The troll block here is "hellbanning." http://toway1234.tumblr.com/post/16021243707/hacker-news-hel...
The problem is that since hellbanning is inherently passive-aggressive -- nobody in the community knows it's been applied, not even the person who's been banned -- it doesn't teach the community lessons on what behavior is beyond the pale. The banned person simply disappears. This makes life easier for the admins (since it lets them avoid both having to confront the person being banned and a time-consuming discussion of why that person was banned), but at the cost of not putting forward an example of How Not To Be for others to learn from and follow. So people make the same mistakes over and over again.
If people really do need to be thrown out of the community, it'd be healthier to have at least some of that happen via public executions rather than silent disappearances. At least when you see someone up on a gibbet, you get the message not to be like that person if you don't want to end up there yourself.
Full disclosure: I was partially hellbanned myself (submissions only, not comments) for a period, despite having 4,000+ karma at the time. I still don't understand what I did wrong, really. The experience knocked my opinion of HN down a few notches.
Really, one-off accounts seem to be the HN version of anonymous posting. Which suggests to me that it may be more beneficial, perhaps, to simply allow anonymous posting with some caveats (any user can choose to block all anonymous posts, anonymous posters couldn't be upvoted or downvoted, it automatically triggers some kind of alert to the mods to watch the post, and maybe it even costs you karma to use the feature.)
I also keep considering that some kind of API which allows users to choose to block other users might be an effective form of self-policing, but that would also kind of by definition stifle and fragment conversation.