The Political Compass

Every compass quiz on the internet tells you where you sit. None of them tell you whether you meant it. This one does both — and the second answer is the one worth having.

Seventy-two statements, two axes. The horizontal runs from collectivist to laissez-faire; the vertical from libertarian to authoritarian. Thirty-six statements feed each, split evenly so that agreeing with everything lands you exactly nowhere. That balance is deliberate: most compass tests quietly reward a yes-man with a libertarian score, and a yes-man is not a free thinker.

What a compass asks about is what its axes end up meaning, so the blank fields matter as much as the statements. This one used to leave four of them blank — school choice, race-conscious remedies, the franchise, and religion in public life — on the grounds that we could not word them without the wording deciding the answer. That was a thumb on the scale wearing a modest hat, and it is gone: all four are asked now, from both sides, each naming what its own position costs. The effect was measurable and it ran one way. The statements we had been willing to ask were ones where the average American leans left of centre, and the four we had not were ones where he does not — so the old bank read him as more collectivist than he is. Asking the full set moves him almost a full unit, and leaves him roughly where the polling actually puts him.

Underneath the placement runs a second instrument. The statements are paired — twenty-four couplings, each resting on one principle that both halves put to the test in different company. Consent in the bedroom against consent in the wage negotiation. Letting a town's only factory fail against letting a bank fail. Deference to an institution against the leak of its secrets. A tax collected at the desk against a tax collected at the border. Answer both halves the same way and you have a principle. Answer them differently and you have a team. Your consistency grade is the share of those pairs you held, and where you broke one, we name it.

The principles are not all of one school, and this is where the first version of this page was wrong. Twelve of the twenty-four test a libertarian premise, five test a collectivist one, four a conservative one, and three cut across the lot. A grader built only from the first kind does not measure coherence — it measures how libertarian you are, and hands a bad mark to anyone who never claimed the premise. A socialist who wants medicine public and poverty relief private is being inconsistent by his own lights, and so is a conservative who defers to every institution except the one holding the documents. Those are the pairs that were missing.

They are no longer missing, but they are not equal either, and the honest thing is to say so on the page rather than bury it in an average. Twelve against five against four is still half the grader resting on one school, and the reason is structural: a pair is only a fair test if both ways of failing it are incoherent, and a universal rule — the state's reach is the same in every room — satisfies that in any room you like. Premises that distinguish between rooms on purpose, as collectivism and conservatism both do, mostly cannot. Manufacturing the missing couplings to even up the count would mean shipping pairs whose other failure mode is a position real people hold for real reasons, which is a trap and not a measurement. Nor would renaming them help: the same coupling of a factory allowed to fail and a bank rescued is failure is information to one school and a public stake comes before a public rescue to another, and the grade is identical either way — the label is prose, the pairing is the instrument. So your result tells you which yardstick produced your letter, premise by premise. On the fixtures we test against, every respondent breaks a principle he never claimed about three times as often as he breaks his own; the libertarian meets his own premise on half the pairs he is asked and the conservative on a sixth. That gap is in your report instead of inside your grade.

Your answers are not stored. The seventy-two are scored in a single round trip and thrown away; the result that comes back is the only copy that ever exists, and nothing kept on this end can reconstruct how you answered any statement. What is kept is the output — your two coordinates, the quadrant they name, the consistency percentage and the figure you landed nearest — under the same daily-rotating visitor hash this site counts page views with, one placement per visitor per day. That is what lets us say where the average reader sits, and it is the whole of it. No account, no cookie, no leaderboard.

And you will not be shown the running average before you finish. A visible target is how a compass test quietly turns into a popularity poll — you would be answering toward the crowd instead of toward yourself, and the instrument would be measuring its own reflection. The tally is for the people who run this site to read on the telemetry page; the result is for you.

One more thing, since it always comes up: the historical figures you are measured against are placed by their governing record where they have one and their written programme where they do not. Nobody is moved to make a quadrant look better. Hitler sits where the evidence puts him, which is not where either side's talking points would prefer. Every placement carries its reasoning, and you can read all of them in your result.

Scored on the server. Your answers are not written down — only the placement they produce.