十个可以反复使用的心智模型
Good thinking tools should last a lifetime
What a Mental Model Is
A mental model is not knowledge. It is a way of handling knowledge. Knowledge tells you what something is; a mental model tells you which direction to think when you meet that kind of thing. Its value is not novelty but reusability — a good model can be called upon for decades.
The ten below are the ones I think are worth keeping permanently. They do not compete with one another; more often they need to be used together.
1. Second-Order Effects
Every action has consequences, and consequences have consequences. First-order effects are usually obvious; second-order effects are often the opposite. When you decide something, ask one more question: "And then what?" Many clever choices fail at the second step.
2. Incentives
To understand someone's behavior, look first at the incentives they face. Before calling a choice irrational, check whether the incentives actually point where you assume. Most "irrational" behavior is rational inside the incentive structure the person actually inhabits.
3. Compounding
Compounding is not only about money. Knowledge, trust, health, and reputation all accumulate the same way. Its counterintuitive feature is that early progress is almost invisible while later progress is exponential. That is why most people quit too soon.
4. Marginal Versus Total
The total tells you where you are; the marginal tells you where you are going. Many errors come from confusing the two: a large industry is not the same as an industry with large growth.
5. Regression to the Mean
Extreme outcomes tend to drift back toward the average. When something looks too good or too bad to be true, suspect randomness first. This model is an antidote to over-attribution.
Before the result arrives, ask yourself: does this look more like skill, or more like luck?
6. Opportunity Cost
Every choice means giving up the others. The real question is never "is this worth it" but "is this worth it compared with the best thing I could otherwise do."
7. Inversion
Instead of asking how to succeed, ask how to fail for certain — then avoid those paths. Avoiding stupidity is often more reliable than pursuing brilliance, because the paths to failure are usually easier to recognize.
8. Systems and Feedback
Many problems recur because they are part of a system's structure rather than anyone's mistake. When a problem keeps returning, shift attention from "who erred" to "what structure is producing this result."
9. Circle of Competence
Knowing what you understand matters more than understanding a lot. The key is not the size of the circle but whether you can honestly state its edge. Stay inside and you look conservative; wander far beyond and you pay for it.
10. Reversible and Irreversible Decisions
Some decisions can be undone; others cannot. For reversible ones, fast iteration beats long deliberation. For irreversible ones, slowing down and gathering information is the responsible move. Treating the two as the same is the root of many mistakes.
A Note on Combination
These models are more useful in combination than alone. Opportunity cost and compounding pull in different directions: one pushes you to abandon anything with a weak marginal return, the other warns that abandoning too early destroys the very thing compounding needs. Inversion and second-order effects work well together, because the fastest way to avoid a bad second step is to imagine the failure first. The point is not to apply a single model everywhere, but to notice when two of them disagree — that disagreement is usually where the real thinking begins.
How to Actually Use Them
Models are worth using, not memorizing. Do not try to hold all ten at once. Pick one or two and deliberately apply them to decisions over the next month. Only after a few real uses does a model become your own tool.
Mental models are not for appearing clever. They are for holding the line when it counts.
None of these ten is a new invention. They are worth keeping precisely because they are old — old enough to have been tested by many people, many times, and to have survived the test.
