有很多现成的反作弊策略可以限制机器人，比如长期来看线性增长的数据量…
GitHub上应该有不少开源策略。
Discussion
找了一会没找着，这在目前大概是个难题。
Any thoughts sharable on handling bots spamly DMing self issue ser? #[4]
Hmm that’s a weird one 😂 I think lots of people are experimenting right now, including spammers.
I think long term the only way to address spam is through modeling and retraining as new data comes in. Rules based spam filters will be defeated.
this kind of spam is a little... tbh, socialpath😂, no point in wasting others' BW if not their own relay