People, books, and breakthrough ideas—connected to the mental models and moves you can use.
3,456 entries
Before acting, write down the numbers that will judge the plan: where you stand today, what result counts as working, what result would prove it wrong, the soonest you can honestly read it, and what ends the experiment.
Say in advance what will count as proof you arrived, who or what judges it, and when you check. It also names the conditions that force you to revise or retire the target instead of quietly sliding the deadline.
Name one observation, or one small safe experiment, whose result would come out differently if your loop is real than if the rival explanation is right. If nothing you can watch or try separates them, call the loop speculative and go gather evidence instead.
Every fact travels through about 11 handoffs between the real world and your decision, and the Evidence Chain is the practice of walking those handoffs backward to find the one that is broken.
Before you gather any data, write down on one page what decision it serves, what the competing explanations predict, and what result would change your mind.
Evidence-based medicine organizes evidence into a hierarchy from weakest to strongest: expert opinion and case reports at the bottom, case-control studies above them, cohort studies higher, randomized controlled trials (…
One written line, set down before you look at the evidence, that names your claim, the strongest competing explanation, the check you will run, and what you will do for each possible result.
No geometry at all. Claims down the side, sources across the top, and in every cell what that source actually gives for that claim. Reading across a row shows how thin or thick the support for one claim is. Reading down a column shows how much weight a single source is carrying across the whole case. A row with one filled cell and a column doing all the work are both visible at a glance, and neither is visible in a bibliography.
Name in advance the observable trace you will count to tell real change from a good mood: something recorded, comparable month to month, and capable of showing no movement at all. If nothing could come back flat, it is not evidence.
This check asks what you do not yet know about an option, and whether keeping it would teach you something. An option built on guesswork survives only if you can restate it as a cheap test that produces an answer instead of a bet.
Biological mutation can generate variation on which selection acts, but evolution also draws on existing variation and other processes
Natural selection occurs when heritable variation is associated with differences in reproductive success under particular conditions
A strategy resistant to invasion by sufficiently rare alternatives under a specified population game.
Conditions change, so this test asks whether you decided in advance when the goal gets reviewed, what evidence would force a rewrite, who holds the authority to change it, and what would trigger stopping or rolling it back.
Ray Dalio’s philosophical view that learning and adaptation can be a central source of accomplishment and satisfaction
Draw the parts at the size of the attention they take rather than the size they are. A habit costing 40 a month becomes the height of the page and a bill of 2,100 a month becomes a postage stamp. The distortion is the claim, so it has to be obvious and it has to be labeled. A reader who mistakes it for a chart has been misled, and that is your doing, not theirs.
Exaptation describes a feature useful in its present role that was not originally shaped by selection for that role
Repurpose a state-transition model for a new domain while checking the adequacy of its state definitions and conditional-independence assumption.
Define ordinary bounds and route everything outside them to a visible human review queue.
A setup that handles the ordinary case perfectly can fail silently on the rare case, and because nothing ever alarms, you keep trusting it.
Munger's term for the universal tendency to overestimate one's own abilities, contributions, and possessions—the Lake Wobegon effect where everyone is above average
Exchange can benefit its participants while leaving consequential effects on others outside the decision’s costs and benefits.
Ask what inputs your frame treats as free, unlimited, or somebody else's problem. Budget, tooling, shared systems, senior attention, headcount. These never appear in the explanation because nobody counted them, yet the work cannot happen without them.
Turn a passive explanation into something you can run, vary, and test, like a checklist, spreadsheet, simulation, or prototype.