@KuJoe It's not really suited for automation because in practice there is always going to some amount of nuance, of course; in a case like brigading, it would of course not be the right choice to trust reports automatically.

But I still find it a very helpful litmus test, because of just how many people would functionally end up at "the offender, unless proven otherwise", and how easy those are to catch this way (assuming a well-intentioned moderator).

Even if "the person harmed" is not *always* the right choice, it should probably at least feature as your default choice in cases of uncertainty, and that's the opposite of the typical situation.

@joepie91 I would agree with you, but my opinion would change after having moderated a Mastodon instance for over a year. I would say 90% of the reports are people trying to win an argument they started or trying to silence people who they disagree with. I guess it depends on who is considered the victim.

@KuJoe The report-based nature of something like Mastodon IMO obscures a really important thing for this litmus test to work: you generally don't have a full view of the situation as it develops.

In almost every moderation conflict, all the parties will be showing behaviour that would *superficially* violate a typical code of conduct; but crucially only one party typically started the conflict, and the others are then defending themselves. This is context that can be very difficult to infer in a report-based mechanism.

So when I'm talking about "moderation conflict", I am not referring to a single report (which may be by either the offender *or* the victim!), but rather the situation that led to the report in the first place, where it's often much clearer who the offender is, and that it isn't a case of silencing.

@retr0id I find that there usually is, but identifying them requires learning to recognize patterns of abusive behaviour (which a lot of moderators are actively unwilling to do)