Spammers sure seem to have figured out how to defeat Bayesian Analysis in the last few months. We've flushed and re-created our Bayesian database a couple of times recently, and I'm beginning to think that it's not worth it. So, as of about 4:45 PM today, we've adjusted our spam scoring and disabled Bayesian. It will be interesting to see what this does for (or to) us.
How are you doing with all of the image-based spam these days?
Posted by: Matt Singley | November 28, 2006 at 06:34 PM
Matt, I've just posted an item on my barracuda blog about our attempt to deal with the image spam problem. It's not perfect, but it's a step. http://tonydye.typepad.com/barracuda/2006/11/stock_image_spa.html
Posted by: Tony Dye | November 28, 2006 at 08:53 PM