Comment Spam Stats

Since January 12th:

  • Valid comments accepted by Akismet: 36
  • Spam comments accepted by Akismet: 17
  • Spam comments rejected by Akismet: 814

I don’t have a number for false positives, but given that I’ve received zero email complaints I’ll assume the number is low if not zero. This gives Akismet about a 98% success rate on catching spam, which is pretty good. It makes my life better. Having more spam comments than real comments get through the gates can be really depressing for a blog owner.

At some point I’ll post my Django newforms/Akismet integration code. It’s very simple, and clearly worth the effort.

Update: James Bennett reminds me that his comment_utils wrap up anti-spam and comment moderation measures in one tidy package.

Alex Ezell commented :

I like to say that comment spam is the reason I always quit blogging, but I suspect it’s because no one wants to read what I write about :)

I guess one of these days, I’ll get around to redoing my blog with Django.

Does anyone else think that Akismet should be doing better? For a distributed spam system, I would expect 99% or better catching spam. I say that based on my, perhaps incorrect, belief that comment spammers usually post the same junk on tons of sites all in a short period of time.

After the first 2 or 3 bloggers mark this stuff, shouldn’t Akismet catch the rest? I probably just don’t know enough about how it happens.

Paul commented :

I suspect that there’s a significant random component to the spam, so that it’s not exactly the same junk everywhere. But given that good Bayesian email filters seem to be able to do better, I do agree that there seems to be room for improvement.

Nonetheless, I can’t begrudge the free service that Akismet is giving me. As we’ve both experienced, comment spam is demoralizing, and it’s nice to be (mostly) free of it.