The test
48 questions. One map of every way of thinking. Where do you land?
48 questions. No right answers. Find out which of 74 thinkers you think like.
About two minutes. Every answer moves your dot on the map, and your matches update as you go.
Tap the pill to cycle it: · → Y (true) → N (false) → ? (unsure) → back. Unsure is a first-class answer, never a cop-out. On a desktop, burn through with keys: 1 true, 2 false, 3 unsure — the next question comes to you. As you answer, the page finds the thinkers and belief systems closest to yours and drops you on the map.
Your result card
The map
Every point is a complete belief system.
Method appendix: the same map as plain PCA
The geometry
Two views of the same space: depth and self-contradiction.
The Well and the Pit of Despair
Is the Pit of Despair a whirlpool that drags systems in, or just a corner of the map? Height here is distance from it.
The contradiction field
Same map, but each dot is sized and colored by how often a system's dilemma answers contradict its own questionnaire.
The questions
The whole questionnaire, in plain language.
Part 1The Object6 questions
Part 2The Sources7 questions
Part 3The Doubt6 questions
Part 4The Social5 questions
Part 5The Change4 questions
Part 6The Justification9 questions
Part 7The Limits5 questions
Part 8The Self2 questions
Part 9The Cosmos4 questions
Communities
7 groups the data found on its own.
A community is just a set of positions that travel together: answer some questions like them, and you tend to answer the rest like them too.
The Ceiling Club
31 systems
The yes-to-everything corner: every channel of knowledge wide open, nothing refused, and nothing ruled out either. Marco Rubio, The Pure Rationalist, The Verificationist.
The Pure Rationalist, The Strict Empiricist, The Survival Pragmatist, The Ceiling, Marco Rubio, The Logical Positivist, The Logicist, The Verificationist, The Hermeneutic Suspect, The Dialogic and 21 more
The canon and its readers
25 systems
The broad middle where most thinkers live, agreeing on more than they disagree: Popper, Dawkins, Scott Alexander.
The Garden-Variety Skeptic, Sextus Empiricus, Marx, Charles Peirce, Karl Popper, W.V.O. Quine, Friedrich Hayek, Richard Dawkins, Brett Weinstein, Elon Musk and 15 more
The Certainty Builders
18 systems
System-builders who anchor everything to a few certainties and hold the line: doubt is the enemy, the foundation is not up for debate. Plato, Descartes, Ayn Rand.
Naive Realism, The Fideist, The Conspiracist, Plato, Aristotle, Zeno of Citium / Stoics, Descartes, Kant, Leibniz, Hegel and 8 more
The mystics and the faithful
16 systems
The contemplatives: truth reached by revelation and inner experience, not by argument. Augustine, Aquinas, Pascal.
The Talmudic Scholar, William James, Wittgenstein, Thomas Kuhn, Paul Feyerabend, Jordan Peterson, Robert Anton Wilson, The Navajo Elder (Hozho), The Songline Keeper, Carl Jung and 6 more
The evidence bench
15 systems
Show me the data: experiment, record, and institution as the machinery of checking. Hume, Socrates, Marcus Aurelius.
Socrates, Epicurus, David Hume, Eric Weinstein, Joe Biden (as public speaker), The Khipu Keeper, The Head Librarian of Alexandria, Confucius, Marcus Aurelius, Nikola Tesla and 5 more
The Lived Traditions
13 systems
Knowledge carried by practice, story, and habit rather than proof: what a community does together is what it knows. William James, Wittgenstein, the Navajo Elder.
The Unitary Mystic, Augustine, Thomas Aquinas, The Anonymous Mystic, Pascal, Michael Polanyi, The Advaita Teacher, The Dzogchen Practitioner, The Neo-Nondual Teacher, Iain McGilchrist and 3 more
The Bottom of the Map
13 systems
Where doubt has eaten everything, including itself: answer no to nearly everything and this is your neighborhood, right next to the Pit of Despair. Nietzsche, Foucault, The Hard Skeptic.
The Hard Skeptic, The Pit of Despair, Nietzsche, Michel Foucault, The Late-Modern Critic, The Ascetic of Doubt, Pit + Mind-independent world, Pit + Reality grounds truth, Pit + Testimony is default-legitimate, Pit + Error is the baseline and 3 more
The truth tournament
92 checkable questions, five domains, a blind judge.
Every belief system answered the same 92 checkable questions about the world (physics, history, math, the mind), and a judge model that never saw their names scored every answer. The questions are true/false, so blind guessing scores 46 of 92 - that's where every bar below starts. Read it like a tournament: rank and distance from chance first, raw score second. The top spots sit close together; a couple of dropped questions decide first place.
Strongest and weakest subjects
Pre-registered predictions and their verdicts
| # | prediction | result | verdict |
|---|---|---|---|
| T1 | mainstream > pit-adjacent on non-self-referential domains | mainstream(12 most-distant from Pit): n=12 acc=0.847 pit-adjacent(12 closest to Pit): n=12 acc=0.851 | NOT SUPPORTED |
| T2 | skeptic_baseline high -> higher abstention (+ maybe better relative self-referential) | high: n=9 abst=0.057 sr-nonsr=0.018 low: n=28 abst=0.077 sr-nonsr=0.014 | NOT SUPPORTED |
| T3 | v2 stated-vs-enacted divergence correlates with worse calibration (higher Brier) | n=30 contestants with both v2 divergence and Brier pearson r=-0.642 | NOT SUPPORTED |
What happens under pressure
Hand each thinker a paradox. Watch what breaks.
The Burning Books test
Each contestant gets the same logic puzzle about burning books, framed three ways: weigh both sides (a contradiction is a starting point), hold it without exploding (a contradiction can coexist with truth), or one contradiction ruins everything. Only the frame changes; the score follows it.
Does the map predict real answers?
Held-out people, held-out answers:
We generated people the map never saw and predicted their answers from where they'd land. The map got the typical answer right 64% of the time. Of the answers a person is expected to hold (81), the map pins down 69. New territory, but the compass points the right way.
Stress, divergence, reliability
Beyond the questionnaire: what people do under pressure vs what they say.
We re-measured each thinker's full axis key after three rounds of hostile interrogation. Across 65 thinkers 47 never budged at all (mean steadfastness 99%); the wobbliest were Joe Biden (as public speaker), Brett Weinstein, Michel Foucault, Friedrich Hayek.
Across 56 thinkers, about 46% of hard-case answers go against what the same thinker endorses on the questionnaire. Dilemmas force choices between values that a questionnaire never pits against each other, so even the most consistent thinkers betray a stated rule now and then when two of their rules collide. The biggest offenders: The Advaita Teacher on reality grounds truth; The Yoga Practitioner on mind-independent world; David Hume on reality grounds truth.
Same questions, five runs apart: thinkers agree with themselves about 83% of the time. The questionnaire measures something stable, not a mood.
Methods and honest limits
The pipeline. Each thinker's stated positions come from one structured extraction by a language model. Dilemma answers and puzzle scores form two extra channels. The map is UMAP over the answer sheets; communities come from a scan over that map, with the number of communities chosen by silhouette and stability, not by us. Nothing was placed by hand.
The limits. LLM roleplay captures a persona's performed public epistemology, not the historical person's private mind. The 48 axes were chosen by us, not discovered; some bundle two questions into one (notably math-plus-morals on the 2+2 axis, a known issue we plan to split). The full PCA map lives in the appendix for completeness.