Question and Vote Methodology

Question corpus

The source system contains 22 authored dilemma families. Controlled premises, consequences and contexts are combined, then every annual season is exhaustively audited for malformed wording, duplicate A/B pairs, unsafe length, unstable IDs and SEO metadata limits.

A normal year contains 1,051,200 rounds and a leap year contains 1,054,080. Stable IDs are derived from the family and variant rather than the schedule, so a question keeps its URL and lifetime total when a new season is shuffled.

Vote collection

One browser choice is accepted per question per round. Votes are submitted in anonymous batches and stored as aggregate A and B counters, not as individual vote records or user profiles. Edge abuse limits and server-side payload validation reduce obvious automated inflation, but the figures are game results rather than scientific polling.

Publication threshold and freshness

A stable question page stays out of search indexing until it has at least 50 recorded votes by default. Public JSON and HTML state when their totals were observed; short caching and write batching can make the durable total a few seconds behind live play.

How to cite the data

Cite the question, A and B totals or percentages, WYR HQ, and the date observed. Example: “WYR HQ, question 1000123, observed 9 August 2026.” Link to the stable canonical question URL.