How this site is made

Who

One person: Alex Seot, who runs the site, decides what goes on it and is answerable for everything published here. There is no editorial team, no review board, and no contributors — claiming any of those would be the first thing on this page that was not true.

How the writing is done

The pages and articles are drafted with AI assistance and edited, checked and approved by a person before they go live. That is worth saying plainly rather than leaving you to guess: a hundred per-question pages and a hundred and one per-score pages are not hand-typed by one developer, and pretending otherwise would be silly.

What the assistance does not do is decide what is true. Every factual claim has to survive a check against something outside the draft, and where no source exists, the page is required to say so instead of asserting a date or a number that reads well. The history article is the clearest example: it grades its own claims as sourced or unsourced in a table, and the widely repeated 1924 origin is marked as folklore, because no primary document has surfaced.

How the questions were settled

There is no official edition of this test to copy from. The list here was reconciled from four live copies published by other sites, compared item by item. The six items those copies disagree about are published with the wording chosen and the reason for each choice — including item 69, which one site prints in full and another censors to a bare question mark.

Where the numbers come from

Nothing on this site is a survey figure, an industry estimate or a number borrowed from another page. Every statistic is computed at request time from scores submitted here. A submission stores the score, which of the seven question sets produced it, the moment it was written, and an age band only when the visitor chose to state one — no identifier of any kind, which is set out in the privacy policy.

Figures are withheld until a test holds 30 results, and the distribution chart waits for six hundred results in total — thirty for each of its twenty buckets — because an average off five submissions moves every time a sixth arrives. When there is nothing worth publishing, the page says so rather than showing a number that will be wrong tomorrow.

What happens when something here is wrong

It gets corrected in place, and the correction stays visible. The article about what purity test sites do with your answers carries one: it claimed this site loads no third-party script, and a measurement showed the CDN was injecting its own. The claim was wrong, the paragraph now says so, and the script is refused by a content security policy rather than quietly left in place.

If you find something here that is wrong, write — a correction is more useful to this site than being right was.