Disclosure: I built this. It is free, no account, no ads.
Most tests of belief sort you by what you conclude. This one asks how you decide what counts as knowing. Fifty-one either/or axes: does testimony deserve default trust, does a single counterexample kill a rule or just dent it, is the world legible at all. You answer the axes in about five minutes, and 137 belief systems, from Socrates to Quine to the Dzogchen practitioners, have already answered the same axes, so your answers land you on the same map as them.
Three things I would rather flag than bury.
The 51 axes are a model I chose, not a structure I discovered. Some bundle two ideas into one question, and each card says when it does.
The historical figures are a language model playing their public epistemic style, not the private minds of the people.
One of my own predictions failed. I expected a collapsed epistemology, a position that answers no to nearly everything including anything can be known, to act like an attractor that systems drift toward under hostile questioning. They drift away from it instead, toward more commitments rather than fewer. That result is on the site as a failed prediction.
The surprise is the last one. Pressure pushed systems outward, not toward a minimal core, which runs against a story I held that hard questioning strips beliefs down. If that holds outside my toy setup, it says something about how disagreements resolve under scrutiny, including among people who share a lot of priors.
I am not sure the map earns its keep. Maybe the axes are arbitrary and the clustering is an artifact of how I wrote the prompts. A useful test would be whether someone else, choosing their own axes, recovers anything like the same neighborhoods. I have not done that, and I would rather hear how to falsify the map than whether you liked it.
https://epistemia.life
