0.11.0 — a poll under the entry

The idea arrived as “three provocative questions about the author’s decisions, so that people want to click and see how others judge him”. Mechanically simple, directionally risky: under an entry about the worst evening of someone’s life, a counter saying “sixty-seven per cent think he was wrong” is a verdict, not engagement.

The direction was changed before the first line was written: **not “judge him”, but “recognise yourself”**. Instead of a question about the rightness of someone else’s act — a question about your own experience. The pull to click stays the same, because curiosity about how many people are like you is stronger than curiosity about somebody else’s sentence. What changes is what the result is: information for the person asked.

The data follows the same principle as comments and ratings. The question lives with the entry and is in the instance’s language — each version has its own wording. The vote lives in the shared database and is bound by the question’s identifier, the same on both sides. **A question has two wordings; a vote has one place.**

The result appears only after you answer. That is not hiding data, it is the whole mechanism: the reader is meant to answer for themselves first and only then see how many people are similar. The other order would turn a poll into a survey to read.

We store nothing that would let one person’s answers be assembled — the question identifier, the answer and the version’s language. That someone has already answered is remembered by their browser. With questions like “have you ever failed to notice that someone was asking for help”, that is not a detail.