26

I just came across, in normal browsing of the site I am a moderator on, another post from a drive-by user which had legitimately been deleted as "spam or rude or abusive".

In a way, that's great, because it means that community flagging and moderation works. No moderator intervention required. Ideally, that should be the end of it.

Unfortunately, my experience is that even just one such post is a pretty good indication that the user is up to no good. Appropriate action to take at that point can range from nothing more than a quick review, all the way up to notifying moderators across the network of the user account in question. But these posts are awfully easy to miss, because by the time a diamond moderator comes around, they may have been pushed off the front page and they may already have been dealt with (so don't show up in the flag summary view either).

Hence a humble request for a new feature: An auto-flag raised on the post whenever a post is deleted as either "spam" or "rude or abusive" purely by community voting/flagging. Since a moderator is already involved if a moderator casts a "spam" or "rude or abusive" flag on the post, there is no need to raise an auto-flag for those cases (if the moderator is for some reason trying to conceal that act from the other moderators, they could just dismiss the autoflag as well, so nothing is gained from the autoflag in that case).

To help reduce the problem with a huge number of flags on high traffic (high spam) sites, this might be selectable by the moderators of each site in communication with the community managers.

This was raised by Monica Cellio in a semi-related answer three years ago, but doesn't seem to have gained much traction (or spurred any discussion) there.

  • 6
    Note that if you were to implement this on sites like Stack Overflow, Super User, Ask Ubuntu or any other site with a heavy spam load, it would likely result in hundreds of extra flags a day, most of which will be useless. – terdon Nov 17 '17 at 18:05
  • @terdon That's a fair enough argument (but also seems a good reason to have a bit of discussion surrounding the idea, including that one). – a CVn Nov 17 '17 at 18:13
  • 4
    This is a subset of this feature request. – Monica Cellio Nov 17 '17 at 18:24
  • It bears pointing out that a huge number of spam accounts are hit-and-run, i.e. they spam once and then never return. As a useful data point, the Metasmoke "blacklisted user" rule has only a 66% hit rate at the moment. Metasmoke maintains an internal blacklist; users whose posts are confirmed by a trusted user as spam are usually automatically added to the blacklist. In the grand scheme of things, 2,660 spam posts by returning spammers is a fairly small minority (the database contains over 80,000 confirmed spam messages). – tripleee Jan 2 '18 at 12:40
  • ... And looking at the recent hits, many of the returning users are regular users who went postal and vandalized multiple posts, not spammers in the strict sense. (Of course you want to catch the vandals, too.) – tripleee Jan 2 '18 at 12:42
16

There's a 'users with flags' list on /admin/users. Such posts need to have 6 or more flags to be deleted, so these kind of users are generally at the top of the list.

For posts reported by SmokeDetector, a list of (potential) spammers per site is maintained in metasmoke; the list is only accessible if you are a ♦ moderator on a site in the network and have a metasmoke account. Hop into Charcoal HQ if you have any questions.

I regularly use these lists to get rid of spammers on Ask Different, for me they're a sufficient substitute for such a feature.

  • If your spam-flagging user base is not active in Charcoal (basically meaning that you get spam which is different from spam on the other sites in the network, and nobody is helping Charcoal detect those posts automatically) then this will not be much use. I hope and believe that there are no such sites in the network, but there is certainly some "dark spam" which does not get detected by Smoke Detector or manually reported in the Charcoal chat room. I'm mainly pointing this out in case there are active anti-spam efforts which are not aligned with Charcoal, now or in the future. – tripleee Jan 2 '18 at 12:29
  • (cont') ... For the obvious purpose of hopefully achieving better coordination. Pop up in the Charcoal HQ chat room on Stack Exchange if you are this future reader. – tripleee Jan 2 '18 at 12:30
  • On some sites it only takes 3 red flags to delete a post, so those don't show up on the mod-visible list. If owners of red-flag-deleted posts always showed up on the list regardless of the number of flags involved, that would help. – Monica Cellio Jan 2 '18 at 15:39
6

I'm going to put this as in review. To be honest, I'd much rather come up with a scenario where users and moderators can nuke spam accounts (and all of their posts across the entire network) from orbit once 2 or 3 trusted users, or a single moderator has made an identification. I hate knowing that we waste time on these sleazy things, even the time we've spent talking about how best to deal with them.

So, ideally, a cross-site solution emerges where the questions of jurisdiction and possible, but narrow caveats for abuse aren't an issue. I don't know if we're ever going to get there, though - giving mods or users on Site A the ability to remove users from other sites (even though it would only happen in very blatant cases) is something I think we all want to find a way to make work, but can't quite get comfortable with.

To that, raising a flag seems like the best idea in the interim so we can be sure that all posts by spammers actually get removed (and the spam systems tuned), which (unfortunately) is still going to require a moderator to roast the account to be sure.

Putting this in review to be picked up once we've finished working out how moderator tools in general need to be tweaked and overhauled so they can more easily be set to "Public" / "Enterprise" / "Channel" configurations which is going to take a little bit.

I still wish we could find a way to just automatically roast these accounts once we get a little signal from a couple of trusted sources, but cross-site actions that destructive seems to set a precedence folks just don't feel all that great about setting, and we can appreciate why.

  • 1
    I agree that it would be great for all of us to waste less time on this. I'm a little leery of automatic cross-site action because sometimes "offensive" needs to be judged in context, and maybe we want some guardrails around that. Religion sites needs to continue to be free to mark proselytizing from other religions as offensive, but maybe you don't want to kill their accounts on Super User or Gardening over it, let alone Christianity or Islam where those exact same posts would probably get upvoted. It's tricky. Maybe decoupling "spam" from "rude/abusive" would suffice? Auto-kill spam only. – Monica Cellio Jan 2 '18 at 18:31
  • How about this: If a user’s only action on the entire network (apart from editing their profile) was a spam post, nuke their entire network account. In this case, nothing valuable can be destroyed (in fact, the user can just make a new account as easily or difficultly as they can now). In the rare event that there was activity on other site, just raise an auto-flag on these. – Wrzlprmft Apr 26 '18 at 13:11
3

I agree. Right now, the only way I know of to find these posts is to search for something like the following on a regular basis:

deleted:yes locked:yes created:7d..

Posts deleted as spam or rude are locked on the way out, so this finds them -- but it also finds posts flag-deleted by moderators, which you (and I) would like to skip, and it finds other things occasionally like migration stubs. So the results are a little noisy.

Further, a manual search like this relies on moderators remembering to do it, and it doesn't lend itself to coordination among moderators on a site. Some of these posts are seen by several mods, others by none. A flag, on the other hand, can be dismissed after one mod has looked at it and done whatever needs to be done.

I'd like SE to tell me about these cases somehow. If there's concern about flag floods, it could be on a delay -- only flag if the user account is still intact some number of hours later, so those that moderators deal with in ordinary site use don't raise flags. (A user with a decent level of activity can't self-delete immediately, so putting spam-deletion flags on a 12-hour delay wouldn't make it easier for vandalism to go undetected.) I don't think it's practical to detect cases where a moderator saw it and decided not to destroy the user, but those are pretty rare in my experience.

  • @MichaelKjörling as a moderator, I deal with spam by using red flags (to impose the penalties). I didn't pick up on the nuance of "community flagging without any of the flags being from mods", sorry. – Monica Cellio Jan 2 '18 at 16:26
  • "I don't think it's practical to detect cases where a moderator saw it and decided not to destroy the user, but those are pretty rare in my experience." I think the only times I've come across such cases are those where a post was incorrectly flagged as spam, and that's rare. – a CVn Jan 2 '18 at 16:44
  • @MichaelKjörling if it was incorrectly flagged as spam by a moderator, then you probably want another moderator to see it anyway. But yes, that's very rare. – Monica Cellio Jan 2 '18 at 16:54
  • Yes, if a moderator incorrectly flags as spam, you probably want another mod to see that; but in those cases, the existence or non-existence of an autoflag like this one isn't going to make any real difference, as the moderator flagging as spam is likely to just dismiss the autoflag as well. So you need other safeguards in place to catch that. I really only think of this autoflag as a way for moderators to become aware of content nuked by non-moderators, so that the account can be reviewed and if appropriate nuked too. – a CVn Jan 2 '18 at 18:13

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .