How Not to Fight Comment Spam
Until recently comment spam at work was easily identifiable because of the following style in the A tag.
’style=”display: none’
It didn’t take long for someone to suggest that all of our comment spam problems could be solved with just fifteen minutes of programmer time. Just add a simple pattern match and ignore comments that matched that pattern. Just!
I think that approach is doomed to failure, spammers adapt. We ended up taking a different approach (delinking) and spammers reacted immediately by testing out a series of new attempts. Here’s my favorite:
<p style=”overflow:auto; height: 1px;”>oreilly (http://www.test.com) </p>
A fight using manual pattern matching is just a test of wills. You may as well remove the posts manually.
July 18th, 2005 at 4:06 pm
Andy Lester agrees with me:
Please stop working on content-based spam filtering