Combining Models: The Latticework

You have spent this whole site collecting models one at a time — inversion in its box, base rates in another, feedback loops, incentives, supply and demand, game theory, each learned, worked, and stress-tested on its own. That was the apprenticeship. This course is the reason for it. Because a real decision does not arrive labelled “this is a base-rate problem” or “apply second-order thinking here.” It arrives as a mess — a bank rumour, a job offer, a viral product, a war — and the expert move is not to reach for your favourite model but to run several over the situation at once and assemble a single judgment from what they collectively say.

This is the capstone, the skill TOPIC.md names as the entire point of the enterprise: reasoning with a latticework. Charlie Munger’s word is exact. Not a list of models, not a toolbox you rummage in, but a latticework — an interlocking mesh of models from different disciplines that you hang your experience on, so that any new fact finds several places to attach and gets read from several angles before you act. A person with one model forces every problem to fit it — Munger’s man with a hammer, to whom everything looks like a nail. A person with a latticework asks, of the same situation, what does the economist see? the biologist? the engineer? the psychologist? — and lets those readings argue.

That argument resolves in exactly three ways, and this course is built around them. Sometimes the models stack: incentives, social proof, commitment and scarcity all shove the same direction, and their forces multiply into a lollapalooza (the effect you met one course ago, now seen from the driver’s seat). Sometimes they check each other: a vivid, seductive story runs straight into a cold base rate; the stated reason collides with the actual incentive; the appealing first-order move gets vetoed by its second-order consequence — and the collision saves you. And sometimes they complete each other: supply and demand explains the price, game theory explains the players, feedback loops explain the dynamics — one situation, several non-competing lenses, a fuller picture than any single model could draw.

Across five teaching lessons you’ll drive an interactive decision desk — hand it a real scenario (a bank run, a viral bet), switch on model-lenses one at a time, and watch them either pin the needle hard to one pole (a stacked reading) or pull against each other (a flagged tension to investigate). You’ll take a full case apart with several models firing at once, and you’ll build the two things that separate a latticework from a party trick: the meta-skill of choosing which models a novel situation actually needs (and knowing, via your circle of competence, when you are out of your depth), and a written checklist / pre-mortem you run over a decision precisely because your in-the-moment judgment can be captured. Then, because a model you can’t criticise owns you, the honest limits: forcing a model where it doesn’t fit, double-counting the same effect wearing two names, confirmation-shopping for the model that flatters the answer you already wanted, and paralysis-by-lattice — the truth that past the point of decision, more models is not better. By the end you won’t just know the models. You’ll know how to make them think together.