Basing any action on 30-day karma seems unfair in this case. There was a relatively short window during which the comments were being downvoted, and now they're banned; typically for a controversial topic, those would have been reversed by upvotes over the next week. I don't know if that would have happened in this case, but now we'll never know.
Systematically downvoted comments indicate some sort of failure mode, even when it's not the failure mode of being wrong, or when it's a failure mode on the part of downvoters. It's usually possible to reframe the debate in a more constructive manner, long before you get remotely close to start getting banned. I'm not aware of any cases where persisting in highly downvoted behavior on LW achieved anything, apart from some unpleasantness (more unpleasantness the more the behavior persists).
In light of recent (and potential) events, I wanted to start a discussion here about a certain method of handling conflicts on this site's discussion threads, and hopefully form a consensus on when to use the measure described in the title. Even if the discussion has no impact on site policy ("executive veto"), I hope administrators will at least clarify when such a measure will be used, and for what reason.
I also don't want to taint or "anchor" the discussion by offering hypothetical situations or arguments for one position or another. Rather, I simply want to ask: Under what conditions should a specific poster, "Alice" be prohibited from replying directly to the arguments in a post/comment made by another poster, "Bob"? (Note: this is referring specifically to replies to ideas and arguments Bob has advanced, not general comments about Bob the person, which should probably go under much closer scrutiny because of the risk of incivility.)
Please offer your ideas and thoughts here on when this measure should be used.