What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 134 Stack Exchange communities.

Sometimes a site gets attacked by spammers/spambots. In such a case, the current procedure for moderators is:

  • Request a blacklist from the community team. This is basically a regex that is tested against the post; if there is a match the post is not allowed through.
  • Keep destroying posts and accounts while waiting for the community team to reply (This affects mods living in The East more, as there is only ~one community team member around)

This is rather inefficient, as moderators have to keep a constant watch on the site and pick of spam. A few weeks ago, Physics (and AU) were being hit by a certain spammer, and I had to keep picking off spam posts from the main page.

Blacklisting is a RIGHT NOW feature -- it is most effective when applied immediately.

After a few hours of waiting, the moderator has probably spent a significant amount of time (which can be better spent doing other things) watching for more spam and killing it. Besides, there's a chance that the spammer may get tired of it too (doesn't always happen, but there's a good chance it will).

With this in mind, I propose the following: Allow mods to set temporary blacklists (which expire in a capped number of hours -- IMO 24 hours is enough). These blacklists can be reviewed by the community team and extended/removed if necessary.

I understand that blacklisting is a feature that is prone to misuse and if not used carefully can lead to many false positives. With this in mind, further restrictions can be imposed:

  • Require two moderators to initiate this, similar to the current system for adding custom OT reasons.
  • Allow the blacklist entry only if it matches N recently deleted posts1.

1. Original idea credit @ChrisF: "It ought to be that if N users get destroyed or M posts get flagged as spam then any urls in the posts are automatically added to the blacklist or at least sent to the team for review."

share|improve this question

1 Answer 1

I think this could be done in a pretty safe way, avoiding all the potential issues that make the current blacklist so dangerous.

Post body blacklists should be limited to pure URLs (only the domain part), no fancy regex or something like that.

If a mod tries to add a blacklist rule, a collateral damage query is run that would list all existing users (for IP blacklists) or all existing posts (for URLs) that would be affected. It could also deny the mod the ability to blacklist if there would be significant collateral damage, so that they would have to run it by the comm team first. Those queries are probably rather expensive, but then again they are not meant to be run often.

share|improve this answer
1  
Note: The black magic spammer we had on Physics/AU did not post URLs (instead pulling from a pool of phone numbers and sometimes email addys). But I guess that's a rare event. –  Manishearth Jun 24 '13 at 10:46
    
@Manishearth Restricting the blacklist to plain text might be okay, but regular expressions are a bit too dangerous IMHO. –  Mad Scientist Jun 24 '13 at 10:57
    
of course. Actually, restricting it to URLs (and possibly email ids) is OK too. Spammers rarely go without posting one of those. I wasn't disagreeing with you, just noting that there are edge cases. :) –  Manishearth Jun 24 '13 at 11:02
1  
Whatever you do, make sure that I can't blacklist the whole Opera-mini proxy farm. ::mad cackling and much dry washing of hands:: –  dmckee Jun 24 '13 at 15:36

You must log in to answer this question.

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