What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 150 Stack Exchange communities.

Is something wrong with StackOverflow's spam filter? Recently there have been huge numbers of spam topics related to football streaming. They have all been spam flagged and deleted but there are more getting posted all the time, I just flagged another 6 or so a few minutes ago. This has been going on for around 3 days now with these topics getting constantly posted over and over again in the tag.

enter image description here

share|improve this question
The spam is annoying, and they keep coming over and over again... no-life spammers, bah. – Tredged Aug 11 '13 at 17:43
"Is there something wrong with out spam filter": nah, you seem to be working fine ;) – Gordon Aug 11 '13 at 17:46
The hell do we have a NFL tag for? – animuson Aug 11 '13 at 17:47
The spam filters (i.e. users flagging) seem to be working fine to me :) I'm not seeing anywhere this number of flags in the mod queue so you guys are knocking them out very quickly indeed. – ChrisF Aug 11 '13 at 17:47
@animuson: I was about to ask the very same thing; nominate for burnination, maybe? Or perhaps anyone asking a question that they tag only 'NFL' has their question dropped, and ip blacklisted? – David Thomas Aug 11 '13 at 17:47
@ani keeping all the spam in one easy to find location. – Flexo Aug 11 '13 at 17:48
@BadWolf - there is some automatic blocking of known spam, but that has to be maintained and can be circumvented. – ChrisF Aug 11 '13 at 17:51
Please don't automatically block it. I'm trying to badge whore here. – PeeHaa Aug 11 '13 at 17:52
I'm curious what will happen now that they have no NFL tag to play with... – animuson Aug 11 '13 at 17:55
@animuson Now that tag is gone they are powerless!!!! chefmonsta.com/wp-content/uploads/2012/02/mini_me.gif – PeeHaa Aug 11 '13 at 17:57
It's not recent, that guy's been at it for months. They seem to be good at working around the spam filter and keeping up with evolutions. – Gilles Aug 11 '13 at 18:01
Man, am I missing out on all the flagging fun?! – Martijn Pieters Aug 11 '13 at 18:18
Once again, the use of link shorteners to work around URL blacklists for this spam has me questioning their utility here. – Brad Larson Aug 11 '13 at 19:02
@psubsee2003 - The tag seemed legit. I was looking at it the other day because of the spam. There were 6 or so questions tagged nfl, several asking about football APIs. I was going to post here recommending we delete the tag, but then remembered this discussion on Sudoku, in which that tag was not deleted. – chue x Aug 11 '13 at 19:11
@chue x: It will be deleted automatically in the next day since there are no questions that use it. It's not the same as sudoku, though: no one is writing a program to run the NFL, and you're right to think that it's unnecessary. If there were an official NFL Data API, then nfl-api or the like would be good, but "Getting schedule data for NFL games" tagged:nfl is like "Getting World Sudoku Championship scores" tagged:sudoku -- the meaning of the data you're trying to get is almost certainly irrelevant to the technical issue. – Josh Caswell Aug 11 '13 at 19:21

3 Answers 3

up vote 88 down vote accepted

We're trying to get better at keeping this at bay. The problem affects all sites that see any kind of traffic, some more than others. I can't go into very much detail because frankly, there isn't a whole lot of detail that belongs to an agreed implementation at this point.

However ....

A big mistake we have made in the past is not fully utilizing the signal that we get from moderators and the community when destructive actions are taken. We know when moderators destroy or delete accounts, but we don't know why they did it, at least not programmatically. A change working its way through implementation now tracks this by doing something remarkable, we actually ask them to indicate a reason for the action.

When this is implemented, we have a much easier time querying across the network to better visualize these fools as they move from site to site, occasionally changing origin. This makes the data that they inadvertently leave behind much more valuable when it comes to automatically reacting to, and subduing these sorts of spam floods.

It's a bit of a trick, we don't want to prevent anyone from reading our sites, and even poorly written perl bots deserve a fighting chance in life. Still, we can be a little more picky about content we accept, especially when we have a much clearer picture about the recent behavior of the remote host. I'm not going to go into specifics that I don't yet have to offer, but it is a problem we're taking seriously.

Phase one of this (collecting reasons) should be out soon, then we need some time with the data we get.

That's not the only reason we implemented this, playing chat tag with moderators to get context when someone writes in asking why they were removed is sub optimal. But this is going to give us a much clearer picture of what we're dealing with.


The feature that requires a reason for account destruction is being pushed now, and will be live shortly. For transparency sake, here's the interface (click image for full resolution):

How can I kaboom thee? Let me count the ways ...

As you can see, certain reasons are much more interesting than others when it comes to tracking repeat abuse, since the extra signal is now being captured. Given the volume of crap that some sites see, it won't be too long before we have enough data to confirm / discover patterns.

share|improve this answer
Yay!​​​​​​​​​​​ – Ryan O'Hara Aug 12 '13 at 4:57
This is awesome! A simple idea, of course, but brilliantly executed. The UI looks well thought out, and everyone loves data. Except spammers. – Cody Gray Aug 12 '13 at 8:03
In a comment above someone wrote "editing spam is a bad idea (resets spam counter)". If the flags are indeed reset upon editing, then care to edit that into Should Spam posts be edited?, to make that official too? – Arjan Aug 12 '13 at 9:29
@Arjan I'm going to check with Jarrod to see what changes may still be in the pipe for spam flags, there's some recent UI oddities around them as well, so I'll answer that once I'm sure I'm correct. – Tim Post Aug 12 '13 at 9:43
-1 for Comic Sans :P – American Luke Aug 12 '13 at 13:54
To be clear, editing doesn't invalidate spam flags - rolling back an edited post removes the flags added after the edit, @Arjan. I've edited that post to clarify. The best reason not to edit blatantly off-topic spam is simply that it makes it harder to find and remove though. – Shog9 Aug 12 '13 at 14:33
@TimPost Just used this on a spammer. Thanks. – MPD Aug 12 '13 at 16:13
@American Luke: That's not even Comic Sans. – BoltClock's a Unicorn Aug 14 '13 at 9:53
@BoltClock'saUnicorn Looks like it at this resolution :P – American Luke Aug 14 '13 at 12:06
I'd love to be flagged on my site when a moderator destroys an account on another site. If someone looked at stackexchange.com/users/3208723/user2709183?tab=accounts you would see the sort of spam that propagating the destroy action to all sites where a linked account is present and has posted at least once would accelerate the cleanup until the filters can be tuned. – bmike Aug 23 '13 at 18:08
This guy is back! – H2CO3 Dec 18 '13 at 22:08
It's been a little while, I'm curious to see what patterns you found so far...? – Taryn East Feb 1 at 1:24
Whole lot of explanation for what a placeholder is… – bjb568 May 12 at 3:01
up vote 122 down vote

I've decided to attack the problem at the root, shutting down the spam blogs themselves.

The secret here is to run Google searches using queries like site:blogspot.com watch football (the site: operator as used here limits the search to the blogspot.com domain), click on and examine any suspicious results, and paste the URL of each spam blog into the reporting form on the "Spam, phishing, or malware on Blogger" page.

A few months ago, I did just that, and managed to shut down about 30 such spam blogs, mostly about boxing. If you do this en masse, this will likely catch the attention of Blogger staff, making it more likely that they will take legal or other action on the operators of these blogs since they are probably run by just a few people.

You'll often find that these spam blogs contain distinctive phrases like:

These spam blogs may also use distinctive images, such as the one in this Google Search by Image query. Using the above queries, I have identified and reported more than 200 spam blogs.

The posts in these blogs often include a long list of related terms like "live broadcast, live sop cast, live telecast, live web cast, live HD coverage". This kind of highly aggressive SEO is meant to overrun legitimate websites in search results, and is a clear indicator of spam. As an example on how hard they push SEO, I found this on one of the websites (in a spoiler block so as to not be disruptive):

Rugby,Rugby Live,Rugby Live Free,Rugby Live Stream,Rugby Live Free Online,Rugby Live Stream Online,Rugby Live Stream Online Free,Rugby Live Stream Online Here Now,Rugby,Rugby Live,Rugby Live Free,Rugby Live Stream,Rugby Live Free Online,Rugby Live Stream Online,Rugby Live Stream Online Free,Rugby Live Stream Online Here Now, Free Rugby,Free Rugby Live,Free Rugby Live Stream,Free Rugby Live Stream Online,Free Rugby Live Stream,Free Rugby Live Online,Free Rugby Live Online,Free Rugby,Free Rugby Live,Free Rugby Live Stream,Free Rugby Live Stream Online,Free Rugby Live Stream,Free Rugby Live Online,Free Rugby Live Online

Of course, I will flag any posts on Stack Exchange that are obviously spam as well.

If you try the searches I suggested in this answer and in the comments, you'll find that many of these blogs are now gone:

Blogger blog removed

share|improve this answer
This is awesome. Salting the roots like this helps to make the Internet a better place. – Tim Post Aug 12 '13 at 5:34
@TimPost: google.com/… – more than 40 blogs reported. I could do this all day... – bwDraco Aug 12 '13 at 6:37
google.com/… (sorry, the previous link got messed up) – bwDraco Aug 19 '13 at 13:20
You are my hero, by the way. – Andrew Barber Aug 19 '13 at 13:45
@AndrewBarber show it by some rep! Oh wait, Mat already did that... ;) – Shadow Wizard Aug 19 '13 at 14:05
@ShaWizDowArd Hmm... yes; that would be a great thing to use some of the rep I've gained from all the posts I've earned 'Reversals' on ;) – Andrew Barber Aug 19 '13 at 14:07
Here's the bounty notice for posterity: "Awesome Fiery Dragon SPAM hunters are awesome." – Mat i.stack.imgur.com/nqMy2.png – bwDraco Aug 19 '13 at 20:21
It would be nice if Google removed the search results at the same time as they destroyed the blogs. – Robert Harvey Aug 19 '13 at 21:26
I came to this answer myself too and came here to post it :-) Thanks. – Preet Sangha Aug 24 '13 at 22:50
A moderator at Academia has referred to this post to handle a similar spam issue. meta.academia.stackexchange.com/questions/903/… – bwDraco Apr 24 '14 at 14:35
This is a good solution in terms of taking a proactive approach, although some of these search terms show more than 1,000,000 results. If we resort to manual flagging it will take uncountable resources and efforts. Since there is spam on a massive scale,google can themselves bring such spam down by running these queries on a periodical basis or we should bring forward a public forum where these links can be listed for blogs like google to take action about or search engines should evolve to learn the spam behavior. – pal4life Apr 29 '14 at 16:05

The spam is not new. Drupal Answers was hit really hard with it starting in Dec/Jan. Other sites have had it bad, too (Ask Ubuntu was one). The SE team has done a fantastic job with the automatic filters, but the spam evolves rather often, so automatic blocking isn't effective for long.

The proper course of action is to flag the post as spam. Three spam flags will remove it from the front page, six will delete it. Don't edit it, don't downvote it, don't use another flag. Flag as spam and move on.

If a mod sees it before the autodeletion, they can destroy the account, which will take all spam from that account with it.

share|improve this answer
+1 for adding to not do anything else to it but flag as spam! This actually affects how quickly the content is nuked! – Andrew Barber Aug 11 '13 at 22:49
I know editing spam is a bad idea (resets spam counter) but why is downvoting it bad? – Richard Tingle Aug 11 '13 at 23:25
@RichardTingle Downvoting can mean that fewer people see the spam, so it ends up with fewer spam flags, so it actually sticks around longer and/or it doesn't end up with enough flags for auto-deletion. – MPD Aug 11 '13 at 23:34
why not use a custom flag in cases where autodeletion appears insufficient? – Jan Dvorak Aug 12 '13 at 9:17
@AndrewBarber - But if the spam is edited out of the post anyway it is less of an issue that it is not acted on quickly. On slower sites it might be half a day before the spam is removed. Surely it is better it does not contain the promotional content for that length of time thus making the spam pointless. – Martin Smith Aug 12 '13 at 10:34
@JanDvorak Spam flags have higher priority in the flag queue. Mods may not see the custom flag until they have cleared others out. – MPD Aug 12 '13 at 10:39
@MartinSmith When we are dealing with the flags, it means that we need to go through the edit histories to verify that the flags were correct. This causes more work for the mods. – MPD Aug 12 '13 at 10:44
@MPD On slower sites I can't imagine this will be a huge problem. I doubt they get that many flags anyway. Probably better to remove the spam from the post and create a small amount of additional work for the moderators then leaving it hanging there for a day. If spammers see that it survives for that length of time it provides an obvious incentive to spam more. – Martin Smith Aug 12 '13 at 10:50
Or perhaps spam flags that are unhandled for (say) 30 mins should get pushed into a network wide spam queue so they can be dealt with by moderators from other sites. On sites with only a few moderators it is unlikely there will be 24/7 moderator coverage. – Martin Smith Aug 12 '13 at 11:19
@Richard, as for invalidating flags, that's not fully accurate. See Shog9's comment. – Arjan Aug 12 '13 at 14:34
Why can't moderators nuke the account after deletion? (At least that's what it sounds like) – Cole Johnson Aug 17 '13 at 15:35
@ColeJohnson Moderators can destroy an account on every moment, if its reputation is not too high. If we notice a flag for spam on an account, we just destroy it, letting the associated posts be automatically deleted. – kiamlaluno Aug 21 '13 at 1:17
@kiamlaluno define "too high" for reputation, and why can't you delete accounts with "too high" of a reputation? – Cole Johnson Aug 21 '13 at 1:29
@ColeJohnson Only accounts with a reputation lower than 500 can be destroyed; the others can be deleted. The difference is that in the first case the posts created by that account are all deleted; in the second case, only the posts with negative score are deleted, while the rest of them is denormalized. Destroying is disabled because users with a reputation of at least 500 have probably positively contributed to the site, and their posts should not be automatically deleted, when the score is not negative. – kiamlaluno Aug 21 '13 at 1:43
@kiamlaluno So it helps with hacked accounts that go on a spamming spree and drown out the good stuff? – Cole Johnson Aug 21 '13 at 2:14

You must log in to answer this question.

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