mental models are debt until exercised

everyone i know owns the same library now. inversion. second-order thinking. circle of competence. via negativa. skin in the game. the latticework. we can all recite them, thread them, put them in the pinned tweet.

and most of the collectors, me included for years, make exactly the same decisions we would have made anyway.

an unused mental model is debt. it costs storage, it signals wisdom you haven't earned, and the interest it accrues is false confidence, which compounds quietly until a real decision comes due.

knowing-about is not knowing-how

a mental model is a compression of someone else's reps. munger compressed fifty years of decisions into "invert, always invert." when you read that sentence, you receive the compression, not the reps. you know about the model. you cannot yet do the model.

the gap would be harmless if it felt like a gap. it doesn't. fluency is the problem: after the third podcast and the tenth thread, the vocabulary sits so comfortably in your mouth that your brain files it as competence. psychologists call it the fluency illusion; builders should call it what it is in our world: reading the gym program and logging it as a workout.

so the collector walks around with the vocabulary of judgment and none of the judgment, which is strictly worse than having neither, because now the meta-cognition is broken too. you don't know that you don't know. the debt has hidden itself inside the balance sheet.

the tell

one question exposes the whole position. what was the last decision a model changed?

not explained. changed. after-the-fact explanation is the debt compounding, models are spectacular at post-hoc narration, and dressing up your gut call in framework language is the most common use of the entire genre. i mean changed: you were going to do a, you ran the model, it said b, and you did b, against your default, with something at stake.

if you can't name one, you own inventory. i ran this audit on myself a few years ago and the answer was humiliating. shelves of frameworks, and every consequential decision made by the same three instincts i'd had since my first build.

there's a darker version too: the model used backwards. conclusion first, then shopping the library for whichever framework dresses it best. "first principles" is the most-worn costume in that wardrobe. if your models only ever agree with you, they're a legal team.

exercising the position

converting model-debt into model-equity is boring, slow, and completely doable. what worked for me:

• one model, one quarter. pick a single model and force it on every real decision for three months. before any meaningful choice, write one paragraph: "what does inversion actually say here?" the forcing is the point. voluntary use always collapses back into decoration.
• stakes only. deciding what to order for lunch with second-order thinking is theater. only decisions that cost something, money, time, reputation, count as reps. the model has to get load-bearing weight on it or the tissue doesn't grow.
• log the disagreements. the interesting moments are when the model says b and your gut screams a. write those down, decide, and revisit in a quarter: who won, the model or the gut? that record is the only honest measure of whether a model deserves a place in your head. i keep a running score. some famous models lose to my gut consistently, which means for me, in my domain, they're wallpaper.
• delete ruthlessly. a toolkit of four models you actually swing beats the encyclopedia every time. the encyclopedia was never for deciding. it was for tweeting.

collecting is the search process

"you can't know in advance which model you'll need. broad exposure is how you find the ones that fit. collecting is the search process."

partially conceded. browsing widely is fine, useful even. the sin is booking reading as training. keep the two accounts separate: exposure is entertainment with occasional upside, exercise is what changes decisions. every hour of the first that gets logged as the second is the debt growing.

and notice who profits from the confusion: an entire content economy runs on selling you the feeling of sharpened judgment, one carousel at a time. the feeling is real. the sharpening isn't. nobody's threads make your decisions. you do, alone, on a tuesday, scared, with incomplete data. that moment is the only exam, and it's closed-book in exactly the way the collecting never prepared you for.

the latticework, honestly

a mental model is a tool. a tool you've never swung under load is a decoration. most people's latticework of mental models is, honestly, wallpaper: intricate, expensive, and structural to nothing.

pick one model. put weight on it. let it fail you a few times and recalibrate. that's a rep. do that for a few years and you'll have something no library can hand you: a small set of models with your own scar tissue on them.