20
$\begingroup$

Has anyone noticed the site getting a lot more spam posts than usual over the past couple days? The ones I've seen repeatedly have "BingX referral codes" or similar codes for Bitcoin or other cryptocurrencies or something. Are the moderators aware of this problem? Is there something normal users can do to help?

EDIT: Something further I've noticed about this recent spate of spam bots is that they sometimes post in languages other than English (I've noticed Turkish in particular but I think there are several).

$\endgroup$
8
  • 8
    $\begingroup$ Flag as spam as soon as you see it. Enough flags and it gets deleted (and some other things happen behind the scenes). $\endgroup$ Commented Apr 18, 2024 at 3:11
  • 5
    $\begingroup$ See also: What should I do when I see a spam post on MO? (Although one could say that this is briefly summarized by David Roberts' comment.) After the next update of SEDE one can use the database to see the number of spam posts - some similar queries can be found here: To which extent should spam posts on meta be considered a problem? And one could also check the stats available in metasmoke. $\endgroup$ Commented Apr 18, 2024 at 11:41
  • 7
    $\begingroup$ @JosephVanName: While there is no obligation to comment when you vote to close, doing so and then making a comment like you did is very unprofessional. I would delete your comment if I were you. $\endgroup$ Commented Apr 18, 2024 at 12:20
  • 7
    $\begingroup$ (I have also upvoted both this post and the answer to counteract what I presume are JvN’s vindictive downvotes) $\endgroup$ Commented Apr 18, 2024 at 12:21
  • 14
    $\begingroup$ We're helping out also, doing all we can to flag as soon as possible. Yesterday alone, there was over 55 reported of those. $\endgroup$ Commented Apr 18, 2024 at 14:56
  • 4
    $\begingroup$ Thanks, @Cow . I also pointed to Charcoal on Alex's answer, before seeing that that was what you linked to. $\endgroup$ Commented Apr 19, 2024 at 2:07
  • 7
    $\begingroup$ On the flip side, think about all the shiny new marshal badges. $\endgroup$ Commented Apr 23, 2024 at 9:30
  • $\begingroup$ I’m voting to close this question. $\endgroup$ Commented May 30, 2024 at 22:06

2 Answers 2

Reset to default
16
$\begingroup$

Yesterday (April 17th) the site was "under attack" between (approximately) 08:00 UTC and 11:00 UTC. During this period, I myself flagged 23 posts as spam. Besides downvoting and flagging as spam, there is nothing that can normal users can do.

IP-based blocking is not trustworthy, because spammers can use Tor or VPNs to hide their origin. On the other hand, moderators could, in principle, install keyword-based filters (for words like "coupon", "referral code", "bonus", "sign up"), so maybe they should step into this discussion and comment on this possibility. (Of course, such filters introduce the possibility of false positives, but for the list of words given above the probability of this happening should be very low.)

$\endgroup$
8
  • 6
    $\begingroup$ "moderators could, in principle, install keyword-based filters" - perhaps you'd be interested to see charcoal-se.org then :-) $\endgroup$ Commented Apr 19, 2024 at 2:02
  • 4
    $\begingroup$ And see also stackoverflow.blog/2020/06/25/… $\endgroup$ Commented Apr 19, 2024 at 2:05
  • 9
    $\begingroup$ Blacklisting coupon could be a nuisance because of questions about variants on the coupon collector's problem. There are three legit questions with that keyword just in the past month. $\endgroup$ Commented Apr 19, 2024 at 9:05
  • 5
    $\begingroup$ I suspect “bonus” is also somewhat frequent (in phrases such as “bonus question”). $\endgroup$ Commented Apr 19, 2024 at 12:33
  • 1
    $\begingroup$ The tendency to co-opt common phraseology into mathematical parlance with the intent of ‘capturing intuition’ may prove an obstacle here. $\endgroup$ Commented Apr 20, 2024 at 2:00
  • 3
    $\begingroup$ Is there a way to do an "n-strikes-blacklist" - so e.g. a post that uses each of "coupon", "bonus", and "referral" gets hit, but using just one of those is fine? $\endgroup$ Commented Apr 20, 2024 at 5:04
  • 2
    $\begingroup$ @AlecRhea: Agreed, but I was thinking about these words being present in the titles, not the bodies of the questions. $\endgroup$ Commented Apr 20, 2024 at 7:21
  • 3
    $\begingroup$ Probably it is worth mentioning that some of the keywords mentioned above have been added to the watchlist by the Charcoal team. Of course, this is a volunteer project, independent from the internal Stack Exchange system. $\endgroup$ Commented Apr 24, 2024 at 14:15
4
$\begingroup$

This answer is community wiki - feel free to edit it if you have some useful information to add.

In case somebody wants to have a look at some stats about spam - for example to compare the current situation with the situation in the past - you can have a look at metasmoke or you could use SEDE.

Some basic information about Charcoal, SmokeDetector and related projects: https://charcoal-se.org/

In metasmoke you can find various stats and details about posts which were caught by SmokeDetector or reported manually. (So this isn't necessarily the same as the list of post which have actually been flagged as spam.)

You can find the list of posts on MathOverflow and MathOverflow Meta. If you look at details of some posts, if there is a small red flag on the top left, you can find logs of flags by Charcoal users. Here is an example of a flag log. (It is certainly valuable that volunteers from the Charcoal project help with removing spams on various sites in this network, including MathOverflow.)

In SEDE, one can find spam flags in the Votes table, using VoteTypeId=12 - which denotes spam. There is also VoteTypeId=4 meaning offensive. In some of the queries below you can change parameters to switch the parameters deciding which VoteTypeId is used.)

You should keep in mind that SEDE is updated only once a week (on Sunday) - so the stats shown here do now count the most recent posts.

  • Number of spam posts per month: main and meta.
  • Number of spam posts (questions, answers) per year: main, meta.
  • Number of spam posts per day: main, meta. (You can change the parameters if you want to see a different date range.)
  • List of the posts flagged as spam between the two given dates: main, meta. (These queries return the links too - but only the users who can view the deleted posts will actually be able to get to the content of those posts, too.)

For example, from those queries we get 182 spam posts in April 2024, which is so far the highest number.

Graph

Perhaps it is useful to see also the stats showing how fast the spam posts are deleted.

  • Time until spam posts were deleted: main, meta.
  • Average time until spam posts were deleted: main, meta.
  • Median time for spam post deletion: main, meta.
  • Median time in 2024: main, meta.

As of June 2025, the above query returns median 19 minutes on main and 43 minutes on meta. After restriction to 2024, the median is 7 minutes on main and 12 minutes on meta.

Some stats on spam are sometimes mentioned in the MO chatroom.

$\endgroup$
4
  • 3
    $\begingroup$ So if I'm reading the data correctly, the current month of April 2024 is indeed way above average in terms of number of spam posts on the main site. $\endgroup$ Commented Apr 24, 2024 at 15:49
  • 4
    $\begingroup$ @SamHopkins Yes, that's correct. Data in SEDE are as of April 20 and it already shows more posts than in any of the previous months. (I plan to add the total number for April after the end of the months.) I wonder whether this is similar to what we would see if MO decides to leave the SE network. $\endgroup$ Commented Apr 24, 2024 at 16:02
  • 3
    $\begingroup$ I have posted an updated graph, so that we can compare the situation with the months after April 2024, too. It seems that this was indeed a peak. I will post some more stats in the MO chatroom. $\endgroup$ Commented Jun 7 at 14:25
  • $\begingroup$ I think a number of posts marked as spam now are actually genAI, and people are flagging them as spam. Since they superficially look like mathematical text, and they aren't blatant commercial spam (advertising a product or business), then they are slower to end up being deleted. $\endgroup$ Commented Jun 17 at 6:06

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.