mcpField
R
reversible-decisions

reversible-decisions

github.com/guia-matthieu生产力150211 次安装4 周前更新
代码仓库

关于 reversible-decisions

这是一项基于 Jeff Bezos Type 1 与 Type 2 框架的决策技能,适合个人和团队。它区分难以撤回且影响重大的决策与可恢复的决策,并据此匹配决策流程。

主要功能

  1. 01按可逆性和影响分类决策
  2. 02根据风险匹配决策流程
  3. 03识别适合委派的决策
  4. 04减少对可恢复决策的过度分析
  5. 05使用决策清单与模板

适用场景

  1. 01加快低风险产品决策
  2. 02评估定价方案变更
  3. 03明确团队决策职责
  4. 04审视迟迟未决的组织调整

安装

复制下面的命令,在终端中运行。

$npx skills add https://github.com/guia-matthieu/clawfu-skills --skill reversible-decisions

安装后在 Claude Code 中输入/reversible-decisions确认已加载。

详细介绍

可逆决策(Type 1 与 Type 2)

了解何时快速行动,何时谨慎行事。掌握 Jeff Bezos 的框架,区分高风险、不可逆决策与低风险、可逆决策。

何时使用此技能

  • 确定决策优先级,了解该把时间投入在哪里
  • 赋能团队,明确哪些事项可以委派,哪些需要升级处理
  • 避免分析瘫痪,不再在无关紧要的决策上过度纠结
  • 风险管理,识别真正需要谨慎对待的事项
  • 在任何情境下权衡速度与周全程度
  • 在组织中建立决策文化

方法论基础

方面详情
来源Jeff Bezos — Amazon 股东信(2015-2016)
核心原则“有些决策不可逆且影响重大(Type 1)。大多数决策可逆且影响较小(Type 2)。针对不同决策采用合适的流程。”
为何重要大多数人把所有决策都当作 Type 1 来处理——缓慢、审慎,并要求掌握完整信息。这会导致决策瘫痪和错失良机。优秀的决策者会对 Type 2 快速行动,对 Type 1 放慢脚步。

Claude 负责什么,你负责什么

Claude 负责你负责
构建内容框架最终信息传达
提出说服技巧建议品牌语调
创建多个草稿版本选择版本
找出优化机会发布时间
分析竞争对手的做法战略方向

此技能的作用

  1. 对决策进行分类 - 这是 Type 1 还是 Type 2?
  2. 根据风险调整流程 - 针对不同决策采用合适的速度
  3. 便于委派 - Type 2 决策可以下放
  4. 避免过度分析 - 不再把可逆决策当作不可逆决策
  5. 提升组织速度 - 团队在正确的事情上行动更快
  6. 减轻决策疲劳 - 不要在低风险选择上浪费精力

使用方法

对决策进行分类
Help me classify this decision:
[Describe the decision]
Is this Type 1 (irreversible) or Type 2 (reversible)?
What process should I use?
加快决策
I'm spending too much time on [decision].
Apply the Type 1/Type 2 framework to help me move faster.
建立决策流程
Help me design a decision-making framework for my team.
Which decisions should require consensus vs. individual judgment?

说明

第 1 步:了解框架
## Type 1 vs. Type 2 Decisions

### Bezos' Definition

**Type 1: One-Way Doors**
"Some decisions are consequential and irreversible or nearly irreversible—
one-way doors—and these decisions must be made methodically, carefully,
slowly, with great deliberation and consultation."

**Type 2: Two-Way Doors**
"But most decisions aren't like that—they are changeable, reversible—
they're two-way doors. If you've made a suboptimal Type 2 decision,
you don't have to live with the consequences for that long.
You can reopen the door and go back through."

### The Problem

"As organizations get larger, there seems to be a tendency to use
the heavy-weight Type 1 decision-making process on most decisions,
including many Type 2 decisions. The end result of this is slowness,
unthoughtful risk aversion, failure to experiment sufficiently,
and consequently diminished invention."

### The Solution

"We must resist this tendency."

Type 2 decisions should be made quickly by high-judgment individuals
or small groups. Type 1 decisions require the full deliberative process.

第 2 步:分类框架
## How to Classify Decisions

### The Two Questions

**Question 1: Is it reversible?**
Can you undo this decision with reasonable effort and cost?

| Reversibility | Examples |
|---------------|----------|
| **Easily reversible** | Pricing change, A/B test, new feature flag, hire (with trial), campaign |
| **Hard to reverse** | Architecture choice, brand name, key hire (C-level), market exit |
| **Irreversible** | Selling company, shutting down product, firing someone, legal action |

**Question 2: What are the consequences?**
If this decision is wrong, what happens?

| Consequence Level | Examples |
|-------------------|----------|
| **Low** | Internal process change, small experiment, minor feature |
| **Medium** | New product launch, pricing tier, team restructure |
| **High** | Major strategic pivot, large investment, partnership |
| **Existential** | Acquisition, shutdown, bet-the-company move |

### The Matrix

                影响
                低          高

可逆性 ┌────────────┬────────────┐ 高 │ TYPE 2 │ TYPE 2 │ (容易) │ (快速) │ (快速并 │ │ │ 持续监测)│ ├────────────┼────────────┤ 低 │ TYPE 2 │ TYPE 1 │ (困难) │ (谨慎) │ (缓慢) │ └────────────┴────────────┘


### Quick Classification

**TYPE 2 (Move Fast):**
- Can be undone
- Low/medium consequences
- Learning opportunity
- Failure is recoverable
- Most business decisions

**TYPE 1 (Move Carefully):**
- Can't be undone
- High/existential consequences
- Mistakes are permanent
- One-way door
- ~5-10% of decisions

第 3 步:让流程与类型匹配
## Decision Process by Type

### Type 2 Process (70% of Decisions)

**Time:** Hours to days (not weeks)
**Who:** Individual or small group with context
**Information:** Good enough, not perfect
**Approval:** None or single level
**Documentation:** Minimal (decision log)

**The Mantra:**
"Disagree and commit" - If you have 70% of the information you wish you had,
make the decision. Waiting for 90% is usually too slow.

**Process:**
1. Identify it's Type 2 (reversible, recoverable)
2. Gather available information quickly
3. Make the call
4. Communicate the decision
5. Monitor and adjust

**Examples:**
- Feature prioritization
- Hiring most roles
- Process changes
- Pricing experiments
- Marketing campaigns
- Internal tools
- Meeting schedules

---

### Type 1 Process (5-10% of Decisions)

**Time:** Weeks to months
**Who:** Senior leadership, broad input
**Information:** As complete as reasonably possible
**Approval:** Multiple stakeholders
**Documentation:** Thorough (rationale, alternatives, risks)

**The Mantra:**
"Measure twice, cut once" - This is permanent. Get it right.

**Process:**
1. Confirm it's Type 1 (irreversible, consequential)
2. Define decision criteria clearly
3. Gather comprehensive information
4. Consider alternatives thoroughly
5. Consult relevant stakeholders
6. Document the reasoning
7. Make the decision
8. Communicate extensively

**Examples:**
- M&A decisions
- Major strategic pivots
- Leadership hires (C-level)
- Market entry/exit
- Large capital allocation
- Shutting down products
- Legal/regulatory choices

第 4 步:常见陷阱
## Decision-Making Traps

### Trap 1: Treating Type 2 as Type 1

**Symptom:** Analysis paralysis on small decisions
**Example:** 2-week committee review for a landing page change
**Problem:** Slows innovation, frustrates teams, misses opportunities
**Fix:** Ask "What's the worst case if we're wrong? Can we fix it?"

### Trap 2: Treating Type 1 as Type 2

**Symptom:** Moving too fast on irreversible choices
**Example:** Acquiring a company in 2 weeks
**Problem:** Permanent mistakes, existential risk
**Fix:** Ask "If this goes wrong, can we undo it?"

### Trap 3: Requiring Consensus on Type 2

**Symptom:** Everyone needs to agree before action
**Example:** 10-person meeting to decide email copy
**Problem:** Slowest person becomes bottleneck
**Fix:** Empower individuals to make Type 2 calls

### Trap 4: Not Recognizing Type 1 in Disguise

**Symptom:** Missing irreversibility hidden in details
**Example:** "Small" technical choice that creates years of debt
**Problem:** Accumulated Type 1 decisions dressed as Type 2
**Fix:** Consider second-order effects

### Trap 5: Using Decision Type as Excuse

**Symptom:** Calling everything Type 1 to avoid responsibility
**Example:** "We need more research" on every decision
**Problem:** Becomes cover for inaction
**Fix:** Default assumption = Type 2 unless proven otherwise

第 5 步:决策流程模板
## Decision Classification Template

### The Decision
[What are you deciding?]

### Classification Checklist

**Reversibility Assessment:**
□ Can this be undone in days/weeks?
□ What's the cost of reversing?
□ Are there path dependencies that make reversal hard?
□ Will this create commitments we can't exit?

**Consequence Assessment:**
□ If wrong, what's the impact on revenue?
□ If wrong, what's the impact on reputation?
□ If wrong, what's the impact on team?
□ If wrong, is the company at risk?

### Classification: ____________

| Factor | Type 1 Indicator | Type 2 Indicator | This Decision |
|--------|------------------|------------------|---------------|
| Reversibility | Can't undo | Can undo | |
| Time to undo | Months/Never | Days/Weeks | |
| Cost to undo | High/Impossible | Low/Moderate | |
| Consequence | Existential | Recoverable | |
| Frequency | Rare | Common | |

### Process to Use

**If Type 2:**
- Who should decide? _______________
- Timeline: _______________ (days, not weeks)
- Information needed: _______________
- Communication: _______________

**If Type 1:**
- Who needs to be involved? _______________
- Timeline: _______________ (weeks/months)
- What information is missing? _______________
- What alternatives exist? _______________
- What's the documentation plan? _______________

示例

示例 1:产品功能决策

输入:

“我们应该给应用添加深色模式吗?我的团队已经为此争论了两周。”

输出:

决策分析:深色模式功能

分类
因素评估
可逆吗?是 - 可以发布,如果效果不好也可以移除
撤销成本低 - 一个冲刺即可移除
决策错误的后果低 - 用户可以关闭该功能,只会浪费少量精力
决策正确的后果中 - 用户体验更好,也能提高部分用户留存率
结论:TYPE 2(双向门)

为一项可逆的功能决策争论两周,是组织运作失调的表现。

建议流程

允许用时: 最多 2-3 天 决策人: 产品经理 + 1 名工程师 所需信息:

  • 快速用户调查(100 份回复)
  • 竞争分析(2 小时)
  • 工程评估(30 分钟)

决策步骤:

  1. 是否有迹象表明用户需要它?(调查中感兴趣的用户 >30% = 是)
  2. 成本是否合理?(工程投入 <1 周 = 是)
  3. 是否与其他事项冲突?(查看路线图)

如果第 1、2 项为是,且第 3 项为否: 发布。 如果不确定: 在功能标记后发布,并进行衡量。

不要做什么

❌ 开两周的会 ❌ 获得高管批准 ❌ 做完美的用户研究 ❌ 在组织内达成共识 ❌ 等待更多数据

贝佐斯测试

“如果我们上线深色模式,结果发现不合适,会怎样?”

答案:我们会在一个冲刺内撤下它。少数用户会短暂感到困惑。不会造成实际损害。

结论: 做出决定,然后继续推进。把审慎思考的精力用在真正重要的事情上。


示例 2:定价策略变更

输入:

“我们正在考虑从按席位定价改为按用量定价。应该如何推进?”

输出:

决策分析:定价模式变更

分类
因素评估
可逆吗?部分可逆——技术上可以,但是……
撤销成本高——客户困惑、客户流失、内部系统调整
路径依赖高——新模式下的新客户不容易迁移
出错的后果高——影响收入、客户流失、竞争劣势
深入分析

为什么这不完全属于 Type 2:

  • 现有客户期待当前定价
  • 改回去会显得不稳定
  • 系统、合同、销售薪酬都需要更新
  • 市场认知会发生变化

为什么这不完全属于 Type 1:

  • 你可以让现有客户继续使用旧方案
  • 你可以并行运行两种模式
  • 你可以先只对新客户进行测试
结论:TYPE 1.5(需要谨慎处理)

这是一个重大的战略决策,能否撤回存在限制。

建议流程

时间: 分析 4-6 周 参与者: CEO、CFO、产品负责人、销售负责人 所需信息:

  • 两种方案的财务建模
  • 客户访谈(了解他们是否愿意切换)
  • 竞争分析
  • 工程和运营评估
  • 对销售薪酬的影响

风险缓解:

  1. 先只对新客户进行测试(Type 2 实验)
  2. 让现有客户继续使用旧定价(降低不可逆风险)
  3. 运行 6 个月后再迁移任何客户
  4. 为两种模式都搭建系统
决策框架
方案可逆性速度风险
全面切换低快高
仅面向新客户高中中
并行运行两种模式高慢低

建议: 先只面向新客户(Type 2)。验证通过后再迁移现有客户(届时就变成 Type 1)。


示例 3:团队/组织决策

输入:

“我们正在讨论是否要重组工程团队。有人支持小队制,有人支持功能团队。我们已经讨论了一个月。”

输出:

决策分析:团队结构

分类
因素评估
可逆吗?是——重组一直都在发生
撤销成本中——变更期间生产力会下降
出错的后果中——一个季度内交付速度会变慢
路径依赖低——人员可以调换团队
结论:TYPE 2(但感觉像 Type 1)

这件事感觉不可逆,因为会影响到人,而且大家情绪都很强烈。 但实际上呢?团队经常重组。这件事可以补救。

真正的问题

讨论一个月并不是在分析,而是在逃避冲突。

实际情况是:

  • 大家各有看法,谁也不肯让步
  • 没有人愿意做决定并承担“责任”
  • 争论让人感到舒适,做决定却令人不安
建议流程

时间: 最多再用 1 周 决策人: 工程负责人(或承担相应责任的人) 流程:

  1. 将两种方案分别整理成文(每份 1 页)
  2. 明确成功标准(哪些指标会改善?)
  3. 选定一个方案
  4. 承诺执行 6 个月(届时再审视)
  5. “求同存异,全力执行”——持不同意见的人也照样执行
推动决策的机制

“我们会在[周五]前做出决定。谁的意见最坚定,谁就来拍板,并负责让方案取得成效。我们都会承诺支持该决定 6 个月,然后再重新评估。”

Type 2 决策

对团队这样说: “这是一个双向门决策。我们之后可以更改。但我们不能永远争论下去。选一个方案,运行 6 个月,衡量结果,再做调整。最糟糕的结果就是陷入瘫痪。”


清单与模板

快速分类清单
## Is This Type 1 or Type 2?

□ Can we undo this in <30 days?
□ If wrong, will we lose <10% of something important?
□ Is this a common decision (we'll make many like it)?
□ Can we experiment/test before committing?
□ Are the consequences contained?

**Mostly YES → Type 2 (Move fast)**
**Mostly NO → Type 1 (Move carefully)**

### Default Rule
"When in doubt, it's Type 2. Most decisions are."

团队决策矩阵模板
## Team Decision-Making Framework

### Type 2 Decisions (Individual/Small Group)
- Feature prioritization
- Bug fixes
- Process improvements
- Hiring (non-leadership)
- Tool selection
- Meeting schedules
- Internal communications

**Process:** Inform, decide, execute
**Timeline:** Hours to days
**Approval:** None needed

### Type 1 Decisions (Leadership/Broader Input)
- Strategic direction
- Major investments (>$X)
- Leadership hiring
- Pricing strategy
- Market entry/exit
- Partnerships
- Shutting down products

**Process:** Analyze, consult, deliberate, decide
**Timeline:** Weeks
**Approval:** [Define levels]

### Escalation Criteria
Escalate Type 2 to Type 1 if:
- Cost exceeds $[X]
- Affects >N customers
- Creates legal/compliance risk
- Changes company strategy
- Irreversible commitment

技能边界

此技能擅长的方面
  • 构建有说服力的内容
  • 运用文案撰写框架
  • 创建不同的草稿版本
  • 分析竞争对手的做法
此技能无法做到的方面
  • 保证转化率
  • 取代品牌语调的塑造工作
  • 了解你的特定受众
  • 做出最终审批决定

参考资料

  • Bezos, Jeff. “Amazon Shareholder Letters”(2015 年、2016 年)——Type 1/Type 2 框架
  • Blank, Steve. “The Four Steps to the Epiphany”——初创企业的速度
  • Ries, Eric. “The Lean Startup”——可逆实验
  • Farnam Street. “Mental Models”——决策框架
  • Amazon. “Leadership Principles”——偏向行动

相关技能


技能元数据

  • 模式:cyborg
name: reversible-decisions
category: thinking
subcategory: decision-making
version: 1.0
author: MKTG Skills
source_expert: Jeff Bezos
source_work: Amazon Shareholder Letters
difficulty: beginner
estimated_value: $2,000 management consulting session
tags: [decisions, Bezos, Amazon, speed, reversibility, management, delegation]
created: 2026-01-25
updated: 2026-01-25

评论

支持 Markdown