SHARP EYES
Predicted right beats voted often.
Voters predict what approval rating the winning feature will earn. Predictions lock at vote close; scoring runs against the always-mean baseline. Bayesian shrinkage. Rolling 12-cycle window.
HOW SCORING WORKSSkill = |mean − actual| − |your prediction − actual|. Positive means you beat the always-mean strategy.
SHRINKAGENew voters start near zero. Score = sum(skill) / (count + 3) — weight grows as predictions accumulate.
WINDOWRolling 12-cycle window. Old form fades. Streaks cool over time.
NO RANKINGS YET
Sharp Eyes scores compute once a cycle ships and accumulates ratings. Until the first cycle ships with locked predictions, the table stays empty. Lock a prediction →