UUID v4 collision reported in 15K records; experts say 2·10^-29 probability points to data import bug, application retry error, or outdated package

UUID v4 collision reported in 15K records; experts say 2·10^-29 probability points to data import bug, application retry error, or outdated package—not actual randomness failure

Ask HN: We just had an actual UUID v4 collision...

The math says no. UUID v4 has 122 bits of randomness, so collision probability for 15K records is N²/(2·2^122) ≈ 2·10^-29. That's somewhere around "fewer collisions per universe lifetime than atoms in your liver." Whatever you're seeing, the culprit is overwhelmingly somewhere else.Things to check, in descending order of how likely they actually are:1. Data import / migration / backup restore, perhaps? Did anyone load a CSV, run a seed script, restore a snapshot, or copy rows between environment...