9

For the past few months, there has been an on and off massive spam wave primarily focuses on Super User, which has encompassed thousands upon thousands of spam posts, all of which need to be nuked through 4 spam flags in order to be deleted. Smokey, an automatic spam detector bot, detects just about all of this and other spam here and around the network, and has been doing so since 2016. In addition, for posts where, based on historical accuracy, the post is 99.75% or more likely to be spam, the post with be autoflagged, giving it 3 out of 4 necessary flags to get it deleted.

Unfortunately, this still leaves at least 1 human who needs to interact with spam posts and flag them. When there is a wave of thousands upon thousands of spam posts coming in, this is quite annoying to do, and takes a reasonable amount of time that could be spent elsewhere. Due to the sheer volume of posts, during low activity periods spam has also been remaining for hours, such as in this image

enter image description here

In many cases, however, there really is no reason any human needs to be involved in getting rid of the spam. The spam could be gotten rid of automatically, through autonuking. In addition to not involving a human, and being more accurate human flaggers, it would have the added benefit of removing the spam immediately (meaning the spammers have no chance of getting anything of out their spam). Wait, but wouldn't autnuking nuke legitimate posts, you ask?

No, it wouldn't, not if done reasonably well. In all the time Smokey has existed, there has never been a post on Super User that has a "reason weight" over 269 or a "reason count" over 6. And yet, in that time, there have been no less than 11027 spam posts matching those conditions, include a huge portion of this recent spam wave. That is more than 34.4% of all spam on Super User during that period.

But we could also autonuke certain keyword that would never appear in legitimate posts and appear in thousands of spam posts. For example, this site alone has over 2,000 spam posts with "loan APP" in the title and another 2,000 with "HELPLINE number" in the tile. These term has never appeared in any non-spam posts, ever, in all the history of every Stack Exchange site. We could probably safely autonuke this too, could we not? And these 2 terms are by far not the only 2 examples of this.

About 40% of all spam (and much larger portion of this spam wave) could be autonuked with no false positives quite simply. This is really a much better way to mitigate the spam problem here.

Really, the only arguement I've heard against this is that by autonuking spam users would no longer be engaged in moderating it. But this is both not entirely true and good thing. For one, while much of the spam wave could be autonuked, a significant portion couldn't, which could still be moderated by the community. But, I think at least, that less people needing to spend valuable time that could be used to improve the site on flagging spam...is a great thing. Why wouldn't we want that?

So, could this site please get autonuking for posts where such autonuking would have 100% accuracy?

8
  • I wonder how who’s would interact with our current, manual spammer nuking processes. We use a mix of the front page, SD rss feeds and user reports. We nuke the accounts not the spam Commented Feb 12, 2025 at 2:50
  • 2
    @JourneymanGeek Almost all the spam is being nuked by the community, not through mods nuking spammer accounts. But it would just remove some work (keep in mind that a 2-3 deleted spam posts will get the account automatically q/a banned for 7 days). Commented Feb 12, 2025 at 12:52
  • @JourneymanGeek Also, could you make this [status-review]? Commented Feb 12, 2025 at 12:53
  • 4
    Not really feeling that this is the best option- I think the human oversight is worth it - and SD is community run so I am unclear why we need a staff ticket for it Commented Feb 12, 2025 at 13:09
  • @JourneymanGeek Per charcoal they would do it but they need CM or that sites mod team approval. There is still human oversight through metasmoke feedbacking Commented Feb 12, 2025 at 13:14
  • 1
    @JourneymanGeek Per SE guidance, a site cannot have automatic spam deletion from Charcoal without the explicit permission of SE staff and the site in question's mods. Creating a staff ticket here would ensure the former. Commented Feb 12, 2025 at 18:51
  • Unfortunately, this is not a feature that SE staff or SU moderators could implement. You would have to talk to Charcoal directly, they are the ones who set up how autoflagging works. Commented Feb 14, 2025 at 3:43
  • @CPlus Per makyen charcoal could easily create a spam wave which would make such posts be autonuked and set that up less than a few minutes. But they would need approval first Commented Feb 14, 2025 at 11:41

3 Answers 3

Reset to default
5

Its still in the works but coming soon - but practically there's an upcoming, new spam prevention mechanism that should help somewhat independently cast a final flag. This wasn't public (and I'm hazy if I was aware of these plans) at the initial time of posting and has no baring on our decision. It should let automated systems handle the worst of it

I also asked about statistics from the community department - there's quite a bit there (mostly pretty good!) but critically (and I asked for permission to share)

To summarize in a handy way: Since Nov 2024, a majority of spam posts on Super User are removed within 2 minutes of posting. Going by the weekly breakdown, that mark is currently at about 3 minutes and has been hovering around there since late March.

Which isn't half bad ;)

1

For example, this site alone has over 2,000 spam posts with "loan APP" in the title and another 2,000 with "HELPLINE number" in the tile. These term has never appeared in any non-spam posts, ever, in all the history of every Stack Exchange site.

But now it has (although not in the title).

Sad fact is that combating spam often requires quoting spam, which confounds Bayesian detectors.

1
  • 2
    Well I wouldn’t think we should put it on the meta site, obviously. Only the main site is getting spammed Commented Apr 15, 2025 at 17:16
-1

I'm going to say no for now.

For most part, the levels of spam are manageable and we're destroying accounts - which feeds spamram, and reporting information for some other tools that help mitigate spam network wide.

Ironically, that people are reporting spam and spammers manually also means better engagement for 'other' curation tasks for us as well

If we get hit by thousands of spam post like webapps was and are overwhelmed - we will certainly ask for help. Right now we're getting a lot more value from keeping the cleanups semi-automated.

We're keeping track of how bad the situation is, working with the community team and others to mitigate the problem, and certainly will reach out if we need help. We're doing fine at the moment however.

8
  • But time spent reporting spam is necessarily time not spend with other activities, like curation of content creation. What value is given from letting spam remain on the site longer than it needs to (yes I know many of these posts are nuked in under a minute, but that's a minute longer than they need to remain there). And keep in mind this wouldn't eliminate all spam. But it could also, hopefully, discourage the spammers from spamming, since there is 0 chance anyone will see most of their spam. Commented Mar 11, 2025 at 18:05
  • I'm not looking at the time reporting spam - If people are looking out for spam, they're also looking at the site for other things - stuff SD misses out, the mysteriously cryptic title that needs decryption. Basically the people who do these things manually are the sort of people I want on my site. There's also other measures for preventing spam that needs me to have visibility on spam coming in which this is helpful on. You're focused on one problem and one solution - I'm trying to think more broadly about site health . Commented Mar 12, 2025 at 2:04
  • Can you reconsider? Its really getting out of hand making it less fun for me, an actual SuperUser to visit the site. This is not just hurting spam accounts, its hurting its actual userbase. Commented Mar 13, 2025 at 9:10
  • We'll discuss this - I'm still skeptical, but we have mods in charcoal who would have better perspectives than I Commented Mar 13, 2025 at 9:24
  • Thanks. It almost seems like they're doing a campaign, based on the sheer volume of spam we're currently receiving. Commented Mar 13, 2025 at 11:17
  • Oh we've been dealing with this for 2 years + across the network too. We occationally have extra bad periods, and we're getting hit by half the spam in the network for some reason. We know there's a problem, we're just disagreeing on how useful this measure would be, compared to everything we're doing now - we're monitoring and using SD reports, user reports, flags and a few other steps, and visibility of spammers is useful to us Commented Mar 13, 2025 at 11:21
  • @JourneymanGeek Whatever happen with the "reconsider"ing? 3 months later and this site is still under massive attack. Could we at least try this for a month or something, see how it goes? Perhaps there's a few users who have engaged in other moderation tasks because of flagging spam, but there are probably far more who are sick of seeing spam everywhere on the site and have just left. Commented May 16, 2025 at 14:56
  • I haven’t seen this spam slow down. We still get 30-40 posts every single night. I even see a few of those posts upvoted. Something should be done. Commented Sep 17, 2025 at 4:06

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.