Sunday, May 30, 2004

Comment Spamming

Just a few thoughts on how to fight comment spamming... First of all a few pointers on the subject:
So, what are the motivations behind comment spammers?
  • Posting millions of links in the web will statistically  make lots of them be clicked in the long run. This is the same motivation e-mail spammers have.
  • Blogs, due to their nature, give a major contribution to Google Pagerank. Having lots of blogs pointing to the spammers webpage makes their Page Rank and their visit number rise.
So, How can we stop them? There are three ways of stopping or controlling spam.
  • Not letting them post or making it harder for them to post.
  • Letting them post but then removing their comments from your blog.
  • Making it less actractive.
Not letting them post:
  • IP throttling - Spammers can easily get a 300000 blog list and comment in each one of them a a time making it hard to use this technic.
  • Black-listing words - This might help solving the problem. I know they can write S E X, or even s3x, instead of sex, but no one looks for that word in Google. If we were talking about spam e-mail it would be a different story.
  • Black-listing IPs - Doesn't seem to really solve anything as if someone can change IP easily that someone are the spammers.
  • Black-listing open proxies - I don't think this would solve anything. The only thing it would accomplish would be to annoy a couple blog users.
  • Mass Black-listing - That's what the blam project is all about. If you have several blogs working together they can do IP throttling much more efficiently.
  • Comment reviewing - Manually review every comment. 100% efficience but try receiving 20000 comments in one hour.
  • Bayesian filters - Might be a nice idea.
  • Disabling comments - Try to comment on my blog now ...
  • Robot detection - Detecting if the poster is a robot or a real human using technics like image analysis.
  • Honey pots - Fake posts used to maintain black-lists
Removing Comments:
  • Mass comment deletion and comment search by keyword and date - This would help where the other methods fail.
Making it pay less:
  • Redirecting all urls through a special page making the links not count towards Page Rank.
Bottomline: There isn't a magic solution that will solve comment spamming one and for all but implementing a wide range of solutions might help controling the problem.
Posted by André Restivo at 11:15:10 | Permanent Link | Comments (4) |
Comments
1 - My comment is here (Comment this)

Written by: Sérgio Carvalho at 2004/05/30 - 18:22:08
2 - Forcing image analysis with a highly usable interface for humans would be my bet. 0,2€ (Comment this)

Written by: Sérgio Nunes at 2004/05/30 - 18:47:51
3 - From now on Blog.com forces image analysis to allow comments to be posted. Issues still to be cared about: usability and accessibility. (Comment this)

Written by: André Restivo at 2004/06/01 - 00:37:51
4 - Just test driving this finnicky image thingy... It works! And it looks kinda nice... (Comment this)

Written by: Sérgio Carvalho at 2004/06/02 - 10:19:24
Write a comment