91

My common workflow when I see a blatant spam post by a 1 rep user is:

  1. flag as spam (when I'm a mod this auto-deletes)
  2. go to user profile
  3. As a mod, I destroy the user. As a regular user I spam flag the other spam posts linked from the profile

These are a bit more clicks than absolutely necessary. While it is not terribly important to optimize this, it would be nice if we could perform these steps with even less effort.

My main assumption is that a 1 rep user that has one of their posts flagged as spam until deletion is almost certainly a pure spammer. I might be wrong there, but I've never seen an example that violates this rule.

Having to go to the profile is an unnecessary step, but I always do that to determine if the user has created more spam posts. Destroying is also not necessary, but it is quicker if the user has multiple posts, and if I have to go to the profile anyway it's also only a few clicks away.

To make this quicker, I suggest that users that have never earned any reputation and that have all their posts deleted by spam flags should be auto-destroyed. Destroying accounts doesn't really slow down spammers much or at all, but it is somewhat tidier and many mods do it anyway, even for single post spammers. It would also serve as feedback that this user can't possibly have more spam posts, preventing other users from checking the user's profile without need.

When I spam flag a post from a 1 rep user as a mod, I'd like to get immediate feedback on how many other posts the user has, and maybe even a quick way to directly preview those posts and destroy the account.

This would of course not happen for posts from real, active users due to the 1 rep limitations. Excessive self-promotion and other borderline spam issues are usually not dealt with by account destruction. This restriction also severely reduces the potential for abuse of this feature.

As for leaving accounts alive to gather information about spam URLs and/or IPs, this is something that should happen behind the scenes anyway that that we mods should not have to think about. The community and mods provide enough information by spam flagging, it would just have to be collected and aggregated by SE internally (and maybe this already happens).

  • 1
    Yes, this would also, theoretically, enable ordinary users to spam flag a user to deletion, which I agree with. It's annoying to finally get rid of a spam post only to have the user try again 20 minutes later because a mod wasn't around to catch it. – ben is uǝq backwards Jan 28 '13 at 12:13
  • Wait, if Mods flag a post as spam it autodeletes? I didn't know that. That means the gold Marshall badge is back on my agenda! – JonW Jan 28 '13 at 12:19
  • 1
    @JonW Unfortunately, mods aren't eligible for Marshall. – yannis Jan 28 '13 at 12:24
  • 2
    @JonW: A mod vote is binding. Spam/offensive flags are votes (at 3, the post is banished from the front page, and at 6, the post is locked+deleted and the user gets a hefty rep penalty) – Manishearth Jan 28 '13 at 12:31
  • @Yannis: Really? :( – Manishearth Jan 28 '13 at 12:31
  • @Manishearth: Ah, I didnt realise a spam/offensive flag counted as a vote. Any reason why a mod would flag instead of directly dealing with it then (aside from the time saving pointed out in this question?) – JonW Jan 28 '13 at 12:33
  • 1
    @JonW: rep penalty, and the locking. 10k users can't touch it. Not necessary if you trust your 10ks, but yeah, it's there. I dive-bomb spam flags all the time :) – Manishearth Jan 28 '13 at 12:35
  • 2
    How prevalent is spam on the big three? On the smaller sites I hang out on we can go days between spam incidents; I gather that's not true everywhere. Are we talking tens per day, hundreds per day, ...? – Monica Cellio Jan 28 '13 at 15:20
42
+100

I'm not sure that I'd want this account auto-deletion, because it might make hunting down other spammers a little harder. (See my update below.)

On Stack Overflow, we often get clusters of spam accounts from the same location, so my usual spammer workflow is to delete the spam post, then pull up the account information for the spammer. If they have other spam answers, I deal with them as well, but I make sure to read each of the answers to avoid removing a legitimate answer inadvertently.

Finally, I use their account as a launching-off point and examine others that have shared a location with this one. This has helped me track down rings of spammers that somehow slipped through the cracks (or were upvoted during our recent /review shenanigans). If the account had been auto-deleted, I would not have been able to perform this kind of investigation. Only when all spammers from that location have been dealt with do I destroy the account(s).

This might be unique to Stack Overflow, given our traffic (and thus spam) volume, but I think the auto-deletion of spammers might hurt us more than it helps. In fact, as a moderator, I might appreciate if the spam flags would hang around even after a post was automatically deleted by a pile of community spam flags. This would point our attention to these users, rather than having their posts be silently deleted.

Update (Sept. 23, 2015): In the time since I wrote the above, we have been given tools to find other accounts at the same location as deleted accounts, so that part of my argument no longer holds. Also, the anti-spam system now does a good job of immediately halting posts from a location hosting identified spammers, which greatly reduces the frequency of repeat spammers from the same location.

After seeing how this has changed my workflow, I now support the automatic deletion of accounts whose only posts were all destroyed via 6 community spam flags or one hard moderator spam flag. I don't see a great reason for keeping those around any longer, and this destruction could save us some time.

  • 21
    Tracking spammer IPs and URLs should be independent of the account destruction and done when spam posts are flagged. We should not have to let the spam accounts live just to investigate this. It would be far easier if we could get automatically aggregated information about this, so that we could see at one glance from where the spammers come from and which URLs they spam. – Mad Scientist Jan 28 '13 at 17:18
  • In response to: "the anti-spam system now does a good job of immediately halting posts" - Does this type of tracking method also contain data for blacklisted/known urls? For example: s9fashion.com which I recently flagged as spam. I also noticed that the account was deleted not long after; was this account (also) auto-deleted? Note/FYI: I found this post while Googling: "are spammers automatically deleted?" which was on page 1 of found results. – Funk Forty Niner Oct 29 '17 at 23:03
  • @Fred-ii- - ChrisF destroyed that user in response to your flag. The system doesn't auto-destroy users yet. We have a limited manual blacklist of URLs, but even moderators aren't exposed to the inner workings of the automated spamblocks. We tend to add sites to the manual blacklist only when they become an extreme problem. The Smoke Detector bot has a broader range of things that it searches for, and it does a pretty good job of identifying things that slip through. Community review is always there as a fallback, and catches pretty much all of the rest. – Brad Larson Oct 30 '17 at 14:30
  • @BradLarson I see, thanks Brad. Well, we're all looking out for those and flagged respectively. Anything I can do to help out, cheers. – Funk Forty Niner Oct 30 '17 at 14:54
12

I think that an important part of spam prevention consists of two parts:

  1. Making spamming more time consuming for the spammer. If it takes more time, it's worth less money.
  2. Making treatment trivial and swift.

I think that if a user is

  • Very new (1 week or less)
  • Has no positive contribution (no upvoted posts)
  • Has his posts flagged as spam six times

We can be pretty confident that the user is a spammer and we want him gone ASAP. Since moderators can now see the last IP address from which deleted and destroyed users come from, an automatic flag about a user being destroyed by spam, along with a possible link to the relevant investigation pages mods have will seriously ease up the detection and removal of spammers.

As of now, if a spam post gets deleted by spam, there's no indication about it to the mods, which makes it all the harder to detect and treat. In 3 years I've been participating in SE, I can probably count the number of times I've seen anyone getting hit by the -100 penalty on one hand. It's ineffective and not deterring at all.

7

Update: I now support automatically deleting users whose only participation was deleted by rude/spam flags. SpamRam has gotten better and trolls worse since I wrote this answer.


I agree with the argument in this answer that auto-deletion would impede investigation of some cases. I've kept spammer accounts around for a few minutes (or even a few hours) while I investigate something, so this is not unique to SO.

What would help me, though, is if community-deleted spam automatically raised a moderator flag so we can clean up the user. Sometimes spam is deleted without mod intervention (good! the tools work and the community helps!), and unless mods come across it in our normal use of the site or someone points it out, we'll never notice and the account will hang around. An auto-flag against one of the user's deleted spam posts, optionally with a one-click option for "destroy user as spammer", would help mods clean up after them efficiently. The "destroy user as spammer" option shouldn't be presented if the user has rep or upvoted posts, but the flag would still tell mods to take a closer look.

Even if the "destroy user as spammer" button isn't feasible, I'm willing to click through to destroy the user -- the auto-flag still helps here.

2

I was about to post a a few months ago when I first saw this. Here are my original thoughts:

If the user meets all of the following criteria, destroy it automatically:

  • The user has only 1 reputation (so those with the association bonus would be exempt from this)
  • The user has at least one post deleted as spam (exclude rude/abusive if possible)
  • The user is no older than 1 week when the first spam deletion occurs
  • It's been another week since the spam post was deleted and the user has no visible activities
  • The user has no other visible posts

I'll explain the last two points of my idea.

One week after spam post deletion

This is to assure that a real user isn't accidentally deleted as spammer, though re-creating an account shouldn't be a big problem if the user sincerely wants to participate in Stack Exchange. In (almost) all cases, a spammer will never re-use an account. He will simply proceed with creating more accounts. A potentially real user should know his fault and never do that again, and probably add another post after learning the basic rules here.

No other visible posts

(Though I think this could be obvious.) If a post is visible, it means that the community accepts its existence, even if it is a bad one. If a post is deleted, or i.e. non-visible, it means that the community thinks it adds no value to the overall quality or the comprehensiveness of the contents on Stack Exchange. That's (IMO) the primary difference between a low quality post and an actually deleted post, so I'd prefer to add this as a criterion for auto-deletion filter.

I think the more filters available for an automated process, the better the process can do. We can (by far) never expect an automation to work as accurately as a real human (moderators), so the best we could do is teaching it more rules to work with.

  • 3
    The user is no longer than 1 week old when the first spam deletion occurs – What’s the point of this criterion? A considerable portion of spammers create accounts that lay low for longer than this before posting and would be excluded due to this. – Wrzlprmft Dec 7 '17 at 16:51

You must log in to answer this question.

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