Bill Brown bio photo

Bill Brown

A complicated man.

Twitter Github

For you Windows users out there (hi, Dad…I know! My own father! Luckily, the apple did fall far from the tree—no pun intended), you might want to check out SpamBayes, a free, open-source Outlook plug-in that catches spam. It requires Python, but there’s an installer that takes care of everything for you.

What does it do? Basically, it uses a technique called Bayesian analysis to establish a pattern of what you consider spam and what you don’t consider spam. After installing the program, you select a bunch of emails that you know are not spam (which the program weirdly calls “ham”) and then you select a bunch of spam. Then, whenever email comes in, you tell it which ones are spam and it gradually learns to distinguish the good from the bad. After its training period, you can either have all of the spam automatically moved to a separate folder or deleted—it’s your call.

Jon Udell, a columnist for InfoWorld, swears by it and I’ve generally found him a credible and reliable source. Arnold Kling, a Corante blogger, has discussed Bayesian filtration numerous times.

I’ve been using Bayesian analysis for some time now and I’ve been mostly happy with it. I don’t even check my junk folder for false positives anymore because I hardly ever found anything in there of value. I still get a couple of spams every now and then, but it’s mostly a case of not having any text in the message. They’re easily found and deleted, of course.

[UPDATE (8/11/03): Slashdot links to an interesting review on different implementations of Bayesian filtration in the Windows world. SpamBayes comes out the winner.]