HN has a rather brutal philosophy in flagging/downvoting items lacking substance and shadowbanning troll-ish folk; it seems to work ok and is usually not too sinister (and when it is, human intervention/moderation can often help). By contrast, large communities like states and national institutions can - with just a hint and a nod - inflict surprisingly chilling effects throughout communities of millions.
So where do we draw the line? I dunno. Whereever we draw it, it's got to be parametric with enough inputs to take into account at least the rough subtleties of the situation. Because the comic is right: I don't need to put up with an asshole's bullshit, but I also don't want to destroy them. Just, you know, tone it down a smidge and stop being a jerk.
I'm fairly passive. Normally I do that by ignoring them, changing the channel, or just avoiding the website. Rarely I'll actively flag or downvote (ugh, so hamfisted). And sometimes, I just go away if they take root. With high-signal web analytics, tracking, stats, and other social-buzz things, my avoidance may actually be logged and noted. Maybe I help starve out what I consider noise, thus boosting what I consider signal.
What if a lot of people in a community are just outraged? Or are very sensitive to easily become outraged? I think that's where things get underdamped fast, and where your worries come into play. My only thought is to starve it of energy 'till it settles down, but we've seen enough self-sustaining bullshit storms that I bet passive resistance doesn't work much anymore for that.
EDIT: speling