How we build the numbers

Estimates combine a literature prior with weighted reports from Reddit counts, investigator reviews, and Open Test completers. Every estimate shows an 80% range. Below threshold, we show an empty state instead of a number.

Votes are separate from evidence. Anonymous votes weigh 1, signed-up votes weigh 5, and a complete review of that treatment weighs 10. Crowd forecasts exclude people enrolled in an Open Test of that treatment.

Personal run verdicts compare baseline fog to the last week on-protocol, then look for a slide-back when you stop. Low adherence or missing days make the verdict unclear on purpose.

Limits: people who post and vote are not a random sample. PubMed coverage is incomplete. We will keep listing known biases here as we learn them.