mcpField
F
first-principles

first-principles

github.com/guia-matthieu生产力150389 次安装4 周前更新
代码仓库

关于 first-principles

面向复杂问题、成本分析和战略选择的推理技能。它区分基本约束与惯例,再从经过核实的基础事实出发构建可行方案。

主要功能

  1. 01审视既有假设
  2. 02区分物理规律与惯例
  3. 03拆解问题的基本组成
  4. 04从基本原理重建方案
  5. 05验证并迭代分析

适用场景

  1. 01重新评估行业普遍接受的假设
  2. 02分析产品或服务成本
  3. 03探索看似无解问题的可能方案

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 first-principles

安装后在 Claude Code 中输入/first-principles确认已加载。

详细介绍

第一性原理思维

将复杂问题拆解为基本真理,然后从这些真理开始推理。掌握由 Aristotle 提出、经 Elon Musk 现代化改造的古老方法,解决看似不可能的问题。

何时使用此技能

  • 面对“不可能”的问题,常规解决方案行不通时
  • 挑战行业假设,尤其是人人都视为真理的假设
  • 进行定价或成本分析,寻找大幅节省成本的方法(如 Musk 对电池的分析)
  • 进行产品创新,迭代现有解决方案仍不够时
  • 做出战略决策,类比其他公司可能产生误导时
  • 突破思维障碍,陷入常规思维时

方法论基础

方面详情
来源Aristotle(公元前 384—322 年),由 Elon Musk 现代化改造
专家Aristotle 称之为“认识事物的第一依据”;Musk 将其应用于 SpaceX、Tesla 及他的每一项事业
核心原则“把事情归结为最基本的真理,然后问:‘我们确定哪些事情是真的?’……接着从那里开始推理。”——Elon Musk

Claude 的工作与由你决定的事项

“Claude 梳理拆解过程。由你挑战假设。”

Claude 负责你提供
系统地列出假设用领域专业知识验证或否定假设
拆解出原子组成部分真实世界的成本和限制数据
运用物理定律与惯例测试判断哪些因素真正不可改变
梳理从基本原理出发重新构建的过程决定要采取哪条路径的战略判断
生成新颖的解决方案框架可行性评估和实施

请记住:第一性原理是一种思维工具。Claude 加快梳理过程;突破性洞见来自你挑战一切的意愿。


此技能的作用

  1. 解构假设——辨别你认为是真的事情与实际为真的事情
  2. 揭示隐藏约束——找出伪装成定律的人为限制
  3. 实现突破性解决方案——开辟类比思维者看不到的路径
  4. 大幅降低成本——找出任何事物的真实最低价格
  5. 建立原创见解——得出无法通过照搬他人而获得的结论

使用方法

挑战传统观念
Apply First Principles Thinking to challenge this assumption:
[describe the "truth" everyone accepts in your industry]

Break it down to fundamental truths and show me what's actually possible.
分析成本或定价
Use First Principles to analyze the true cost of:
[product, service, or process]

What are the raw components? What should this actually cost?
解决“不可能”的问题
I'm told [problem] is impossible because [reasons].
Apply First Principles to find a path forward.
What fundamental truths apply here? What assumptions can we challenge?

说明

运用第一性原理思维时,请遵循以下系统化流程:

步骤 1:明确问题和当前假设
## Current State Analysis

**Problem:** [What are you trying to solve?]

**Current "Truth":** [What does everyone believe about this?]

**Analogies in Use:** [How do people currently reason about this?]
- "It's always been done this way"
- "Company X does it like this"
- "Industry standard is..."

**Hidden Assumptions:** (List everything assumed to be true)
1.
2.
3.
4.
5.

关键问题:

  • “专家”认为哪些事情不可能或不可改变?
  • 对于这个问题,你从未质疑过什么?
  • 完全的局外人会觉得你的哪些假设很奇怪?

步骤 2:拆解为基本真理
## First Principles Breakdown

For each assumption, ask: "Is this a LAW OF NATURE or a CONVENTION?"

| Assumption | Type | Evidence |
|------------|------|----------|
| [Assumption 1] | Law / Convention | [Why?] |
| [Assumption 2] | Law / Convention | [Why?] |
| [Assumption 3] | Law / Convention | [Why?] |

**Laws of Nature** (Cannot be changed):
- Physics (gravity, thermodynamics, etc.)
- Mathematics (proven theorems)
- Biology (human needs)

**Conventions** (CAN be changed):
- Industry practices
- "Best practices"
- Historical precedent
- Regulatory (can be challenged)
- Social norms

物理定律测试: 问:“物理定律是否会阻止这件事?”如果不会,它就是可改变的。

Elon Musk 的 3 步流程:

  1. 确定并定义当前假设
  2. 将问题拆解为基本原理
  3. 从头开始创建新的解决方案

步骤 3:找出原子组成部分
## Atomic Breakdown

**What is this thing MADE OF at the most basic level?**

Example: Battery Pack
├── Lithium: $X per kg
├── Cobalt: $X per kg
├── Nickel: $X per kg
├── Aluminum: $X per kg
├── Steel: $X per kg
├── Polymers: $X per kg
└── Manufacturing energy: $X

**Raw Material Cost:** $___
**Current Market Price:** $___
**Gap to Explain:** $___

The gap represents: convention, inefficiency, margin, or artificial scarcity

对于非物理问题:

  • 所需的原子级操作是什么?
  • 所需的原子级输入是什么?
  • 真正的原子级约束是什么?

步骤 4:从基本原理开始推理
## Rebuild from First Principles

Starting ONLY from verified truths, ask:

1. **What is physically possible?**
   [Based on laws of nature]

2. **What is the theoretical minimum?**
   [Cost, time, resources, steps]

3. **What new combination could work?**
   [Ignoring how it's "supposed" to be done]

4. **What would we do if starting from scratch?**
   [No legacy constraints]

## New Solution Design

**Fundamental Requirements:**
- [Only TRUE requirements]

**Eliminated "Requirements":**
- [Things assumed necessary but aren't]

**Novel Approach:**
[Description of solution built from first principles]

步骤 5:验证并迭代
## Validation Checklist

- [ ] Solution respects actual laws of nature
- [ ] No hidden conventional assumptions snuck back in
- [ ] Can explain WHY each element is necessary
- [ ] Theoretical minimum is understood
- [ ] Gap between current and theoretical is explained

## Implementation Path

**What makes this hard (real obstacles):**
1.

**What makes this SEEM hard (false obstacles):**
1.

**First experiment to validate:**
[Smallest test of the core assumption]

示例

示例 1:SpaceX 降低火箭成本

背景: Elon Musk 想为火星任务购买火箭。报价:每枚火箭 6,500 万美元。

输入:

“大家都说火箭要花 6,500 万美元,成本就是这么高。这是真的吗?请运用第一性原理。”

输出:

现状分析:

假设质疑
火箭很贵惯例——基于历史定价
必须向成熟供应商购买惯例——也可以自己制造
火箭是一次性的惯例——飞机并非一次性用品
航空航天产品必须定高价惯例——这不是物理定律

第一性原理拆解:

“火箭由什么构成?”

  • 航空航天级铝
  • 钛
  • 铜
  • 碳纤维
  • 燃料(煤油、液氧)

原子级成本分析:

Raw materials for Falcon 9:
├── Aluminum alloys: ~$35,000
├── Titanium: ~$15,000
├── Carbon fiber: ~$20,000
├── Avionics components: ~$100,000
├── Engines (9x Merlin): ~$1M (at scale)
└── Other materials: ~$30,000

Theoretical material cost: ~$1.2M
Industry price: $65M
Gap: 54x markup

逐步推导:

  1. **物理规律允许:**火箭可以重复使用(飞机可以降落后再次飞行)
  2. **最低成本:**材料成本约 120 万美元,加上制造和利润,成本有望达到 500 万至 1,000 万美元
  3. 全新方案:
    • 内部制造(消除承包商利润)
    • 按可重复使用的方式设计(摒弃一次性使用的假设)
    • 垂直整合(掌控供应链)

**结果:**Falcon 9 目前每次发射成本约为 2,800 万美元,可重复使用使边际成本降至约 1,500 万美元。其他公司仍收取 1 亿美元以上。

**为何有效:**Musk 没有把“火箭要花 6,500 万美元”当作事实。他发现这只是低效做法形成的惯例,而非物理规律。


示例 2:重新思考员工培训成本

**背景:**公司每年为每位员工花费 2,000 美元进行培训。CFO 说“这是行业标准”。

输入:

“我们每位员工的培训支出是 2,000 美元。HR 说这很正常。真的是这样吗?请用第一性原理分析。”

输出:

现状分析:

假设类型
培训必须由讲师授课惯例
必须使用外部供应商惯例
必须同步进行惯例
认证必须参加付费课程惯例
培训 = 课堂学习时间惯例

原子级拆解——培训实际需要什么?

Training = Transfer of Knowledge + Skill Practice + Verification

├── Knowledge Transfer
│   ├── Content (can be recorded once)
│   ├── Delivery (can be async/digital)
│   └── Q&A (can be AI-assisted or peer)
│
├── Skill Practice
│   ├── Exercises (can be self-paced)
│   ├── Feedback (can be automated or peer)
│   └── Repetition (individual, no instructor needed)
│
└── Verification
    ├── Testing (automated)
    └── Observation (sampling, not 100%)

成本明细:

Current $2,000/person:
├── Vendor fees: $800 (40%)
├── Instructor time: $500 (25%)
├── Travel/facilities: $400 (20%)
├── Materials: $200 (10%)
└── Admin overhead: $100 (5%)

First Principles minimum:
├── One-time content creation: $50/person (amortized)
├── Platform/LMS: $30/person/year
├── Peer mentoring time: $100/person
├── Verification tools: $20/person
└── Buffer: $50/person

Theoretical minimum: ~$250/person

逐步推导——新模式:

  1. **内容只需制作一次:**录制优秀培训师讲解每个主题的内容,成本在 5 年内分摊。
  2. **异步交付:**员工按自己的节奏学习,无需承担排期成本
  3. **AI 问答:**Claude/GPT 处理 80% 的问题
  4. **同伴练习:**员工一起练习(免费)
  5. **自动验证:**数字化评估,并抽查实际操作

新方案:

  • 第 1 年:每人 500 美元(内容制作投入)
  • 第 2 年及以后:每人 250 美元(持续降低 75%)

验证:

  • 仍能传递知识吗?是
  • 仍能培养技能吗?是(同伴练习)
  • 仍能验证学习效果吗?是(自动化 + 抽样)
  • 符合学习规律吗?是

清单与模板

第一性原理分析模板
## First Principles Analysis: [Topic]

### 1. THE ASSUMPTION AUDIT

**What "everyone knows" about this:**
1.
2.
3.

**Source of these beliefs:**
- [ ] Physics/Math (unchangeable)
- [ ] Biology (unchangeable)
- [ ] Regulation (changeable but hard)
- [ ] Industry practice (changeable)
- [ ] Historical precedent (changeable)
- [ ] Someone told me (verify)

### 2. THE ATOMIC BREAKDOWN

**Physical components:**
- Component 1: [cost/quantity]
- Component 2: [cost/quantity]
- ...

**Required actions (minimum):**
1.
2.
3.

**True constraints (laws of nature):**
-

**False constraints (conventions):**
-

### 3. THE REBUILD

**If we started from scratch with only true constraints:**

**Theoretical minimum:** [cost/time/resources]

**Current state:** [cost/time/resources]

**Gap:** [X%]

**Gap explained by:**
- [ ] Inefficiency
- [ ] Margin/profit
- [ ] Convention
- [ ] Lack of innovation
- [ ] True complexity (irreducible)

### 4. THE NEW SOLUTION

**Core insight:**

**New approach:**

**First experiment to test:**

**What could go wrong:**

第一性原理快速检查清单
## Before Accepting Any "Truth"

□ Is this a law of physics/math/biology?
□ Or is it "how things are done"?
□ Who benefits from this being "true"?
□ What would a complete outsider think?
□ Has anyone ever done it differently?
□ What would this look like if we started fresh?

## The Musk Test

□ Can I explain the atomic components?
□ Do I know the theoretical minimum cost?
□ Is there a 10x gap I can't explain?
□ Am I reasoning by analogy or from fundamentals?

常见的第一性原理应用领域
## Where to Apply First Principles

### COST REDUCTION
- What are the raw materials?
- What is labor actually required?
- What steps can be eliminated?

### PRODUCT DESIGN
- What does the user fundamentally need?
- What is convention vs. requirement?
- What would this look like in 10 years?

### PROCESS IMPROVEMENT
- What are the required steps (physics)?
- What are the assumed steps (convention)?
- What is the theoretical minimum time?

### PRICING
- What is the cost to deliver?
- What is convention in pricing?
- What would a new entrant charge?

### STRATEGY
- What does the customer fundamentally want?
- What is assumed about competition?
- What constraints are real vs. imagined?

警示信号:何时你并未使用第一性原理
## Warning Signs

You're reasoning by ANALOGY (not First Principles) when you say:

- [ ] "Industry standard is..."
- [ ] "Competitors do it this way..."
- [ ] "We've always done it like this..."
- [ ] "Best practice says..."
- [ ] "Experts recommend..."
- [ ] "It's common knowledge that..."
- [ ] "That's just how it works..."

None of these are REASONS. They're appeals to convention.

## The Fix

For each statement above, ask:
"But WHY? What fundamental truth makes this necessary?"

If you can't answer with physics, math, or biology - it's changeable.

技能边界(识别适用范围)

此技能适用于:
  • 成本削减分析(找出“真实底价”)
  • 质疑人人接受的行业假设
  • 突破渐进式改进的产品创新
  • 类比可能误导的战略决策
  • 定价策略(了解事物的合理成本)
此技能不适用于:
  • 需要快速决策的情况,速度比深入分析更重要 → 改用启发式方法
  • 执行任务,实施路径已经明确 → 使用第一性原理分析过度了
  • 情绪或人际关系问题 → 并非所有问题都能归结为物理规律
  • 确实无法改变的监管限制 → 接受限制并在其范围内开展工作
  • 纯创意工作,逻辑并非瓶颈 → 改用其他创意工具
质量检查要点

在接受分析之前,请确认:

  • 你确实识别了真正的替代方案(而不只是直接竞争对手)
  • “物理定律”确实是物理规律,而不只是难以改变的惯例
  • 原子级成本拆解使用了真实的市场数据
  • 创新方案没有偷偷带回传统假设
  • 你有办法在投入之前验证第一个洞见

迭代指南

“第一性原理是一场对话,而不是一次性分析。”

推荐迭代模式
轮次重点要提出的问题
第 1 轮罗列假设“列出这个问题中的所有假设”
第 2 轮挑战假设“对于每个假设,它是物理规律还是惯例?”
第 3 轮原子级拆解“它实际上由什么构成/实际需要什么?”
第 4 轮重新构建“从仅有的真实约束出发,有哪些可能?”
实用的后续提示词

首次分析后,可以尝试:

  • “我认为[假设]实际上是物理规律,因为[理由]。说服我事实并非如此。”
  • “原子级拆解遗漏了[组成部分]。把它纳入后重新分析。”
  • “大家都把[约束]视为既定事实。继续深入探究它是否真的不可改变。”
  • “这个‘创新方案’感觉仍然很传统。还有哪些完全不同的方案?”

学习曲线

使用次数你将体验到什么
第 1 次系统地拆解假设,了解这种方法
第 3 次你会发现自己在用类比推理,并及时停下来
第 10 次面对“不可能”的问题时,第一性原理将成为一种本能

专业提示:当你愿意冒着显得愚蠢的风险去质疑“显而易见”的事实时,收获最大。如果大家早就知道某件事可以改变,那它就不是第一性原理洞见。


参考资料

  • Aristotle. “Posterior Analytics”(约公元前 350 年)——第一性原理概念的起源
  • Musk, Elon. 关于 SpaceX 方法论的各类访谈(2012 年至今)
  • Clear, James. “First Principles: Elon Musk on the Power of Thinking for Yourself”
  • Farnam Street. “First Principles: The Building Blocks of True Knowledge”
  • Munger, Charlie. “Elementary Worldly Wisdom”——相邻心智模型思维

相关技能


技能元数据

  • 模式:cyborg
name: first-principles
category: strategy
subcategory: problem-solving
version: 2.0
author: GUIA
source_expert: Aristotle, Elon Musk
source_work: Posterior Analytics (Aristotle), SpaceX methodology (Musk)
difficulty: intermediate
mode: both  # Both = can be structured (Centaur) or exploratory (Cyborg) depending on use
estimated_value: $2,000 strategy consulting session
tags: [problem-solving, innovation, cost-reduction, assumptions, Musk, thinking]
created: 2026-01-25
updated: 2026-01-28

此技能属于 GUIA Premium Marketing Skills Library——连接 AI 基础知识与技术实现的 201 层。

评论

支持 Markdown