What I find off-putting about StackOverflow is their crowd-sourced moderation system (which I got access to at some point due to karma threshold). It’s based on pre-categorised moderation queues with flagged submissions that you can (have to) go through, and it’s designed in a way that you mostly choose from a set of pre-fabricated standard responses to “resolve” the issue. The eventual moderation decision is “democratic”, so when 2 or 3 moderators happen to come to the same result independently, then that’s what it is.
To me, that “moderation process” mostly felt like a robotic assembly line which isn’t designed to genuinely help people, but it’s primarily optimised for throughput. The frustration expressed by end-users is a direct consequence of this design that’s not really surprising.
What’s most bizarre is that the system intersperses phony test submissions every once in a while, just to verify whether you as moderator are “doing it right” (or, are still alive, for that matter).