Core Thinking Models
Model 1: First Principles Thinking
One-sentence definition: Strip away all analogies and conventions, return to the factual foundation at the physics level, and derive solutions from zero.
Applicable scenarios：
- When facing industry "impossible" consensus ("batteries are too expensive" "rockets cannot be reused")
- Cost structure analysis—why does this thing "have to" be this expensive?
- Disruptive product/technology design
- When a new entrant in any field wants to break the existing landscape
Execution steps：
- List all assumptions: Write down the things "everyone knows"
- Question each one: Is this a law of physics, or a human convention?
- Break down to basic facts: What is the material cost of the battery? What is the raw material price of aluminum-lithium alloy?
- Rebuild from the foundation: If you had only these basic facts, how would you design it?
- Compare the differences: Where is the gap between the new plan and the existing plan? What can be eliminated?
Classic case: SpaceX reduced rocket launch costs from about $600 million to about $60 million. The core method: the raw material cost of a rocket accounts for only about 2% of the total cost—the other 98% is manufacturing processes, single-use, and supply chain markups. If it can be recovered and reused, the cost can drop a hundredfold.
Model 2: Physics-First Bottom Line
One-sentence definition: Use the basic laws of physics as insurmountable boundary conditions, and seek the optimal solution within these boundaries.
Applicable scenarios：
- Technical feasibility assessment
- Competitor analysis—are they approaching the physical limit?
- Investment decisions—does this technology direction have a future in physics?
Execution steps：
- Determine the core physical constraints in the field (energy density, thermodynamic limits, speed-of-light latency, etc.)
- Calculate the theoretical optimum
- Assess the percentage of the current technology's distance from the theoretical optimum
- Judge whether the room for improvement is worth investing in
- If it is still far from the physical limit, it means there is huge room for improvement—this is the opportunity
Classic case: When Musk assessed the feasibility of electric vehicles, he did not look at "how existing electric vehicles are" but calculated the theoretical energy density of lithium-ion batteries and the theoretical efficiency of motors, and found that it is completely feasible in physics—it is only an engineering problem.
Model 3: Probabilistic Path Planning
One-sentence definition: Accept the high probability of failure, but ensure that the payoff from success is enough to cover all failure costs.
Applicable scenarios：
- High-risk innovation projects
- Pursuing multiple lines in parallel under limited resources
- Strategic planning for startups
Execution steps：
- List all critical paths to achieving the goal
- Assess the probability of success for each path
- Assess the cost of each path (time, money, opportunity cost)
- Ensure that even if most fail, the successful path can cover the entire investment
- Pursue multiple paths in parallel, and cut branches that have been proven to be dead ends at any time
Classic Case: SpaceX's first three Falcon 1 launches all failed, and Musk had only enough funds left for one last attempt. He accepted a 90% probability of failure, but success would make SpaceX worth tens of billions—probability × payoff = positive expected value.
Model 4: Extreme Time Compression
One-Sentence Definition: Parallelize everything that can be parallelized, trade organizational capability for time, and turn an "impossible deadline" into reality.
Applicable Scenarios：
- Compressing product development cycles
- Catch-up strategies when competitors are faster than you
- Crisis response and emergency handling
Execution Steps：
- Identify which steps in a serial process can be parallelized
- Accept the rework risk brought by parallelism (rework is cheaper than waiting)
- Set extremely short iteration cycles (daily rather than weekly)
- Daily stand-ups, daily decisions, eliminating time spent waiting for approvals
- Tolerate an "inelegant" first version—get it running first, then optimize
Classic Case: During Tesla Model 3's production hell, Musk slept directly in the factory, compressing what would normally be 12 months of production line debugging into a few months, and used "anyone can communicate directly with anyone" to break down organizational hierarchy.
Model 5: Vertical Integration & Control
One-Sentence Definition: Core links must be controlled in-house, and only non-core links should be outsourced—when the external supply chain becomes a bottleneck, do it yourself.
Applicable Scenarios：
- Supply chain bottleneck analysis
- Build vs Buy decisions
- Resource allocation for startups
Execution Steps：
- List all key links of the product/service
- Mark which links are monopolized by a small number of suppliers
- Mark which links have unreasonable cost structures
- For monopolized/unreasonable links: evaluate the cost and time of building in-house
- If the long-term ROI of building in-house > being continuously bottlenecked for 2 years, then build in-house
Classic Case: Tesla developed its own FSD chip, 4680 battery, and seats—because external suppliers were either too slow, too expensive, or did not follow Tesla's timeline. SpaceX developed its own engines, avionics, and software—because the aerospace supply chain operates with a "government contract mindset" and cannot keep up with commercial pace.
Decision Framework
收到问题/机会
│
▼
[第1层：物理底线检查]
这个方向在物理上可行吗？理论极限在哪？
│
├─ 不可行 → 放弃，不浪费时间
│
▼
[第2层：第一性原理分析]
现有方案为什么这样做？哪些是必要的，哪些是惯性？
│
▼
[第3层：10倍改进空间评估]
这个方向能比现有方案好10倍吗？
（好10%不值得投入——增量改进是别人的事）
│
├─ 不能 → 降低优先级或放弃
│
▼
[第4层：关键路径与瓶颈识别]
从当前到目标，最大的瓶颈是什么？
│
▼
[第5层：资源集中攻击]
把最优秀的资源集中到瓶颈上，其他一切让路
│
▼
[第6层：极限迭代]
最短周期内完成→测试→反馈→改进，循环直到成功
Decision Principles：
- Look at the downside, not the upside: First list all possible reasons for failure, and eliminate them one by one
- Shortest Path: If you drop a ball from the 50th floor, what is the fastest way for it to hit the ground? Follow that shortest path
- Changing people is not as good as changing methods: If the method is wrong, changing people is useless. If the method is right, ordinary people can also produce extraordinary results
- Delayed gratification but rapid validation: The grand vision can be far away, but the validation loop must be very short
Classic Quotes
-
"I think it's possible for ordinary people to choose to be extraordinary." — USC Marshall School of Business commencement speech, 2014
-
"When something is important enough, you do it even if the odds are not in your favor." — 60 Minutes interview, 2012 (on persistence after SpaceX's first three launch failures)
-
"I think it is possible for physics to be the guide to the future." — Royal Astronomical Society speech, on first-principles thinking
-
"Persistence is very important. You should not give up unless you are forced to give up." — Repeatedly mentioned in multiple interviews
-
"If you get up in the morning and think the future is going to be better, it is a bright day. Otherwise, it's not." — A declaration of optimism in multiple interviews
-
"I'd rather be optimistic and wrong than pessimistic and right." — Discussion about climate change and sustainable energy
-
"Some people don't like change, but you need to embrace change if the alternative is disaster." — Argument about the necessity of the sustainable energy transition
-
"Failure is an option here. If things are not failing, you are not innovating enough." — SpaceX's internal cultural principle, and also the conceptual basis for SpaceX accepting launch failures
Practical Templates
Template 1: First-Principles Cost Analysis
## 第一性原理分析：[问题/产品]
### 1. 行业共识（需要被质疑的假设）
- 假设1：[描述]
- 假设2：[描述]
- 假设3：[描述]
### 2. 拆解到基本事实
| 组件 | 市场价格 | 原材料/基本成本 | 加价比例 | 加价原因 |
|------|---------|---------------|---------|---------|
| [A] | ¥X | ¥Y | Z% | [原因] |
### 3. 物理底线
- 理论最优值：[计算]
- 当前最优值：[数据]
- 差距百分比：[X%]
- 结论：[是否值得投入]
### 4. 从基座重建的方案
- 如果只有基本原材料和物理定律，我会怎么设计？
- [方案描述]
### 5. 可行性评估
- 技术：[可行/需要突破/不可行]
- 成本：[预算]
- 时间：[估算]
- 风险：[列出]
Template 2: 10x Improvement Challenge
## 10倍改进挑战：[领域/产品]
### 当前最佳
- 指标：[当前数值]
- 瓶颈：[什么在限制它]
### 10倍目标
- 目标值：[当前×10]
- 为什么现在做不到：[列出所有原因]
### 逐条攻克
| 限制因素 | 性质 | 攻克方案 | 预期改进 |
|---------|------|---------|---------|
| [因素1] | 物理/工程/惯例 | [方案] | [X倍] |
### 剩余差距
- 还差 [X倍]，方案：[描述]
### 结论
- [可行/条件性可行/不可行]
Template 3: Risk Path Matrix
## 风险路径矩阵：[项目名]
### 路径列表
| # | 路径描述 | 成功概率 | 成本 | 成功回报 | 期望值 |
|---|---------|---------|------|---------|-------|
| 1 | [描述] | P% | $C | $R | P×R-C |
### 关键依赖
- 路径1依赖：[条件]
- 路径2依赖：[条件]
### 砍线标准
- 如果 [条件] 在 [日期] 前未满足 → 砍掉路径[N]
### 总体评估
- 全部失败的最坏情况：[描述]
- 部分成功的保底情况：[描述]
- 全部成功的理想情况：[描述]
- 总期望值：正/负 → [继续/放弃]
Application Scenarios
When to activate this skill：
- Cost structure analysis: When you need to understand how much a product/service is "actually worth"
- Disruptive innovation assessment: When evaluating a technology direction that looks "impossible"
- Bottleneck breakthrough: When a project is stuck at a key bottleneck and you need to concentrate firepower to break through it
- Rapid prototyping/iteration: When you need to quickly validate an idea from 0 to 1
- Supply chain/outsourcing decisions: When evaluating "do it yourself or buy from others"
- Risk decisions: When facing decisions with a high probability of failure but high returns
- Resource allocation under time pressure: When the deadline is very tight and extreme compression is needed
Typical trigger situations：
- "Everyone says this industry cannot be changed"
- "This cost structure can definitely be optimized"
- "Can we be 10 times faster than others"
- "Where is the physical limit of this technology"
Anti-patterns
-
❌ Replacing first principles with analogical reasoning: Musk explicitly opposes "doing what others do, so we do it too"—"You have to build up the reasoning from the first principles rather than by analogy."
-
❌ Satisfaction with incremental improvement: Being 10% better is not the goal; being 10 times better is. "The path to the breakthrough is not incremental."
-
❌ Excessive outsourcing of core links: When external suppliers become the bottleneck, not doing it yourself means giving up control.
-
❌ Not trying because of fear of failure: How many rockets did SpaceX blow up before successfully recovering one? Failure is an essential part of iteration.
-
❌ Long-cycle grand plans: Musk opposes 3-year plans and 5-year plans—the longer the planning cycle, the more assumptions there are, and the farther it deviates from reality. The shortest verifiable cycle is king.
-
❌ Pursuing perfection before release: Musk's style is "release, iterate, fix"—Twitter's rapid iteration philosophy directly influenced Tesla's OTA update strategy.
-
❌ Believing "experts say it cannot be done": Musk has repeatedly faced expert consensus that "rockets cannot be reusable" and "electric vehicles cannot become widespread"—if it is physically feasible, the experts' "cannot" is only an engineering challenge, not a law of physics.
Compiled based on Elon Musk's public interviews, biographies (Walter Isaacson's "Elon Musk", Ashlee Vance's "Elon Musk: Tesla, SpaceX, and the Quest for a Fantastic Future"), TED talks, and public posts on Twitter/X. Last updated: 2026-04-14