Ryan's Journal

My filters, plus anything that went through the W3C mail filters and came spam-rated to me, have caught 24,023 pieces of spam so far this year. We upgraded our spam filters, which resulted in 76 false positives (real mail that isn't spam but was still marked as such). The server-side filters missed 975 pieces of spam that Thunderbird recognized; Thunderbird missed 713 pieces of spam that I had to manually mark, which comes out to about 2 per day, or a 2.7% systems failure rate.

Last year: 19,136 pieces of spam server-side catch, 2,049, manual and Thunderbird catches (I wasn't making the distinction then), and 15 false positives.

One reason the false positive rate is so high is my email commenting system on this site was finally discovered by spammers. Up until around July, 2006, I almost never got spam from it; now I get several per day. Because of the way the system is structured, there's not much of use except part of the actual comment body in determining a spam rating. The rest fluctuates between looking like spam and looking like good mail. It may be better to move some of the information into the comment mail headers.

Other reasons are automated mail from frequently spoofed corporations with which I have actual dealings, and Yahoo!, which attaches ads as signatures to all mail, making the spam rating go up.

I'll wait until the year is out before making more graphs.

You must login to leave a comment