Tools
Each of these takes one move from the framework and makes it something you can do — on a question you actually care about — instead of something to read about. They're built in public and added as they're ready. Three so far: one you can use now, two on the way.
Interactive tool
Take a claim — an argument, or one of your own beliefs — and watch it come apart into the parts evidence can settle, the parts that come down to what you value, and the parts that are bets either way. It names what's really at stake, lays out both sides, and traces them down to the single hard question underneath.
Open the tool →Interactive tool
Takes any claim and places it on the constrainability gradient — from tightly constrained, where careful people converge, to loosely constrained, where thoughtful people genuinely diverge. It tells you what confidence the claim actually warrants, and names what would settle it, if anything — in short, what kind of question is this?
Interactive tool
Takes any LLM response and flags the model's overconfidence, surfaces the assumptions hidden in the answer, catches where fact and value have been quietly collapsed into one confident voice, and returns a recalibrated rewrite — one that marks what's settled, what's contested, and what's a bet. It builds on the essay Knowledge Without Process and the navigation-versus-computation distinction — the most ambitious of the three, and the furthest out.
Each tool is one move from a larger whole. If you want the move it comes from: the idea in full, and the project behind it.