Forecast trace · verifiable lifecycle of one forecast
The question is stored in Russian (the original language); no translation is available yet — the original is shown.
forecast id a5f78e401c3b4fdab1aeda0401f44c60 · question id a5f78e401c3b4fdab1aeda0401f44c60:q
No outcome recorded yet: the question is open until the deadline and nothing is inferred before the resolver writes it.
The outcome says what happened. It does not say whether the forecast was good: forecast quality is a separate question, answered below only by the persisted score.
No score: the forecast has no final yes/no outcome.
Quality is read from the persisted proper score only, never from the outcome label: the same «did not occur» outcome yields different scores for different frozen probabilities.
An individual proper score of the frozen probability against the recorded outcome. Skill is defined only as a cohort aggregate against a baseline (Brier Skill Score), never per row — this number is not a skill claim.
Entries are ordered by the time each record appeared. Later checks, resolutions and revisions carry their own timestamps and never alter the earlier snapshot they follow.
Question opened
a5f78e401c3b4fdab1aeda0401f44c60:qProbability frozen: 10%
a5f78e401c3b4fdab1aeda0401f44c60:s0Outcome not yet resolved
Our own system's self-assessment — explicitly marked as such, not an external source.
What would refute this: our own LLM importance assessment does not reach 4/5 before the deadline