[nostrademons]

That's how most large sites handle abuse & anti-spam protection. One of the cardinal rules of spam-defense is that you don't let spammers know that you've caught them, you just silently quarantine their spam. Otherwise, they'll adopt countermeasures and you're stuck in a rat race.

[qq66]

If you're going to do that, you have to be 100% sure that a person is a spammer. One bad comment a year ago doesn't make someone a spammer.

[nostrademons]

I think PG's testing the hypothesis of "What if we don't have to be 100% sure? What if we probabilistically ban people based on comments that are likely to have no value? Will that result in a higher or lower community quality?"

Presumably the assumption is that some good people will be banned and leave, and some other good people will just leave out of sympathy, but it's better to have good people leave than have bad people enter the community. It's an interesting hypothesis, and HN has been going strong for 5+ years now with minimal moderation effort, though I have to admit that it seems to violate basic rules of fairness and empathy. Then again, a lot of businesses are built on being unfair to people.

[qq66]

My argument is that you have some obligation to the person you are hellbanning, that a decent person can't just probabilistically waste a year of a stranger's life as part of an experiment on community quality.