8

Of course spam posts should be flagged, and if they come up in review, rejected.

Over on Mi Yodeya during the past 24 hours I’ve helped delete and reject four spam posts through the review queue, and I can only imagine that others were posted over the course of the night. To clarify, I’m referring to edits and answers that were about 99.274% words that I don’t think any decent person would use in everyday discussions, even if someone got them really angry.

Thing is, if the only thing we can do as a community is delete his posts/reject his edits, and as moderators to ban his account, can’t he just create a new account? Further, some of these posts . Maybe the mods can ban his IP address? I’m not one, so I have no idea what kind of power they wield in that respect. But he can always just get a new computer.

I don’t have the numbers to see what this is like on other sites, but I know this doesn’t happen often on Mi Yodeya. I wonder, then, whether this is the optimal solution, or whether an automatic spam filter be put into place - say, a new review queue that automatically deletes posts or rejects edits that can then be voted on whether to be reposted if it’s a false alarm. Perhaps it can be automatically triggered by certain keywords (like ****, %$#^, or @$+^).

Now, all of these posts that I voted to delete/reject came up in Suggested Edits and First Posts. Maybe such edits would show up in Low Quality in general, and that just wasn’t the case here because of precedence rules. I don’t know. But perhaps the above would be a better solution, if this is even as big of a problem as I’m making it out to be.

  • 2
    You might be interested in this. – Glorfindel May 18 '18 at 13:42
  • As a matter of terminology, what you are describing sounds like rude/abusive. The spam flag is intended for repetitive promotional posts, which is what Charcoal focuses on. There are other efforts e.g. for identifying abusive posts and comments (and lots of other automation) in the Stack Overflow SObotics community, which recently also branched out to cover the rest of the Stack Exchange network with SEbotics. – tripleee May 19 '18 at 15:54
16

It should, and it does. The system is named SpamRam, and there are a handful of existing meta posts about it.

It gets trained by people like you flagging various badness. It works on first-class posts (i.e. questions and answers); not sure about comments or suggested edits. It's not account-based, but spammer accounts are typically deleted quickly through other means.

There's also SmokeDetector, a community project that tries (very effectively) to catch spam missed by other means.

Automatic word filters have been discussed and rejected to avoid "clbuttic" type problems.

  • dang. Ninjaed - have an upvote. I'd add though, while its a little outdated, a good amount of spam is blocked before it gets in meta.stackexchange.com/questions/228043/… – Journeyman Geek May 18 '18 at 13:48
  • 1
    In addition there are IP bans from the system on IPs where there's been recorded spam posts. AFAIK they don't start out with an instant ban, but it gets blocked pretty quickly. – Olivia May 18 '18 at 13:49
  • 3
    In more detail - SpamRam works more with post metadata than it does with post content; SmokeDetector is the opposite. If you're seeing a lot of spam on a site, come talk to us in Charcoal HQ and we'll see if there's something SmokeDetector can do to catch more of it. – ArtOfCode May 18 '18 at 13:49
  • Awesome, thanks! Like I said, I haven’t been seeing an awful lot of it, but when you see an answer that’s entirely curse words, which, when I leave a polite comment on the OP to welcome him to the site and apologize for the rudeness that’s followed up by a second curse-full answer... It seemed to me that something wasn’t working right. If this persists throughout the day I’ll absolutely be coming over to y’all @ArtOfCode. – DonielF May 18 '18 at 14:38
  • though, hm, none of our tools catch bogus edit attempts... – Journeyman Geek May 20 '18 at 12:34
0

In addition to SpamRam, used for posts, there is also a content filter used for suggested edits from anonymous (logged-out) users. From my experience with it, the filter covers a very wide scope and is frequently subject to false positives. When someone trips, their edit submission silently fails, and if they try to submit another edit, they'll see a notice that they've been banned from suggested edits.

Also, to reduce the incentive for spammers to spam via suggested edits, links to edits rejected as spam show as false 404 errors, so spammers can't show whoever's paying them a page with spam.

You must log in to answer this question.

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