Mental Models for Engineers

The more experienced an engineer, the less time they spend writing code - and the more time they spend thinking. Mental models are what separates 'someone who just executes' from 'someone who understands what and why.'

• Elon Musk uses First Principles to reduce rocket costs.
• Charlie Munger (Berkshire Hathaway) uses a latticework of 100+ mental models for investment decisions.
• Inversion is core to Amazon's pre-mortem / working-backwards process.

Предварительные знания

• Argumentation: Logic vs Rhetoric

First Principles: Think From the Ground Up

**First Principles Thinking** - break a problem down to basic, indisputable truths and rebuild the solution from scratch, without relying on analogies or 'that's how it's always been done.' The method of Descartes, Aristotle, and Musk. The opposite: **reasoning by analogy** - 'it's like X.'

**Three First Principles questions:** 1. What is known for certain (indisputable facts)? 2. What has been assumed to be true that is actually an analogy or assumption? 3. When building from scratch knowing only #1, where does the reasoning lead? The hardest step is seeing assumptions AS assumptions rather than facts.

Inversion and Second-Order Effects

**Inversion:** instead of asking 'how do I achieve success?' ask 'what would guarantee failure?' and avoid those things. It's often easier to prevent failure than to optimize for success. **Second-Order Effects:** 'And what happens after the consequence of our decision?' The first-order effect is obvious; the second and third orders hide the unexpected consequences.

**The Law of Unintended Consequences:** every decision has second- and third-order effects that often outweigh the obvious first-order benefits. Add a feature → scope increases → time-to-market slows → team burns out. Invert it: what could go wrong with this feature?

Extended Toolkit: Occam, Hanlon, Maps

Several additional models that form an engineer's daily decision-making toolkit.

**How to apply mental models in practice?** Applying all of them at once is counterproductive. Choosing 2-3 most relevant to the current problem is sufficient. Over time they become automatic patterns of thought - the 'muscle memory' of reasoning.

Key Ideas

• First Principles: break down to indisputable truths, discard analogies, rebuild
• Inversion: instead of 'how to achieve X' ask 'what will definitely prevent X'
• Second-Order Effects: consequences of consequences often matter more than the initial effect
• Occam: the simplest explanation is the starting point for diagnosis
• Hanlon: error explains more than malice
• Map ≠ Territory: models simplify reality; know the limits of the abstractions in use

What's Next

Mental models are theory. The next step is applying them to real technical decisions - studying cases where they were applied or ignored, with concrete consequences.

• Analyzing Real Technical Decisions — Case studies: how mental models apply to real architectural decisions

Вопросы для размышления

• Take the last architectural decision the team made. Were second-order effects checked? What was missed?
• What is a good 'Occam's Razor' heuristic for debugging? What should be checked first at any incident?
• Where are the boundaries of the Circle of Competence for a senior engineer? What is well-known, and where should external expertise be sought?

Связанные уроки

• st-01-feedback-loops