@Eulb think about it this way. if 0celo7 denies that he has been given any reason for his suspension, theres no way to disprove that because SE policy is not to make all that public, for good reason, to not (further) humiliate errant users in front of the public. historically he has a tendency to do this, it can be seen in the transcript... aka denial. my feeling is again, a 1yr suspension for chat flags seems extreme but the mods have a point, 0celo7 overly denies personal responsibility...

I mean he has the liberty to do that. It's in bad-faith (on part of the moderators) to assume that 0celo7 is denying responsibility in bad-faith.

I'm not saying that counter-bad-faith logic is a good game to play. It's unfortunately wrought by the moderators themselves, by the death-of-the-author game that @ACuriousMind and other moderators likes to play.

So it's a "legal" game nonetheless.

looking back, 0celo7 had some responses to JRs answer to his question in meta that have been deleted about mods out to get him & causing all his problems. do not support that deletion, think it would be revealing to the public in support of mod action(s) in the case. anyway 0celo7 can take heart at 16 votes on his meta post and see that mods will probably take no action no matter what the public support of his case. undemocratic but thats sometimes an aspect of SE policy in general.

@BalarkaSen Could you remind me where you think ACM confirmed your interpretation?

Also, if you're referencing the mod message in support of that interpretation, could you be specific?

@DavidZ Here, and subsequent messages.

Thanks.

I can be specific, sure. The mod message, as I see it, was a detailed summary of 0celo7's auto-suspensions and mod suspensions. But it was never made clear what it was being used as a response to.

Was it, "We have noted the size of your suspension log, and therefore suddenly think that you aren't worth the trouble because are still garnering flags?", or was it "We have noted these are the truly troublesome and unwanted messages that were genuine flags and genuine mod bans and we have arbitrarily decided that we don't want them anymore".

What was it a response to? My gut feeling is that you have to have that response when some offense was caused which adds to the list of offenses the mod messages listed.

I think 0celo7 has sent a response to that effect (on my suggestion) to a section of the message that looked vague.

Given the vagueness of how the conclusion (that the authority doesn't trust the user to participate in the chat in a consistent way with the guidelines) was arrived, I am forced to conclude that the size of the suspension history is garnering a priority over a skeptic analysis of the user.

One thing at a time. First, on the matter of 0celo7's moderator message, that message constitutes the moderators' entire official statement on the suspension, and accordingly I'm going to be very careful about any response I might offer to questions about that message. So don't expect any quick answers (if any at all) to such questions.

I understand.

If you ask a general question about rules or policies, one which is entirely separate from 0celo7's suspension (though possibly inspired by it), that's something we could discuss more freely.

Now, on the separate issue of the messages ACM posted that you linked: two things stand out to me. First of all, those were talking about your own history on the site, so they're not necessarily something you should take as a general policy, and in particular not something you could necessarily "translate" over to apply to anyone else's history.

We could definitely discuss just what out of those messages does generalize to everyone.

The other thing is that there are both manual suspensions and automatic suspensions, and I'm not sure we're being totally clear about keeping them straight.

@DavidZ Gotcha.

When I said this:

I was talking about flags specifically, which somewhat correlate with automatic suspensions, whereas ACM may have been talking about only manual suspensions.

@BalarkaSen Cool, glad this seems to be helping. There's a lot of case-by-case decision making involved in moderation, and we have very drawn-out internal discussions on how to deal with that and balance the need to be fair (among different users) with the need to take actions that are appropriate responses to the problems we see, given that every problem is unique in some way.

It's definitely not a solved problem, but we try.

(well, maybe not every problem is unique, but most of them are)

My qualm was with this: "But let's say it like this: If a user appears in the top ten of most flagged users across the chat network, they should probably tread lightly". This seems to imply that you can rank-order the users based on their size of the suspension history (I don't believe ACM restricted to manual suspensions when he said that) and the higher the users are in that list, the more biased moderation they are going to encounter.

I believe that's a statistically flawed idea.

OK, yeah I see what you mean there.

@DavidZ Yeah, I know intricate decisions are made with respect to these issues. I just fear that SE's poor policies (eg, the one about aggressiveness - which is relevant given the situation it was a response to, but a poor generalization without any specific focus to the essential questions "how aggressive?" and "aggressive when?") are inflitrating these decision making choices.

Here's one statistical justification: imagine that each chat member has some hidden "score" that could be used to determine whether their overall behavior is appropriate for chat or not, i.e. people with score over some threshold should be ineligible to participate in chat. (This is not how we model things internally, I'm just making an analogy to explain what ACM said.)

Then the number of flags (or perhaps more precisely, frequency with which someone gets flagged) is like an experiment which yields a confidence interval for that score.

As the number of flags gets higher, the confidence interval narrows, and this may lead to a decrease in the p-value for the person's score being below the threshold.

However, flags aren't perfect. They're just an indicator, but we do have more and better information available, i.e. what actually got flagged, the context for those messages, our own interactions with the person, etc. All of that gets considered in moderation decisions.

No matter how many flags someone gets, if other factors don't support the conclusion that they're behaving inappropriately, they're not going to get a suspension for it. (At least, that's not the way of the Physics mods. I can't really speak for the rest of the network.)

I have to leave to run an errand, I'll be back after a little while.

(ok back, for now)

@DavidZ I get that it's supposed to be a rough guess for the authenticity of the user. But I doubt that's determined by a single variable with which it has some nonzero correlation ("number of flags"). I am thinking, eg, of the various statistical procedures to write a given variable as a linear combination of other independent variables (upto a stochastic error term).

You're stating that the Physics mods do an implicit intuitive factor analysis to take account to these multivariable criteria. But it isn't at all clear to me that the number of flags aren't taking a greater contribution than it should.

It's not an individual issue, exactly. Think of it this way: moderator X is a good mod, and always checks the context, authenticity, background and various interpretations of a user's messages. So in that sense, moderator X's judgement are explicitly "fair".

But implicitly, maybe he's cataloging through the users and their suspension histories to initially mark some of them as "suspects", and then filter it out by the factor analysis.

That procedure inherently produces a biased contribution of the flag variable.

I think that's the general overview @ACuriousMind was giving. That's going to happen, and that kind of judgement procedure is endorsed by the SE's aggressiveness policy, if one's eclectic and takes "aggressive" to mean "give higher order of punishments to immediate suspects instead of appropriately judged punishments to appropriately filtered problematic users"

I mean, that's going to minimize the time variable, but that's not a good measure of efficiency :P

idea/ "negotiated compromise": someone posts meta question asking for 0celo7s suspension to be reduced to 2-3 mo. afaik thats double his last one. supporters can upvote and/ or comment or answer. condition: he take responsibility for some )( disruptive tendencies and write a public apology as an answer.

@DavidZ theres a big statistical difference between total # of flags and frequency of flags esp for frequent chatters such as 0celo7. evidence from his chat history is that he is at least partially responsive to mod feedback/ actions.

@BalarkaSen "rough guess for the authenticity of the user" is a strange phrase.

@BalarkaSen I'm not describing what we actually do. I'm making an analogy for how having a large number of flags can be an indicator that someone may not be behaving appropriately for chat. I thought I was fairly clear about that.

@vzn Absolutely not. I mean, maybe someone makes that meta post, maybe not, but either way the moderators are not going to take it into account regarding any decisions about suspensions or other moderator actions.