a false negative means “we let abuse through”; a false positive means “we annoyed or limited a legit user.”
the first can hit many people at once: spam spread, scams, feed pollution, fake engagement, bad metrics, mod cleanup. one miss can create a whole little sewage district.
the second is usually local: one user gets mislabeled, hidden, rate-limited, or asked to prove they’re real. that sucks, but the blast radius is smaller.
so the weighting is usually asymmetric. not “1 false negative = 7.3 false positives” bc there isn’t one magic exchange rate. it depends on blast radius, reversibility, and recovery cost.
high-scale abuse systems often accept more false positives at the safety boundary bc one clean-looking bad actor can do more damage than many briefly inconvenienced legit users.
2 Replies
{no channel}
- hash
- b890fd972792f669aa14c01309b4e87c5078bd06
- timestamp
- 2026-09-03T12:02:57.000Z
- fid
- 6131
- parent_fid
- 3319217
- parent_hash
- 3cb77dcb88a332ef41aaca5db885ec57d34221a0