decision-engine-lite（决策引擎精简版）

Decision Engine Lite

*"In any moment of decision, the best thing you can do is the right thing. The worst thing you can do is nothing."* — Theodore Roosevelt

🧠 Want the full framework library with bias detection and decision policies? Full version → agentofalpha.com

What This Skill Does

Brings structure to high-stakes decisions so you stop going in circles. Classify the decision, surface the real tradeoffs, and stress-test it before you commit.

Included in Lite:

• ✅ Decision classification (4 types — know how much rigor to apply)
• ✅ Structured pro/con framework (weighted, not just a list)
• ✅ Simple pre-mortem exercise (find the hidden risks before they find you)
• ✅ Clear next-step recommendation

Upgrade to Full for:

• ❌ Full framework library (OODA loop, Eisenhower matrix, RICE scoring, Regret Minimization, Opportunity Cost, Bayesian updating, and more)
• ❌ Cognitive bias detection checklist (15 biases with specific mitigations)
• ❌ Group decision-making (RAPID framework, disagree-and-commit protocol)
• ❌ Scenario planning with expected value calculations
• ❌ Organizational decision policies (convert recurring decisions into rules)
• ❌ Decision quality scoring rubric (100-point framework)
• ❌ Decision record templates and decision log

How to Use

Tell me the decision you're facing. Include:

• What you're deciding
• The options you're considering
• Any constraints (budget, timeline, must-haves)
• What's at stake if you get it wrong

I'll run you through the framework.

Phase 1: Decision Classification

Before applying any framework, classify the decision. This tells you how much time and rigor to invest.

The 4 Decision Types

Classification Questions

Ask yourself:

1. If this goes wrong, can we fix it within 30 days at reasonable cost? → Yes = Type 2
2. Is the cost of being wrong more than 10× the cost of analysis? → Yes = Type 1
3. Have we made this exact decision 3+ times before? → Yes = Type 3 (build a policy)
4. Does someone closer to the work have better information to decide this? → Yes = Type 4

Delegation Test (Type 4 criteria)

Delegate when ALL of these are true:

• The decision is reversible within an acceptable window
• The downside is less than 5% of the relevant budget or resource
• Someone closer to the problem can decide better than you
• Speed matters more than perfection here

What classification tells you

• Type 1: Don't rush. Run the full pre-mortem. Get outside perspective.
• Type 2: Decide now with the information you have. Don't let analysis paralysis set in.
• Type 3: The answer is a policy, not a decision. Stop solving this individually.
• Type 4: Delegate and move on. Deciding this yourself is a poor use of your judgment.

Phase 2: Structured Pro/Con Analysis

A basic pro/con list is weak because all factors are treated as equal. This version weights them.

Step 1: List the Options

Name each option clearly. If you only have one option and one "status quo," that's fine — write both down.

Step 2: Define Criteria

What actually matters for this decision? List 3-6 criteria. Examples:

• Financial impact
• Speed of execution
• Risk level
• Alignment with long-term goals
• Team/stakeholder impact
• Reversibility

Assign a weight to each criterion (1-5):

• 5 = Critical — a bad score here could be a dealbreaker
• 3 = Important — matters but won't make or break the decision
• 1 = Nice to have — relevant but minor

Step 3: Score Each Option

Score each option against each criterion (1-10):

• 9-10: Excellent
• 7-8: Strong
• 5-6: Acceptable
• 3-4: Below average
• 1-2: Poor / major concern

Calculate: Weighted score = Σ (criterion weight × option score)

Step 4: Gut Check

After calculating scores — how do you feel about the winner?

If the math says Option A but your gut says Option B, that's data. Name the feeling. Ask: "What criterion did I underweight or miss?"

Your gut is not infallible, but it often detects factors you haven't articulated yet.

Output Format

Decision: [What we're deciding]

Criteria & Weights:
- [Criterion 1]: Weight X/5
- [Criterion 2]: Weight X/5
- [Criterion 3]: Weight X/5

Scoring:
| Criterion (Weight) | Option A | Option B |
|--------------------|----------|----------|
| [Criterion 1] (×X) | X | X |
| [Criterion 2] (×X) | X | X |
| [Criterion 3] (×X) | X | X |
| **Weighted Total** | **XX** | **XX** |

Winner by score: [Option]
Gut check: [Does the winner feel right? Any flag?]

Phase 3: Pre-Mortem Exercise

This is the most important step most people skip.

How It Works

Imagine it's 12 months from now. You made this decision. It failed spectacularly.

Not a minor setback — a real failure. What went wrong?

The Exercise

Step 1: Write failure scenarios

List 5-7 specific ways this decision could go badly. Don't be optimistic — be honest. Think about:

• What assumptions could prove wrong?
• What external factors could change?
• What internal execution risks exist?
• What might you have underestimated?

Step 2: Rate each scenario

For each failure scenario:

• Likelihood: Low / Medium / High
• Impact if it happens: Minor / Significant / Catastrophic
• Detectability: Would you see it coming, or only after it's too late?

Step 3: Focus on High + Catastrophic

Any scenario rated "High likelihood + Significant/Catastrophic impact" needs a mitigation plan or needs to change your decision.

Step 4: Update your decision

After the pre-mortem:

• Does any failure scenario change which option you pick?
• Can you add safeguards that reduce your biggest risks?
• Are there go/no-go criteria you should set in advance? ("If X happens within 90 days, we reverse the decision")

Pre-Mortem Output Format

Pre-Mortem: [Decision]

Failure Scenarios:
1. [Scenario] | Likelihood: [L/M/H] | Impact: [Minor/Significant/Catastrophic]
   → Mitigation: [How to reduce likelihood or damage]

2. [Scenario] | Likelihood: [L/M/H] | Impact: [Minor/Significant/Catastrophic]
   → Mitigation: [How to reduce likelihood or damage]

[Continue for each scenario]

Kill Criteria (set in advance):
- If [observable signal], we reverse or pivot by [date]

Updated Confidence: [Do you feel better or worse about the decision after this exercise?]

Putting It Together

Final Recommendation Output

## Decision: [Clear statement of what we're deciding]

**Classification:** Type [1/2/3/4] — [One-way door / Two-way door / Recurring / Delegatable]
**Urgency:** [How quickly does this need to be decided?]

---

### Options Considered
- Option A: [Brief description]
- Option B: [Brief description]
[Additional options if any]

---

### Weighted Analysis
[Table from Phase 2]
**Score winner:** Option [X] with [XX] vs [XX]

---

### Pre-Mortem Summary
**Top risks identified:**
1. [Biggest risk + mitigation]
2. [Second risk + mitigation]

**Kill criteria:** [What would make you reverse this within 90 days?]

---

### Recommendation
**Choose:** [Option]
**Why:** [2-3 sentences tying together the score, the gut check, and the risk assessment]
**By when:** [Decision deadline — Type 2 decisions should be decided now]
**First action:** [What do you do in the next 24 hours?]

Quick Shortcuts

When you're stuck and going in circles: → You probably have a Type 2 decision. Decide with 70% of the information you wish you had. The cost of delay is exceeding the cost of a suboptimal choice.

When everyone's agreeing too fast: → Run the pre-mortem. Assign one person to actively argue against the leading option before you commit.

When it feels wrong but the analysis says go: → Name the feeling. What criterion did you underweight? Adjust the model or adjust the decision — but don't ignore the signal.

When you're making the same decision repeatedly: → This is Type 3. Stop deciding case-by-case. Write a policy.

Where the Lite Version Ends

You can make a significantly better decision with just these three phases. Classification prevents you from over-thinking easy decisions and under-thinking hard ones. The weighted analysis surfaces the real tradeoffs. The pre-mortem catches what optimism hides.

What you won't get here:

• The OODA loop (Observe-Orient-Decide-Act) for time-pressured decisions in dynamic environments
• The Eisenhower matrix + RICE scoring for prioritizing competing options
• Second-order thinking — consequences of consequences (where most strategic decisions go wrong)
• Regret Minimization framework for personal / career decisions
• Cognitive bias checklist — 15 specific biases with targeted mitigations and a bias risk score
• RAPID framework for group decisions (who decides, who advises, who executes — with one clear decider)
• Scenario planning with expected value calculations across bull/base/bear cases
• Decision policies — convert recurring Type 3 decisions into rules that eliminate repeated deliberation
• Decision quality scoring rubric — rate the process, not just the outcome
• Decision record templates for building an organizational decision log

The full version is used by founders and executives facing strategic pivots, hiring calls, investment decisions, and product prioritization — any choice where the cost of being wrong is high.

🧠 Want the full framework library with bias detection and decision policies? Full version → agentofalpha.com

Example Queries

• "Help me decide whether to take this job offer"
• "We're choosing between two vendors — walk me through it"
• "Should we build or buy this feature?"
• "I can't decide whether to raise a round now or wait — help me think through it"
• "Run a pre-mortem on our decision to enter this new market"
• "We keep revisiting our pricing strategy — how do we just decide?"