Defensio Update

Well, here are the stats for another week of Defensio (in conjunction with the Project Honey Pot a.k.a. the HTTP:bl addon) for my photoblog.

  • Recent accuracy: 100.00%
  • 523 spam
  • 1 legitimate comments
  • 0 false negatives (undetected spam)
  • 0 false positives (legitimate comments identified as spam)

Funny thing is I got 6 comments which were not rated by Defensio. Both parameters for each comment were empty (e.g. NULL). I’m still figuring out how this has happened. x-(

In the mean time I’m resetting my SPAM flood protection, the token support and the maximum # of URI allowed in comments because 523 spam comments a week is ridiculous (That is 75 comments a day :-o ).

In case you were wondering: the Project Honey Pot a.k.a. the HTTP:bl addon blocked nearly 230 comments this week. x-(

Defensio, the outcome

As mentioned earlier I wrote an addon for Pixelpost which uses Defensio for blocking spam. It’s been about a week and time to evaluate how it has worked.

Well, without further ado here are the stats for my photoblog for the past week (please remember I disabled all other anti-spam measures):

  • Recent accuracy: 100.00%
  • 173 spam
  • 2 legitimate comments
  • 0 false negatives (undetected spam)
  • 0 false positives (legitimate comments identified as spam)

Quite impressive, there was only one comment placed in the moderation queue and you might consider this as a false negative (undetected spam). The comment in question:

Thanks for the post.
Great info.

You can imagine why Defensio didn’t flag it. All in all I’m quite satisfied with the performance (and indirectly with my coding ;-) ). The next step is to evaluate how it works combined with the Project Honey Pot a.k.a. the HTTP:bl addon.

Based on data collected from honeypots (spamtraps) set on sites you can block certain ip-addresses to access your site. So basically, if a known spammer is trying to access your page, they are directed to the honeypot. This also reduces spam and the best part is: the spam never makes it in the database.

Now, if you all excuse me, I have a quarantine I need to empty. :-)

Defensio; outsmarting evil spam

Recently I received some information about a new way to fight spam. Some guys from Canada really hate spam and developed a new anti-spam routine called Defensio.

This new routine is more accurate then anything known at this point (*cough*Akismet*cough*) and actually makes managing spam comments a breeze. This project is still in beta right now, but is scheduled to go live pretty soon. The last few days I’ve been working on a Pixelpost addon that uses Defensio. It wasn’t easy, we had to upgrade the look of the comments section in Pixelpost to make it work but I finally managed. And today is the day I put it to the real test.

I disabled all other anti-spam measurements on my photoblog and rely solely on Defensio (and my coding skills). Hopefully Defensio does a better job then Akismet.

Let me provide you with some stats for the last 275 days on my blog. Around 98% of all the comments where spam and these were all blocked by my defensive lines. My defensive lines consists of several things: accepting no urls in comments, using the token and the Project Honey Pot addon (PP1.7 only)

Additionally, 64% of the 2% that got through my defensive lines were also spam which I had to deleted manually. My best lines of defense are currently the Token support and the Project Honey Pot addon. But still these things will let spam through and hopefully Defensio will do a better job.

In a couple of days I’ll post my results here and in the mean time I’ll just keep my fingers crossed ;-)