How exactly? It's not like the only other option is "always assume bad faith".
Since reality is a bit more complex, I think such things should be decided on a case by case basis. I've seen plenty of both extremes here on HN, and tons of unknowns
By being an effective moderation tool. I like shadowbanning more than a simple ban, because I think it works better by not giving the bad actor an instant response. They can't use the feedback for further bad action, creating a sense of confusion, or being ignored. And then if we suppose that it works well, then with other useful tools in the moderators' bag of tricks, they can do a good job keeping a community's discussion above a certain level, thereby benefiting the community at a larger scale. I'm sure that if used in bad faith, it can be used to control a narrative, but in this regard it doesn't do a better job than other moderation tools, like editing a comment or banning users.
