A properly designed Twitter wouldn't have anyone sitting at HQ trying to figure out which accounts to ban and which ones to leave alone
Conversation
Replying to
there we agree. PeopleRank or suchlike with sock puppet detection via machine learning for whack-a-mole trolls
1
7
Replying to
I'm skeptical of 99.99% of 'free speech' derping. 0.001% is courageous speech under threat. Rest is "waah, she bit back" crybabies
1
1
Replying to
there is always subjectivity; only diff is whether it's algo designer thinking of use cases or operations person thinking live cases
1
1
Replying to
then either your systems haven't been stressed past their abstraction leak thresholds or you're not acknowledging subjective elements
1

