 +1,462 @@
+---
+name: forecast-premortem
+description: Use to stress-test predictions by assuming they failed and working backward to identify why. Invoke when confidence is high (>80% or <20%), need to identify tail risks and unknown unknowns, or want to widen overconfident intervals. Use when user mentions premortem, backcasting, what could go wrong, stress test, or black swans.
+---
+
+# Forecast Pre-Mortem
+
+## Table of Contents
+- [What is a Forecast Pre-Mortem?](#what-is-a-forecast-pre-mortem)
+- [When to Use This Skill](#when-to-use-this-skill)
+- [Interactive Menu](#interactive-menu)
+- [Quick Reference](#quick-reference)
+- [Resource Files](#resource-files)
+
+---
+
+## What is a Forecast Pre-Mortem?
+
+A **forecast pre-mortem** is a stress-testing technique where you assume your prediction has already failed and work backward to construct the history of how it failed. This reveals blind spots, tail risks, and overconfidence.
+
+**Core Principle:** Invert the problem. Don't ask "Will this succeed?" Ask "It has failed - why?"
+
+**Why It Matters:**
+- Defeats overconfidence by forcing you to imagine failure
+- Identifies specific failure modes you hadn't considered
+- Transforms vague doubt into concrete risk variables
+- Widens confidence intervals appropriately
+- Surfaces "unknown unknowns"
+
+**Origin:** Gary Klein's "premortem" technique, adapted for probabilistic forecasting
+
+---
+
+## When to Use This Skill
+
+Use this skill when:
+- **High confidence** (>80% or <20%) - Most likely to be overconfident
+- **Feeling certain** - Certainty is a red flag in forecasting
+- **Prediction is important** - Stakes are high, need robustness
+- **After inside view analysis** - Used specific details, might have missed big picture
+- **Before finalizing forecast** - Last check before committing
+
+Do NOT use when:
+- Confidence already low (~50%) - You're already uncertain
+- Trivial low-stakes prediction - Not worth the time
+- Pure base rate forecasting - Premortem is for inside view adjustments
+
+---
+
+## Interactive Menu
+
+**What would you like to do?**
+
+### Core Workflows
+
+**1. [Run a Failure Premortem](#1-run-a-failure-premortem)** - Assume prediction failed, explain why
+**2. [Run a Success Premortem](#2-run-a-success-premortem)** - For pessimistic predictions (<20%)
+**3. [Dragonfly Eye Perspective](#3-dragonfly-eye-perspective)** - View failure through multiple lenses
+**4. [Identify Tail Risks](#4-identify-tail-risks)** - Find black swans and unknown unknowns
+**5. [Adjust Confidence Intervals](#5-adjust-confidence-intervals)** - Quantify the adjustment
+**6. [Learn the Framework](#6-learn-the-framework)** - Deep dive into methodology
+**7. Exit** - Return to main forecasting workflow
+
+---
+
+## 1. Run a Failure Premortem
+
+**Let's stress-test your prediction by imagining it has failed.**
+
+```
+Failure Premortem Progress:
+- [ ] Step 1: State the prediction and current confidence
+- [ ] Step 2: Time travel to failure
+- [ ] Step 3: Write the history of failure
+- [ ] Step 4: Identify concrete failure modes
+- [ ] Step 5: Assess plausibility and adjust
+```
+
+### Step 1: State the prediction and current confidence
+
+**Tell me:**
+1. What are you predicting?
+2. What's your current probability?
+3. What's your confidence interval?
+
+**Example:** "This startup will reach $10M ARR within 2 years" - Probability: 75%, CI: 60-85%
+
+### Step 2: Time travel to failure
+
+**The Crystal Ball Exercise:**
+
+Jump forward to the resolution date. **It is now [resolution date]. The event did NOT happen.** This is a certainty. Do not argue with it.
+
+**How does it feel?** Surprising? Expected? Shocking? This emotional response tells you about your true confidence.
+
+### Step 3: Write the history of failure
+
+**Backcasting Narrative:** Starting from the failure point, work backward in time. Write the story of how we got here.
+
+**Prompts:**
+- "The headlines that led to this were..."
+- "The first sign of trouble was when..."
+- "In retrospect, we should have known because..."
+- "The critical mistake was..."
+
+**Frameworks to consider:**
+- **Internal friction:** Team burned out, co-founders fought, execution failed
+- **External shocks:** Regulation changed, competitor launched, market shifted
+- **Structural flaws:** Unit economics didn't work, market too small, tech didn't scale
+- **Black swans:** Pandemic, war, financial crisis, unexpected disruption
+
+See [Failure Mode Taxonomy](resources/failure-mode-taxonomy.md) for comprehensive categories.
+
+### Step 4: Identify concrete failure modes
+
+**Extract specific, actionable failure causes from your narrative.**
+
+For each failure mode: (1) What happened, (2) Why it caused failure, (3) How likely it is, (4) Early warning signals
+
+**Example:**
+| Failure Mode | Mechanism | Likelihood | Warning Signals |
+|--------------|-----------|------------|-----------------|
+| Key engineer quit | Lost technical leadership, delayed product | 15% | Declining code commits, complaints |
+| Competitor launched free tier | Destroyed unit economics | 20% | Hiring spree, beta leaks |
+| Regulation passed | Made business model illegal | 5% | Proposed legislation, lobbying |
+
+### Step 5: Assess plausibility and adjust
+
+**The Plausibility Test:**
+
+Ask yourself:
+- **How easy was it to write the failure narrative?**
+  - Very easy → Drop confidence by 15-30%
+  - Very hard, felt absurd → Confidence was appropriate
+- **How many plausible failure modes did you identify?**
+  - 5+ modes each >5% likely → Too much uncertainty for high confidence
+  - 1-2 modes, low likelihood → Confidence can stay high
+- **Did you discover any "unknown unknowns"?**
+  - Yes, multiple → Widen confidence intervals by 20%
+  - No, all known risks → Confidence appropriate
+
+**Quantitative Method:** Sum the probabilities of failure modes:
+```
+P(failure) = P(mode_1) + P(mode_2) + ... + P(mode_n)
+```
+
+If this sum is greater than `1 - your_current_probability`, your probability is too high.
+
+**Example:** Current success: 75% (implied failure: 25%), Sum of failure modes: 40%
+**Conclusion:** Underestimating failure risk by 15%, **Adjusted:** 60% success
+
+**Next:** Return to [menu](#interactive-menu) or document findings
+
+---
+
+## 2. Run a Success Premortem
+
+**For pessimistic predictions - assume the unlikely success happened.**
+
+```
+Success Premortem Progress:
+- [ ] Step 1: State pessimistic prediction (<20%)
+- [ ] Step 2: Time travel to success
+- [ ] Step 3: Write the history of success
+- [ ] Step 4: Identify how you could be wrong
+- [ ] Step 5: Assess and adjust upward if needed
+```
+
+### Step 1: State pessimistic prediction
+
+**Tell me:** (1) What low-probability event are you predicting? (2) Why is your confidence so low?
+
+**Example:** "Fusion energy will be commercialized by 2030" - Probability: 10%, Reasoning: Technical challenges too great
+
+### Step 2: Time travel to success
+
+**It is now 2030. Fusion energy is commercially available.** This happened. It's real. How?
+
+### Step 3: Write the history of success
+
+**Backcasting the unlikely:** What had to happen for this to occur?
+- "The breakthrough came when..."
+- "We were wrong about [assumption] because..."
+- "The key enabler was..."
+- "In retrospect, we underestimated..."
+
+### Step 4: Identify how you could be wrong
+
+**Challenge your pessimism:**
+- Are you anchoring too heavily on current constraints?
+- Are you underestimating exponential progress?
+- Are you ignoring parallel approaches?
+- Are you biased by past failures?
+
+### Step 5: Assess and adjust upward if needed
+
+If success narrative was surprisingly plausible, increase probability.
+
+**Next:** Return to [menu](#interactive-menu)
+
+---
+
+## 3. Dragonfly Eye Perspective
+
+**View the failure through multiple conflicting perspectives.**
+
+The dragonfly has compound eyes that see from many angles simultaneously. We simulate this by adopting radically different viewpoints.
+
+```
+Dragonfly Eye Progress:
+- [ ] Step 1: The Skeptic (why this will definitely fail)
+- [ ] Step 2: The Fanatic (why failure is impossible)
+- [ ] Step 3: The Disinterested Observer (neutral analysis)
+- [ ] Step 4: Synthesize perspectives
+- [ ] Step 5: Extract robust failure modes
+```
+
+### Step 1: The Skeptic
+
+**Channel the harshest critic.** You are a short-seller, a competitor, a pessimist. Why will this DEFINITELY fail?
+
+**Be extreme:** Assume worst case, highlight every flaw, no charity, no benefit of doubt
+
+**Output:** List of failure reasons from skeptical view
+
+### Step 2: The Fanatic
+
+**Channel the strongest believer.** You are the founder's mother, a zealot, an optimist. Why is failure IMPOSSIBLE?
+
+**Be extreme:** Assume best case, highlight every strength, maximum charity and optimism
+
+**Output:** List of success reasons from optimistic view
+
+### Step 3: The Disinterested Observer
+
+**Channel a neutral analyst.** You have no stake in the outcome. You're running a simulation, analyzing data dispassionately.
+
+**Be analytical:** No emotional investment, pure statistical reasoning, reference class thinking
+
+**Output:** Balanced probability estimate with reasoning
+
+### Step 4: Synthesize perspectives
+
+**Find the overlap:** Which failure modes appeared in ALL THREE perspectives?
+- Skeptic mentioned it
+- Even fanatic couldn't dismiss it
+- Observer identified it statistically
+
+**These are your robust failure modes** - the ones most likely to actually happen.
+
+### Step 5: Extract robust failure modes
+
+**The synthesis:**
+
+| Failure Mode | Skeptic | Fanatic | Observer | Robust? |
+|--------------|---------|---------|----------|---------|
+| Market too small | Definitely | Debatable | Base rate suggests yes | YES |
+| Execution risk | Definitely | No way | 50/50 | Maybe |
+| Tech won't scale | Definitely | Already solved | Unknown | Investigate |
+
+Focus adjustment on the **robust** failures that survived all perspectives.
+
+**Next:** Return to [menu](#interactive-menu)
+
+---
+
+## 4. Identify Tail Risks
+
+**Find the black swans and unknown unknowns.**
+
+```
+Tail Risk Identification Progress:
+- [ ] Step 1: Define what counts as "tail risk"
+- [ ] Step 2: Systematic enumeration
+- [ ] Step 3: Impact × Probability matrix
+- [ ] Step 4: Set kill criteria
+- [ ] Step 5: Monitor signposts
+```
+
+### Step 1: Define what counts as "tail risk"
+
+**Criteria:** Low probability (<5%), High impact (would completely change outcome), Outside normal planning, Often exogenous shocks
+
+**Examples:** Pandemic, war, financial crisis, regulatory ban, key person death, natural disaster, technological disruption
+
+### Step 2: Systematic enumeration
+
+**Use the PESTLE framework for comprehensive coverage:**
+
+- **Political:** Elections, coups, policy changes, geopolitical shifts
+- **Economic:** Recession, inflation, currency crisis, market crash
+- **Social:** Cultural shifts, demographic changes, social movements
+- **Technological:** Breakthrough inventions, disruptions, cyber attacks
+- **Legal:** New regulations, lawsuits, IP challenges, compliance changes
+- **Environmental:** Climate events, pandemics, natural disasters
+
+For each category, ask: "What low-probability event would kill this prediction?"
+
+See [Failure Mode Taxonomy](resources/failure-mode-taxonomy.md) for detailed categories.
+
+### Step 3: Impact × Probability matrix
+
+**Plot your tail risks:**
+
+```
+High Impact
+│
+│  [Pandemic]        [Key Founder Dies]
+│
+│
+│  [Recession]       [Competitor Emerges]
+│
+└─────────────────────────────────────→ Probability
+  Low                              High
+```
+
+**Focus on:** High impact, even if very low probability
+
+### Step 4: Set kill criteria
+
+**For each major tail risk, define the "kill criterion":**
+
+**Format:** "If [event X] happens, probability drops to [Y]%"
+
+**Examples:**
+- "If FDA rejects our drug, probability drops to 5%"
+- "If key engineer quits, probability drops to 30%"
+- "If competitor launches free tier, probability drops to 20%"
+- "If regulation passes, probability drops to 0%"
+
+**Why this matters:** You now have clear indicators to watch
+
+### Step 5: Monitor signposts
+
+**For each kill criterion, identify early warning signals:**
+
+| Kill Criterion | Warning Signals | Check Frequency |
+|----------------|----------------|-----------------|
+| FDA rejection | Phase 2 trial results, FDA feedback | Monthly |
+| Engineer quit | Code velocity, satisfaction surveys | Weekly |
+| Competitor launch | Hiring spree, beta leaks, patents | Monthly |
+| Regulation | Proposed bills, lobbying, hearings | Quarterly |
+
+**Setup monitoring:** Calendar reminders, news alerts, automated tracking
+
+**Next:** Return to [menu](#interactive-menu)
+
+---
+
+## 5. Adjust Confidence Intervals
+
+**Quantify how much the premortem should change your bounds.**
+
+```
+Confidence Interval Adjustment Progress:
+- [ ] Step 1: State current CI
+- [ ] Step 2: Evaluate premortem findings
+- [ ] Step 3: Calculate width adjustment
+- [ ] Step 4: Set new bounds
+- [ ] Step 5: Document reasoning
+```
+
+### Step 1: State current CI
+
+**Current confidence interval:** Lower bound: __%, Upper bound: __%, Width: ___ percentage points
+
+### Step 2: Evaluate premortem findings
+
+**Score your premortem on these dimensions (1-5 each):**
+
+1. **Narrative plausibility** - 1 = Failure felt absurd, 5 = Failure felt inevitable
+2. **Number of failure modes** - 1 = Only 1-2 unlikely modes, 5 = 5+ plausible modes
+3. **Unknown unknowns discovered** - 1 = No surprises, all known, 5 = Many blind spots revealed
+4. **Dragonfly synthesis** - 1 = Perspectives diverged completely, 5 = All agreed on failure modes
+
+**Total score:** __ / 20
+
+### Step 3: Calculate width adjustment
+
+**Adjustment formula:**
+
+```
+Width multiplier = 1 + (Score / 20)
+```
+
+**Examples:**
+- Score = 4/20 → Multiplier = 1.2 → Widen CI by 20%
+- Score = 10/20 → Multiplier = 1.5 → Widen CI by 50%
+- Score = 16/20 → Multiplier = 1.8 → Widen CI by 80%
+
+**Current width:** ___ points, **Adjusted width:** Current × Multiplier = ___ points
+
+### Step 4: Set new bounds
+
+**Method: Symmetric widening around current estimate**
+
+```
+New lower = Current estimate - (Adjusted width / 2)
+New upper = Current estimate + (Adjusted width / 2)
+```
+
+**Example:** Current: 70%, CI: 60-80% (width = 20), Score: 12/20, Multiplier: 1.6, New width: 32, **New CI: 54-86%**
+
+### Step 5: Document reasoning
+
+**Record:** (1) What failure modes drove the adjustment, (2) Which perspective was most revealing, (3) What unknown unknowns were discovered, (4) What monitoring you'll do going forward
+
+**Next:** Return to [menu](#interactive-menu)
+
+---
+
+## 6. Learn the Framework
+
+**Deep dive into the methodology.**
+
+### Resource Files
+
+📄 **[Premortem Principles](resources/premortem-principles.md)** - Why humans are overconfident, hindsight bias and outcome bias, the power of inversion, research on premortem effectiveness
+
+📄 **[Backcasting Method](resources/backcasting-method.md)** - Structured backcasting process, temporal reasoning techniques, causal chain construction, narrative vs quantitative backcasting
+
+📄 **[Failure Mode Taxonomy](resources/failure-mode-taxonomy.md)** - Comprehensive failure categories, internal vs external failures, preventable vs unpreventable, PESTLE framework for tail risks, kill criteria templates
+
+**Next:** Return to [menu](#interactive-menu)
+
+---
+
+## Quick Reference
+
+### The Premortem Commandments
+
+1. **Assume failure is certain** - Don't debate whether, debate why
+2. **Be specific** - Vague risks don't help; concrete mechanisms do
+3. **Use multiple perspectives** - Skeptic, fanatic, observer
+4. **Quantify failure modes** - Estimate probability of each
+5. **Set kill criteria** - Know what would change your mind
+6. **Monitor signposts** - Track early warning signals
+7. **Widen CIs** - If premortem was too easy, you're overconfident
+
+### One-Sentence Summary
+
+> Assume your prediction has failed, write the history of how, and use that to identify blind spots and adjust confidence.
+
+### Integration with Other Skills
+
+- **Before:** Use after inside view analysis (you need something to stress-test)
+- **After:** Use `scout-mindset-bias-check` to validate adjustments
+- **Companion:** Works with `bayesian-reasoning-calibration` for quantitative updates
+- **Feeds into:** Monitoring systems and adaptive forecasting
+
+---
+
+## Resource Files
+
+📁 **resources/**
+- [premortem-principles.md](resources/premortem-principles.md) - Theory and research
+- [backcasting-method.md](resources/backcasting-method.md) - Temporal reasoning process
+- [failure-mode-taxonomy.md](resources/failure-mode-taxonomy.md) - Comprehensive failure categories
+
+---
+
+**Ready to start? Choose a number from the [menu](#interactive-menu) above.**
--- a/skills/forecast-premortem/resources/backcasting-method.md
+++ b/skills/forecast-premortem/resources/backcasting-method.md
@@ -0,0 +1,378 @@
+# Backcasting Method
+
+## Temporal Reasoning from Future to Present
+
+**Backcasting** is the practice of starting from a future state and working backward to identify the path that led there.
+
+**Contrast with forecasting:**
+- **Forecasting:** Present → Future (What will happen?)
+- **Backcasting:** Future → Present (How did this happen?)
+
+---
+
+## The Structured Backcasting Process
+
+### Phase 1: Define the Future State
+
+**Step 1.1: Set the resolution date**
+- When will you know if the prediction came true?
+- Be specific: "December 31, 2025"
+
+**Step 1.2: State the outcome as a certainty**
+- Don't say "might fail" or "probably fails"
+- Say "HAS failed" or "DID fail"
+- Use past tense
+
+**Step 1.3: Emotional calibration**
+- How surprising is this outcome?
+  - Shocking → You were very overconfident
+  - Expected → Appropriate confidence
+  - Inevitable → You were underconfident
+
+---
+
+### Phase 2: Construct the Timeline
+
+**Step 2.1: Work backward in time chunks**
+
+Start at resolution date, work backward in intervals:
+
+**For 2-year prediction:**
+- Resolution date (final failure)
+- 6 months before (late-stage warning)
+- 1 year before (mid-stage problems)
+- 18 months before (early signs)
+- Start date (initial conditions)
+
+**For 6-month prediction:**
+- Resolution date
+- 1 month before
+- 3 months before
+- Start date
+
+---
+
+**Step 2.2: Fill in each time chunk**
+
+For each period, ask:
+- What was happening at this time?
+- What decisions were made?
+- What external events occurred?
+- What warning signs appeared?
+
+**Template:**
+```
+[Date]: [Event that occurred]
+Effect: [How this contributed to failure]
+Warning sign: [What would have indicated this was coming]
+```
+
+---
+
+### Phase 3: Identify Causal Chains
+
+**Step 3.1: Map the causal structure**
+
+```
+Initial condition → Trigger event → Cascade → Failure
+```
+
+**Example:**
+```
+Team overworked → Key engineer quit → Lost 3 months → Missed deadline → Funding fell through → Failure
+```
+
+---
+
+**Step 3.2: Classify causes**
+
+| Type | Description | Example |
+|------|-------------|---------|
+| **Necessary** | Without this, failure wouldn't happen | Regulatory ban |
+| **Sufficient** | This alone causes failure | Founder death |
+| **Contributing** | Makes failure more likely | Market downturn |
+| **Catalytic** | Speeds up inevitable failure | Competitor launch |
+
+---
+
+**Step 3.3: Find the "brittle point"**
+
+**Question:** Which single event, if prevented, would have avoided failure?
+
+This is your **critical dependency** and highest-priority monitoring target.
+
+---
+
+### Phase 4: Narrative Construction
+
+**Step 4.1: Write the headlines**
+
+Imagine you're a journalist covering this failure. What headlines mark the timeline?
+
+**Example:**
+- "Startup X raises $10M Series A" (12 months before)
+- "Startup X faces regulatory scrutiny" (9 months before)
+- "Key executive departs Startup X" (6 months before)
+- "Startup X misses Q3 targets" (3 months before)
+- "Startup X shuts down, cites regulatory pressure" (resolution)
+
+---
+
+**Step 4.2: Write the obituary**
+
+"Startup X failed because..."
+
+Complete this sentence with a single, clear causal narrative. Force yourself to be concise.
+
+**Good:**
+"Startup X failed because regulatory uncertainty froze customer adoption, leading to missed revenue targets and inability to raise Series B."
+
+**Bad (too vague):**
+"Startup X failed because of various challenges."
+
+---
+
+**Step 4.3: The insider vs outsider narrative**
+
+**Insider view:** What would the founders say?
+- "We underestimated regulatory risk"
+- "We hired too slowly"
+- "We ran out of runway"
+
+**Outsider view:** What would analysts say?
+- "82% of startups in this space fail due to regulation"
+- "Classic execution failure"
+- "Unit economics never made sense"
+
+**Compare:** Does your insider narrative match outsider base rates?
+
+---
+
+## Narrative vs Quantitative Backcasting
+
+### Narrative Backcasting
+
+**Strengths:**
+- Rich, detailed stories
+- Reveals unknown unknowns
+- Good for complex systems
+
+**Weaknesses:**
+- Subject to narrative fallacy
+- Can feel too "real" and bias you
+- Hard to quantify
+
+**Use when:**
+- Complex, multi-causal failures
+- Human/organizational factors dominate
+- Need to surface blind spots
+
+---
+
+### Quantitative Backcasting
+
+**Strengths:**
+- Precise probability estimates
+- Aggregates multiple failure modes
+- Less subject to bias
+
+**Weaknesses:**
+- Requires data
+- Can miss qualitative factors
+- May feel mechanical
+
+**Use when:**
+- Statistical models exist
+- Multiple independent failure modes
+- Need to calculate confidence intervals
+
+---
+
+## Advanced Technique: Multiple Backcast Paths
+
+### Generate 3-5 Different Failure Narratives
+
+Instead of one story, create multiple:
+
+**Path 1: Internal Execution Failure**
+- Team burned out
+- Product quality suffered
+- Customers churned
+- Revenue missed
+- Funding dried up
+
+**Path 2: External Market Shift**
+- Competitor launched free tier
+- Market commoditized
+- Margins compressed
+- Unit economics broke
+- Shutdown
+
+**Path 3: Regulatory Kill**
+- New law passed
+- Business model illegal
+- Forced shutdown
+
+**Path 4: Black Swan**
+- Pandemic
+- Supply chain collapse
+- Force majeure
+
+---
+
+### Aggregate the Paths
+
+**Calculate probability for each path:**
+- Path 1 (Internal): 40%
+- Path 2 (Market): 30%
+- Path 3 (Regulatory): 20%
+- Path 4 (Black Swan): 10%
+
+**Total failure probability:** 100% (since we assumed failure)
+
+**Insight:** But in reality, your prediction gives 25% failure. This means you're underestimating by 75 percentage points, OR these paths are not independent.
+
+**Adjustment:**
+If paths are partially overlapping (e.g., internal failure AND market shift), use:
+```
+P(A or B) = P(A) + P(B) - P(A and B)
+```
+
+---
+
+## Temporal Reasoning Techniques
+
+### The "Newspaper Test"
+
+**Method:**
+For each time period, imagine you're reading a newspaper from that date.
+
+**What headlines would you see?**
+- Macro news (economy, politics, technology)
+- Industry news (competitors, regulations, trends)
+- Company news (your specific case)
+
+**This forces you to think about:**
+- External context, not just internal execution
+- Leading indicators, not just lagging outcomes
+
+---
+
+### The "Retrospective Interview"
+
+**Method:**
+Imagine you're interviewing someone 1 year after failure.
+
+**Questions:**
+- "Looking back, when did you first know this was in trouble?"
+- "What was the moment of no return?"
+- "If you could go back, what would you change?"
+- "What signs did you ignore?"
+
+**This reveals:**
+- Early warning signals you should monitor
+- Critical decision points
+- Hindsight that can become foresight
+
+---
+
+### The "Parallel Universe" Technique
+
+**Method:**
+Create two timelines:
+
+**Timeline A: Success**
+What had to happen for success?
+
+**Timeline B: Failure**
+What happened instead?
+
+**Divergence point:**
+Where do the timelines split? That's your critical uncertainty.
+
+---
+
+## Common Backcasting Mistakes
+
+### Mistake 1: Being Too Vague
+
+**Bad:** "Things went wrong and it failed."
+**Good:** "Q3 2024: Competitor X launched free tier. Q4 2024: We lost 30% of customers. Q1 2025: Revenue dropped below runway. Q2 2025: Failed to raise Series B. Q3 2025: Shutdown."
+
+**Fix:** Force yourself to name specific events and dates.
+
+---
+
+### Mistake 2: Only Internal Causes
+
+**Bad:** "We executed poorly."
+**Good:** "We executed poorly AND market shifted AND regulation changed."
+
+**Fix:** Use PESTLE framework to ensure external factors are considered.
+
+---
+
+### Mistake 3: Hindsight Bias
+
+**Bad:** "It was always obvious this would fail."
+**Good:** "In retrospect, these warning signs were present, but at the time they were ambiguous."
+
+**Fix:** Acknowledge that foresight ≠ hindsight. Don't pretend everything was obvious.
+
+---
+
+### Mistake 4: Single-Cause Narratives
+
+**Bad:** "Failed because of regulation."
+**Good:** "Regulation was necessary but not sufficient. Also needed internal execution failure and market downturn to actually fail."
+
+**Fix:** Multi-causal explanations are almost always more accurate.
+
+---
+
+## Integration with Forecasting
+
+### How Backcasting Improves Forecasts
+
+**Before Backcasting:**
+- Forecast: 80% success
+- Reasoning: Strong team, good market, solid plan
+- Confidence interval: 70-90%
+
+**After Backcasting:**
+- Identified failure modes: Regulatory (20%), Execution (15%), Market (10%), Black Swan (5%)
+- Total failure probability from backcasting: 50%
+- **Realized:** Current 80% is too high
+- **Adjusted forecast:** 60% success
+- **Adjusted CI:** 45-75% (wider, reflecting uncertainty)
+
+---
+
+## Practical Workflow
+
+### Quick Backcast (15 minutes)
+
+1. **State outcome:** "It failed."
+2. **One-sentence cause:** "Failed because..."
+3. **Three key events:** Timeline points
+4. **Probability check:** Does failure narrative feel >20% likely?
+5. **Adjust:** If yes, lower confidence.
+
+---
+
+### Rigorous Backcast (60 minutes)
+
+1. Define future state and resolution date
+2. Create timeline working backward in chunks
+3. Write detailed narrative for each period
+4. Identify causal chains (necessary, sufficient, contributing)
+5. Generate 3-5 alternative failure paths
+6. Estimate probability of each path
+7. Aggregate and compare to current forecast
+8. Adjust probability and confidence intervals
+9. Set monitoring signposts
+10. Document assumptions
+
+---
+
+**Return to:** [Main Skill](../SKILL.md#interactive-menu)
--- a/skills/forecast-premortem/resources/failure-mode-taxonomy.md
+++ b/skills/forecast-premortem/resources/failure-mode-taxonomy.md
@@ -0,0 +1,497 @@
+# Failure Mode Taxonomy
+
+## Comprehensive Categories for Systematic Risk Identification
+
+---
+
+## The Two Primary Dimensions
+
+### 1. Internal vs External
+
+**Internal failures:**
+- Under your control (at least partially)
+- Organizational, execution, resource-based
+- Can be prevented with better planning
+
+**External failures:**
+- Outside your control
+- Market, regulatory, competitive, acts of God
+- Can only be mitigated, not prevented
+
+---
+
+### 2. Preventable vs Unpreventable
+
+**Preventable:**
+- Known risk with available mitigation
+- Happens due to negligence or oversight
+- "We should have seen this coming"
+
+**Unpreventable (Black Swans):**
+- Unknown unknowns
+- No reasonable way to anticipate
+- "Nobody could have predicted this"
+
+---
+
+## Four Quadrants
+
+|  | **Preventable** | **Unpreventable** |
+|---|---|---|
+| **Internal** | Execution failure, bad hiring | Key person illness, burnout |
+| **External** | Competitor launch (foreseeable) | Pandemic, war, black swan |
+
+**Premortem focus:** Mostly on **preventable failures** (both internal and external)
+
+---
+
+## Internal Failure Modes
+
+### 1. Execution Failures
+
+**Team/People:**
+- Key person quits
+- Co-founder conflict
+- Team burnout
+- Cultural toxicity
+- Skills gap
+- Hiring too slow/fast
+- Onboarding failure
+
+**Process:**
+- Missed deadlines
+- Scope creep
+- Poor prioritization
+- Communication breakdown
+- Decision paralysis
+- Process overhead
+- Lack of process
+
+**Product/Technical:**
+- Product quality issues
+- Technical debt collapse
+- Scalability failures
+- Security breach
+- Data loss
+- Integration failures
+- Performance degradation
+
+---
+
+### 2. Resource Failures
+
+**Financial:**
+- Ran out of money (runway)
+- Failed to raise funding
+- Revenue shortfall
+- Cost overruns
+- Budget mismanagement
+- Fraud/embezzlement
+- Cash flow crisis
+
+**Time:**
+- Too slow to market
+- Missed window of opportunity
+- Critical path delays
+- Underestimated timeline
+- Overcommitted resources
+
+**Knowledge/IP:**
+- Lost key knowledge (person left)
+- IP stolen
+- Failed to protect IP
+- Technological obsolescence
+- R&D dead ends
+
+---
+
+### 3. Strategic Failures
+
+**Market fit:**
+- Built wrong product
+- Solved non-problem
+- Target market too small
+- Pricing wrong
+- Value prop unclear
+- Positioning failure
+
+**Business model:**
+- Unit economics don't work
+- CAC > LTV
+- Churn too high
+- Margins too thin
+- Revenue model broken
+- Unsustainable burn rate
+
+**Competitive:**
+- Differentiation lost
+- Commoditization
+- Underestimated competition
+- Failed to defend moat
+- Technology leapfrogged
+
+---
+
+## External Failure Modes
+
+### 1. Market Failures
+
+**Demand side:**
+- Market smaller than expected
+- Adoption slower than expected
+- Customer behavior changed
+- Willingness to pay dropped
+- Switching costs too high
+
+**Supply side:**
+- Input costs increased
+- Suppliers failed
+- Supply chain disruption
+- Talent shortage
+- Infrastructure unavailable
+
+**Market structure:**
+- Market consolidated
+- Winner-take-all dynamics
+- Network effects favored competitor
+- Platform risk (dependency on another company)
+
+---
+
+### 2. Competitive Failures
+
+**Direct competition:**
+- Incumbent responded aggressively
+- New entrant with more capital
+- Competitor launched superior product
+- Price war
+- Competitor acquired key talent
+
+**Ecosystem:**
+- Complementary product failed
+- Partnership fell through
+- Distribution channel cut off
+- Platform changed terms
+- Ecosystem shifted away
+
+---
+
+### 3. Regulatory/Legal Failures
+
+**Regulation:**
+- New law banned business model
+- Compliance costs too high
+- Licensing denied
+- Government investigation
+- Regulatory capture by incumbents
+
+**Legal:**
+- Lawsuit (IP, employment, customer)
+- Contract breach
+- Fraud allegations
+- Criminal charges
+- Bankruptcy proceedings
+
+---
+
+### 4. Macroeconomic Failures
+
+**Economic:**
+- Recession
+- Inflation
+- Interest rate spike
+- Currency fluctuation
+- Credit crunch
+- Stock market crash
+
+**Geopolitical:**
+- War
+- Trade restrictions
+- Sanctions
+- Political instability
+- Coup/revolution
+- Expropriation
+
+---
+
+### 5. Technological Failures
+
+**Disruption:**
+- New technology made product obsolete
+- Paradigm shift (e.g., mobile, cloud, AI)
+- Standard changed
+- Interoperability broke
+
+**Infrastructure:**
+- Cloud provider outage
+- Internet backbone failure
+- Power grid failure
+- Critical dependency failed
+
+---
+
+### 6. Social/Cultural Failures
+
+**Public opinion:**
+- Reputation crisis
+- Boycott
+- Social media backlash
+- Cultural shift away from product
+- Ethical concerns raised
+
+**Demographics:**
+- Target demographic shrunk
+- Generational shift
+- Migration patterns changed
+- Urbanization/de-urbanization
+
+---
+
+### 7. Environmental/Health Failures
+
+**Natural disasters:**
+- Earthquake, hurricane, flood
+- Wildfire
+- Drought
+- Extreme weather
+
+**Health:**
+- Pandemic
+- Endemic disease outbreak
+- Health regulation
+- Contamination/recall
+
+---
+
+## Black Swans (Unpreventable External)
+
+### Characteristics
+- Extreme impact
+- Retrospectively predictable, prospectively invisible
+- Outside normal expectations
+
+### Examples
+- 9/11 terrorist attacks
+- COVID-19 pandemic
+- 2008 financial crisis
+- Fukushima disaster
+- Technological singularity
+- Asteroid impact
+
+### How to Handle
+**Can't prevent, can:**
+1. **Increase robustness** - Survive the shock
+2. **Increase antifragility** - Benefit from volatility
+3. **Widen confidence intervals** - Acknowledge unknown unknowns
+4. **Plan for "unspecified bad thing"** - Generic resilience
+
+---
+
+## PESTLE Framework for Systematic Enumeration
+
+Use this checklist to ensure comprehensive coverage:
+
+### Political
+- [ ] Elections/regime change
+- [ ] Policy shifts
+- [ ] Government instability
+- [ ] Geopolitical conflict
+- [ ] Trade agreements
+- [ ] Lobbying success/failure
+
+### Economic
+- [ ] Recession/depression
+- [ ] Inflation/deflation
+- [ ] Interest rates
+- [ ] Currency fluctuations
+- [ ] Market bubbles/crashes
+- [ ] Unemployment
+
+### Social
+- [ ] Demographic shifts
+- [ ] Cultural trends
+- [ ] Public opinion
+- [ ] Social movements
+- [ ] Consumer behavior change
+- [ ] Generational values
+
+### Technological
+- [ ] Disruptive innovation
+- [ ] Obsolescence
+- [ ] Cyber attacks
+- [ ] Infrastructure failure
+- [ ] Standards change
+- [ ] Technology convergence
+
+### Legal
+- [ ] New regulations
+- [ ] Lawsuits
+- [ ] IP challenges
+- [ ] Compliance requirements
+- [ ] Contract disputes
+- [ ] Liability exposure
+
+### Environmental
+- [ ] Climate change
+- [ ] Natural disasters
+- [ ] Pandemics
+- [ ] Resource scarcity
+- [ ] Pollution/contamination
+- [ ] Sustainability pressures
+
+---
+
+## Kill Criteria Templates
+
+### What is a Kill Criterion?
+
+**Definition:** A specific event that, if it occurs, drastically changes your probability.
+
+**Format:** "If [event], then probability drops to [X%]"
+
+---
+
+### Template Library
+
+**Regulatory kill criteria:**
+```
+If [specific regulation] passes, probability drops to [X]%
+If FDA rejects in Phase [N], probability drops to [X]%
+If government bans [activity], probability drops to 0%
+```
+
+**Competitive kill criteria:**
+```
+If [competitor] launches [feature], probability drops to [X]%
+If incumbent drops price by [X]%, probability drops to [X]%
+If [big tech co] enters market, probability drops to [X]%
+```
+
+**Financial kill criteria:**
+```
+If we miss Q[N] revenue target by >20%, probability drops to [X]%
+If we can't raise Series [X] by [date], probability drops to [X]%
+If burn rate exceeds $[X]/month, probability drops to [X]%
+```
+
+**Team kill criteria:**
+```
+If [key person] leaves, probability drops to [X]%
+If we can't hire [critical role] by [date], probability drops to [X]%
+If team size drops below [X], probability drops to [X]%
+```
+
+**Product kill criteria:**
+```
+If we can't ship by [date], probability drops to [X]%
+If NPS drops below [X], probability drops to [X]%
+If churn exceeds [X]%, probability drops to [X]%
+```
+
+**Market kill criteria:**
+```
+If TAM shrinks below $[X], probability drops to [X]%
+If adoption rate < [X]% by [date], probability drops to [X]%
+If market shifts to [substitute], probability drops to [X]%
+```
+
+**Macro kill criteria:**
+```
+If recession occurs, probability drops to [X]%
+If interest rates exceed [X]%, probability drops to [X]%
+If war breaks out in [region], probability drops to [X]%
+```
+
+---
+
+## Failure Mode Probability Estimation
+
+### Quick Heuristics
+
+**For each failure mode, estimate:**
+
+**Very Low (1-5%):**
+- Black swans
+- Never happened in this industry
+- Requires multiple unlikely events
+
+**Low (5-15%):**
+- Happened before but rare
+- Strong mitigations in place
+- Early warning systems exist
+
+**Medium (15-35%):**
+- Common failure mode in industry
+- Moderate mitigations
+- Uncertain effectiveness
+
+**High (35-70%):**
+- Very common failure mode
+- Weak mitigations
+- History of this happening
+
+**Very High (>70%):**
+- Almost certain to occur
+- No effective mitigation
+- Base rate is very high
+
+---
+
+### Aggregation
+
+**If failure modes are independent:**
+```
+P(any failure) = 1 - ∏(1 - P(failure_i))
+```
+
+**Example:**
+- P(regulatory) = 20%
+- P(competitive) = 30%
+- P(execution) = 25%
+
+```
+P(any) = 1 - (0.8 × 0.7 × 0.75) = 1 - 0.42 = 58%
+```
+
+**If failure modes are dependent:**
+Use Venn diagram logic or conditional probabilities (more complex).
+
+---
+
+## Monitoring and Signposts
+
+### Early Warning Signals
+
+For each major failure mode, identify **leading indicators:**
+
+**Example: "Key engineer will quit"**
+
+**Leading indicators (6-12 months before):**
+- Code commit frequency drops
+- Participation in meetings declines
+- Starts saying "no" more often
+- Takes more sick days
+- LinkedIn profile updated
+- Asks about vesting schedule
+
+**Action:** Monitor these monthly, set alerts
+
+---
+
+### Monitoring Cadence
+
+| Risk Level | Check Frequency |
+|------------|----------------|
+| Very High (>50%) | Weekly |
+| High (25-50%) | Bi-weekly |
+| Medium (10-25%) | Monthly |
+| Low (5-10%) | Quarterly |
+| Very Low (<5%) | Annually |
+
+---
+
+## Practical Usage
+
+**Step-by-Step:** (1) Choose categories (Internal/External, PESTLE), (2) Brainstorm 10-15 failure modes, (3) Estimate probability for each, (4) Aggregate, (5) Compare to forecast, (6) Identify top 3-5 risks, (7) Set kill criteria, (8) Define monitoring signposts, (9) Set calendar reminders based on risk level.
+
+**Return to:** [Main Skill](../SKILL.md#interactive-menu)
--- a/skills/forecast-premortem/resources/premortem-principles.md
+++ b/skills/forecast-premortem/resources/premortem-principles.md
@@ -0,0 +1,292 @@
+# Premortem Principles
+
+## The Psychology of Overconfidence
+
+### Why We're Systematically Overconfident
+
+**The Planning Fallacy:**
+- We focus on best-case scenarios
+- We ignore historical delays and failures
+- We assume "our case is different"
+- We underestimate Murphy's Law
+
+**Research:**
+- 90% of projects run over budget
+- 70% of projects run late
+- Yet 80% of project managers predict on-time completion
+
+**The fix:** Premortem forces you to imagine failure has already happened.
+
+---
+
+## Hindsight Bias
+
+### The "I Knew It All Along" Effect
+
+**What it is:**
+After an outcome occurs, we believe we "always knew" it would happen.
+
+**Example:**
+- Before 2008 crash: "Housing is safe"
+- After 2008 crash: "The signs were obvious"
+
+**Problem for forecasting:**
+If we think outcomes were predictable in hindsight, we'll be overconfident going forward.
+
+**The premortem fix:**
+By forcing yourself into "hindsight mode" BEFORE the outcome, you:
+1. Generate the warning signs you would have seen
+2. Realize how many ways things could go wrong
+3. Reduce overconfidence
+
+---
+
+## The Power of Inversion
+
+### Solving Problems Backward
+
+**Charlie Munger:**
+> "Invert, always invert. Many hard problems are best solved backward."
+
+**In forecasting:**
+- Hard: "Will this succeed?" (requires imagining all paths to success)
+- Easier: "It failed - why?" (failure modes are more concrete)
+
+**Why this works:**
+- Failure modes are finite and enumerable
+- Success paths are infinite and vague
+- Humans are better at imagining concrete negatives than abstract positives
+
+---
+
+## Research on Premortem Effectiveness
+
+### Gary Klein's Studies
+
+**Original research:**
+- Teams that did premortems identified 30% more risks
+- Risks identified were more specific and actionable
+- Teams adjusted plans proactively
+
+**Key finding:**
+> "Prospective hindsight" (imagining an event has happened) improves recall by 30%
+
+---
+
+### Kahneman's Endorsement
+
+**Daniel Kahneman:**
+> "The premortem is the single best debiasing technique I know."
+
+**Why it works:**
+1. **Legitimizes doubt** - In group settings, dissent is hard. Premortem makes it safe.
+2. **Concrete > Abstract** - "Identify risks" is vague. "Explain the failure" is concrete.
+3. **Defeats groupthink** - Forces even optimists to imagine failure.
+
+---
+
+## Outcome Bias
+
+### Judging Decisions by Results, Not Process
+
+**What it is:**
+We judge the quality of a decision based on its outcome, not the process.
+
+**Example:**
+- Drunk driver gets home safely → "It was fine"
+- Sober driver has accident → "Bad decision to drive"
+
+**Reality:**
+Quality of decision ≠ Quality of outcome (because of randomness)
+
+**For forecasting:**
+A 90% prediction that fails doesn't mean the forecast was bad (10% events happen 10% of the time).
+
+**The premortem fix:**
+By imagining failure BEFORE it happens, you evaluate the decision process independent of outcome.
+
+---
+
+## When Premortems Work Best
+
+### High-Confidence Predictions
+
+**Use when:**
+- Your probability is >80% or <20%
+- You feel very certain
+- Confidence intervals are narrow
+
+**Why:**
+These are the predictions most likely to be overconfident.
+
+---
+
+### Team Forecasting
+
+**Use when:**
+- Multiple people are making predictions
+- Groupthink is a risk
+- Dissent is being suppressed
+
+**Why:**
+Premortems legitimize expressing doubts without seeming disloyal.
+
+---
+
+### Important Decisions
+
+**Use when:**
+- Stakes are high
+- Irreversible commitments
+- Significant resource allocation
+
+**Why:**
+Worth the time investment to reduce overconfidence.
+
+---
+
+## When Premortems Don't Help
+
+### Already Uncertain
+
+**Skip if:**
+- Your probability is ~50%
+- Confidence intervals are already wide
+- You're confused, not confident
+
+**Why:**
+You don't need a premortem to tell you you're uncertain.
+
+---
+
+### Trivial Predictions
+
+**Skip if:**
+- Low stakes
+- Easily reversible
+- Not worth the time
+
+**Why:**
+Premortems take effort; save them for important forecasts.
+
+---
+
+## The Premortem vs Other Techniques
+
+### Premortem vs Red Teaming
+
+**Red Teaming:**
+- Adversarial: Find flaws in the plan
+- Focus: Attack the strategy
+- Mindset: "How do we defeat this?"
+
+**Premortem:**
+- Temporal: Failure has occurred
+- Focus: Understand what happened
+- Mindset: "What led to this outcome?"
+
+**Use both:** Red team attacks the plan, premortem explains the failure.
+
+---
+
+### Premortem vs Scenario Planning
+
+**Scenario Planning:**
+- Multiple futures: Good, bad, likely
+- Branching paths
+- Strategies for each scenario
+
+**Premortem:**
+- Single future: Failure has occurred
+- Backward path
+- Identify risks to avoid
+
+**Use both:** Scenario planning explores, premortem stress-tests.
+
+---
+
+### Premortem vs Risk Register
+
+**Risk Register:**
+- List of identified risks
+- Probability and impact scores
+- Mitigation strategies
+
+**Premortem:**
+- Narrative of failure
+- Causal chains
+- Discover unknown unknowns
+
+**Use both:** Premortem feeds into risk register.
+
+---
+
+## Cognitive Mechanisms
+
+### Why Premortems Defeat Overconfidence
+
+**1. Prospective Hindsight**
+Imagining an event has occurred improves memory access by 30%.
+
+**2. Permission to Doubt**
+Social license to express skepticism without seeming negative.
+
+**3. Concrete Failure Modes**
+Abstract "risks" become specific "this happened, then this, then this."
+
+**4. Temporal Distancing**
+Viewing from the future reduces emotional attachment to current plan.
+
+**5. Narrative Construction**
+Building a story forces causal reasoning, revealing gaps.
+
+---
+
+## Common Objections
+
+### "This is too negative!"
+
+**Response:**
+Pessimism during planning prevents failure during execution.
+
+**Reframe:**
+Not negative - realistic. You're not hoping for failure, you're preparing for it.
+
+---
+
+### "We don't have time for this."
+
+**Response:**
+- Premortem: 30 minutes
+- Recovering from preventable failure: Months/years
+
+**Math:**
+If premortem prevents 10% of failures, ROI is massive.
+
+---
+
+### "Our case really is different!"
+
+**Response:**
+Maybe. But the premortem will reveal HOW it's different, not just assert it.
+
+**Test:**
+If the premortem reveals nothing new, you were right. If it reveals risks, you weren't.
+
+---
+
+## Practical Takeaways
+
+1. **Use for high-confidence predictions** - When you feel certain
+2. **Legitimate skepticism** - Makes doubt socially acceptable
+3. **Concrete failure modes** - Forces specific risks, not vague worries
+4. **Widen confidence intervals** - Adjust based on plausibility of failure narrative
+5. **Set kill criteria** - Know what would change your mind
+6. **Monitor signposts** - Track early warning signals
+
+**The Rule:**
+> If you can easily write a plausible failure narrative, your confidence is too high.
+
+---
+
+**Return to:** [Main Skill](../SKILL.md#interactive-menu)