决策框架

---
name: decision-frameworks
description: "When to activate: decision frameworks, RAPID, RACI, decision matrix, reversible decisions, pre-mortem, second-order thinking, decision making, prioritization"
origin: zakr
metadata:
  namespace: zakr/devtools-pack
  version: 1.0.0
  category: business
  domain: productivity
  platforms: [claude-code, codex, gemini-cli, cursor, aider, windsurf]
---

# Decision Frameworks

## Decision Classification

### Reversible vs Irreversible (Bezos Type 1/2)
| Type | Description | Process |
|------|-------------|---------|
| **Type 1** (One-way door) | Hard to reverse, high stakes | Slow, careful, get more data |
| **Type 2** (Two-way door) | Reversible, low stakes | Fast, decentralize, iterate |

Most decisions are Type 2. Move fast on Type 2 — speed matters more than perfection.
Identify Type 1 early and apply appropriate rigor.

### Decision Urgency Matrix
```
           High Impact    Low Impact
Urgent    → Decide now   → Delegate
Not urgent → Schedule    → Automate/Ignore
```

## RAPID Framework

### Role Definitions
| Letter | Role | Responsibility |
|--------|------|---------------|
| **R** | Recommend | Gathers input, proposes a recommendation |
| **A** | Agree | Must formally agree (veto power) |
| **P** | Perform | Executes the decision |
| **I** | Input | Consulted, provides relevant knowledge |
| **D** | Decide | Makes the final call (one person only) |

### RAPID Assignment Template
```
Decision: [What are we deciding?]
Deadline: [By when]

R (Recommend): [Name] — will provide recommendation by [date]
A (Agree): [Names] — must sign off (keep to ≤2 people)
P (Perform): [Name/team] — will implement
I (Input): [Names] — consulted, not deciding
D (Decide): [ONE person] — makes final call

Common mistake: multiple D's → paralysis. Always one D.
```

## Decision Matrix (Weighted Criteria)

### Building a Decision Matrix
```
Step 1: List options (rows)
Step 2: Define criteria (columns)  
Step 3: Assign weights to criteria (must sum to 100%)
Step 4: Score each option on each criterion (1-10)
Step 5: Multiply score × weight, sum across criteria

Example: Selecting a database

Criteria:      Perf  Scale  Cost  Ease  Ecosystem
Weight:         30%   25%   20%   15%    10%

PostgreSQL:      8     7     9     8      9
MySQL:           7     7     9     9      8
MongoDB:         8     9     7     6      8
DynamoDB:        9    10     6     5      7

Weighted scores:
PostgreSQL: 8(.30)+7(.25)+9(.20)+8(.15)+9(.10) = 7.95 ✓
MongoDB:    8(.30)+9(.25)+7(.20)+6(.15)+8(.10)  = 7.75
DynamoDB:   9(.30)+10(.25)+6(.20)+5(.15)+7(.10) = 7.90

Result: PostgreSQL wins — but decision matrix shows DynamoDB close on perf/scale
```

## Pre-Mortem

### Pre-Mortem Protocol
```
When: Before committing to a major decision or project

Step 1 (5 min): State the plan clearly — what exactly are we deciding?

Step 2 (10 min, solo): "Imagine it's 12 months from now. 
The decision was made. It was a disaster. 
Write down everything that went wrong."

Step 3 (15 min): Share and cluster failure scenarios

Step 4 (10 min): For each top scenario, ask:
- How likely (1-5)?
- How severe (1-5)?  
- Can we mitigate it? How?

Step 5: Decide with eyes open, or modify the plan

Pre-mortem catches: overconfidence, groupthink, planning fallacy
```

## Second-Order Thinking

### Second-Order Thinking Framework
```
First-order: What happens immediately?
Second-order: Then what happens?
Third-order: And then what?

Example: "We should lower prices to grow revenue"
1st order: More customers sign up → revenue up
2nd order: Existing customers demand same price → margin hit
3rd order: Competitors match price → race to bottom
3rd order: Resources cut → quality suffers → churn rises
→ Decision is more complex than it appeared
```

### "And then what?" Template
```
Decision: [What you're considering]

Immediate effect (1st order):
→

What happens next (2nd order):
→

What happens after that (3rd order):
→

Unintended consequences to watch:
•
•

Does this still look like a good idea?
```

## 10/10/10 Framework (Suzy Welch)

```
How will I feel about this decision:
- In 10 minutes? (immediate, emotional)
- In 10 months? (medium-term, practical)
- In 10 years? (long-term, values)

Best for: Personal/career decisions, not technical ones

Example: Should I speak up about a process problem in the team meeting?
10 min: Uncomfortable, anxiety
10 months: Either problem persists OR it improved
10 years: Almost certainly glad I raised it

→ Speak up
```

## Regret Minimization Framework (Bezos)

```
Project yourself to age 80, looking back at this decision.
Which choice will you regret less?

Best for: Career choices, entrepreneurial bets, life decisions
Not for: Tactical/operational decisions

Script: "When I'm 80 and looking back at my life, 
will I regret not having tried [X]?"
```

## Avoiding Decision Traps

### Common Cognitive Biases in Decisions
| Bias | Description | Counter |
|------|-------------|---------|
| **Anchoring** | Over-weighting first number heard | Generate independent estimates before discussing |
| **Confirmation bias** | Seeking data that confirms existing view | Assign devil's advocate role |
| **Sunk cost** | "We've already invested so much" | Ask: "If we hadn't invested anything, would we start now?" |
| **Status quo bias** | Default to doing nothing | Evaluate inaction as a choice with its own consequences |
| **Availability heuristic** | Overweighting vivid recent examples | Seek base rates and statistical data |
| **Planning fallacy** | Underestimating time/cost/risk | Use reference class forecasting (how long do similar projects take?) |

### Devil's Advocate Protocol
```
For important decisions, explicitly assign someone to argue the opposite:

"[Name], your job for the next 10 minutes is to argue 
 against this proposal as strongly as possible. 
 We're not judging your actual opinion — we want the best arguments against."

This separates role from person and reduces social pressure
to conform to apparent consensus.
```

## Decision Documentation

### Decision Record Template
```
Decision: [What was decided]
Date: [When]
Decider: [Who had final call]

Context: [Why this decision needed to be made]

Options considered:
1. [Option A] — pros/cons
2. [Option B] — pros/cons
3. [Status quo] — always list this

Decision rationale: [Why this option was chosen]

Tradeoffs accepted: [What we're giving up]

Success criteria: [How we'll know if it worked]

Review date: [When to revisit — especially for Type 1 decisions]
```