No Free Lunch was proved about search algorithms, but its shape — every one-size-fits-all winner is an illusion; performance comes from the fit — turns up everywhere humans argue about “the best” way to do something. Once you can see the pattern, a hundred tired debates suddenly make sense. This lesson tours the idea’s disguises and then shows how it locks together with two models you already met in What Are Mental Models.
Before you read — take a guess
Guess: what do 'the best diet,' 'the best investing strategy,' and 'the best management style' debates all have in common?
”It depends” — with a theorem behind it
When an expert answers “what’s the best X?” with an infuriating “it depends,” they’re not dodging. They’re stating No Free Lunch in plain English. “It depends” means: the winner changes with the structure of your specific situation, so there’s no context-free answer. The follow-up questions — depends on what? — are where the real expertise lives. Let’s see the pattern wear four costumes.
Diet
There is no single optimal diet for every human. A diet that controls one person’s blood sugar may leave another sluggish; an athlete in training and a desk worker recovering from illness have different “problem structures” (activity, genetics, goals, what they’ll actually stick to). A diet is a method; a body-plus-goal is a problem. The winner depends on the match. Anyone selling the diet for everyone is selling a free lunch.
Investing
No strategy beats the market in every regime. Value investing shines in some decades and lags for years in others; momentum works until it violently doesn’t; holding cash is genius in a crash and a slow bleed in a boom. Each strategy has a prior about how markets will behave, and gets paid when reality matches it. “What’s the best strategy?” has no answer without “in what kind of market, over what horizon, for whose risk tolerance?”
Management
There’s no leadership style that’s best for every team and moment. Tight, directive management can rescue a team in crisis and suffocate a team of experts; hands-off autonomy frees great people and abandons struggling ones. The style is a method; the team-in-its-situation is the problem. Great managers don’t have a style — they read the situation and match.
Study technique
No study method is universally best. Flashcards crush vocabulary and are hopeless for essay structure; re-reading feels productive but fails for problem-solving, where practice testing wins. The best technique depends on the kind of knowledge you’re after — another problem-structure in disguise.
The tell in every case
Whenever a claim reaches for “the best ___ for everyone,” a free lunch is being smuggled in. The honest version always names a condition: best for this body, this market, this team, this kind of material. If there’s no “for,” be suspicious.
State the shared pattern.
Pick the right option for each blank, then check.
In diets, investing, management, and study, there is no best method; the winner depends on the of the specific situation — which is why experts answer 'what's best?' with .
Spot the free-lunch pitch
Here’s where the model earns its keep in daily life: it’s a lie-detector for “one weird trick” claims. Sort the honest, conditional claims from the free-lunch pitches.
Sort each claim: is it an honest 'depends on the situation' claim, or a free-lunch pitch promising a universal winner?
Place each item in the right group.
- "This diet is optimal for everyone, no matter their body or goals."
- "Practice testing works best when the goal is durable recall."
- "In a high-uncertainty crisis, a more directive style often steadies a team."
- "For endurance athletes in heavy training, higher carbs tend to help."
- "This productivity system is the best for every person and every job."
- "This one strategy beats the market in every era."
How it locks into the latticework
No Free Lunch doesn’t sit alone — it’s the theorem underneath two foundational models you already know.
Circle of competence. Your circle is the set of problems whose structure you genuinely understand — where you can tell which method fits. No Free Lunch explains why the circle matters: since no method is universally right, the crucial skill is knowing which problems you can correctly match a method to. Straying outside your circle isn’t dangerous because the problems are hard; it’s dangerous because you can no longer tell whether your favourite tool’s prior fits.
The latticework of models. Why carry many mental models instead of one favourite? Because No Free Lunch guarantees your favourite has problems it’s bad at. A single model is a single prior; it wins on the problems that match it and misleads you on the rest. A latticework is a toolkit of priors — so that whatever structure a new problem has, you’ve got a model whose assumptions fit. No Free Lunch is precisely why the man with one hammer gets in trouble.
The through-line
One method = one prior = good on some problems, bad on others (No Free Lunch). So: know which problems you can judge (circle of competence) and carry many tools for the ones you can (the latticework). Three models, one spine.
How does No Free Lunch explain why a 'man with a hammer' (one favourite model for everything) gets into trouble?
A worked case: the same person, two problems
Watch the matching happen. Sam is deciding how to tackle two very different tasks this week.
- Problem A: memorise 200 medical terms for an exam. The structure is arbitrary paired associations, tested by recall. The method that matches: spaced flashcards. Re-reading the textbook would feel productive and fail.
- Problem B: figure out why a patient’s symptoms don’t add up. The structure is diagnosis under uncertainty, needing many disciplines. The method that matches: a latticework — run it past several models (base rates, physiology, incentives to malinger, measurement error). Flashcards are useless here.
Same person, same week, and the best method flips completely because the problem structure flipped. That flip is No Free Lunch in one life.
Match each model to how No Free Lunch connects to it.
Pick a term, then click its definition.
Recap
Big picture
The many faces of No Free Lunch
- One idea, many disguises
- Everyday debates
- Best diet / strategy / style / study — no universal winner
- "It depends" = the theorem in plain words
- Circle of competence
- Know which problems you can match a tool to
- Latticework of models
- Many priors, so one fits any structure
- Man with a hammer
- One prior forced everywhere — fails off-match
- Everyday debates
So No Free Lunch isn’t only about algorithms — it’s the quiet logic under “it depends,” under your circle of competence, and under why you collect models at all. One lesson left: what No Free Lunch permits. Because there absolutely is a lunch you can eat — you just have to pay for it.