How a track record lies without a single false statement

You never have to fabricate anything. Every entry can be real, sourced, correctly dated and accurately graded, and the overall picture can still be false. Selection does all the work.

Pick a public figure. Go looking for their predictions after you already know how the last two years went. Whatever you were hoping to show, the material is there — enough hits to build a genius, enough misses to build a fool. Both compilations are honest line by line.

Why the ordering is the whole thing

A prediction registered before its outcome carries information that one registered afterwards cannot. Not because the words differ — they are the same words — but because of what the act of registering proves.

Register in advance and you did not know. You could not have selected for the result, because there was no result yet. The set is whatever you thought was worth logging at the time, which is exactly the set an honest track record needs.

Register afterwards and every item passed through a filter with the answer already in view. The filter might be generous or hostile. Either way it is a filter, and the resulting percentage describes it rather than the forecaster.

How this archive handles it

Records here are marked pre-outcome or historical, and the two never mix. A historical record — added when the answer was already knowable — is kept, labelled, and permanently excluded from any track record, even when it settled cleanly.

Two of the records on this site are historical and both resolved “not met”. They are real, sourced, and they count for nothing, because a set of after-the-fact submissions is precisely the thing this rule exists to keep out of the numbers. Fans import hits. Critics import misses. Let them; just never blend the two.

This costs us. The pre-outcome archive is the smaller and slower half — nothing locked in advance has settled yet, and the first one lands in days. Historical records would have made the site look established immediately. That is exactly why they are fenced off.

What to check on any track record

One question does most of the work: was each item logged before its outcome was known, and can you verify that independently?

Not “does the site say so” — anyone can write a date. Whether there is something outside the site’s control that fixes when the entry existed. Here that is a hash committed to a public blockchain in a batch, which proves the record existed in that exact form by that block time and can be recomputed by anyone. It proves nothing about whether the claim is true. It proves the entry was not written after the fact, which is the part selection depends on.