Frameworks:

What is a “Framework”?

A framework is a disciplined way of turning confusion into something you can test. People use “framework” to mean everything from a political vibe to a personal identity to a favorite set of metaphors. That’s not what it means here. Here, a framework is an operational system for thinking: a structured method that tells you what counts as an explanation, what counts as evidence, how you update when you’re wrong, and what would force you to abandon or revise the view.

A framework is not a belief. It’s not a slogan. It’s not a worldview you “hold.” It’s a toolchain you run. If it can’t be run, checked, and improved, it’s not doing framework work.

At minimum, a framework has three parts.

First, it has a constraint vocabulary. It forces every claim to cash out in terms that can be evaluated: mechanisms, boundary conditions, costs, tradeoffs, and scale. This prevents the mind from “explaining” things with words that don’t connect to anything that could, even in principle, be measured or challenged. A framework makes you specify what is doing what, where, when, and under what limits.

Second, it has an inference discipline. It specifies what evidence counts, how strong that evidence is, and how competing explanations are compared. This includes rules for avoiding predictable reasoning failures: cherry-picking, post-hoc rationalization, moving goalposts, and “it’s true because it feels profound.” A framework earns its keep by making it harder to lie to yourself.

Third, it has an intervention hook. It doesn’t just describe the world in hindsight. It tells you what you could do differently and what should change if the model is right. Without this, you can’t distinguish between two stories that both sound plausible. A framework must create discriminating predictions or at least discriminating tests: if you push here, that should move there. If nothing changes no matter what you do, your “framework” is just commentary.

This also implies a hard line: a framework must have failure conditions. If there is no imaginable observation, experiment, or outcome that would make you say “I was wrong,” then what you have is not a framework. It’s an immunization strategy. Frameworks are allowed to be incomplete, approximate, and provisional. They are not allowed to be unfalsifiable while pretending to be knowledge.

Finally, frameworks are judged by performance, not by elegance. A good framework reduces error over time, compresses complexity without losing causal structure, and improves decision-making under constraints. It makes better predictions, supports better interventions, and fails in ways that teach you something. A bad framework protects itself from contact with reality. It explains everything and therefore explains nothing.

So the clean definition is this: a framework is a repeatable method for generating explanations that can be tested, revised, and used to guide action, with explicit rules for what counts as success and what counts as failure.