Lovely spam, wonderful spam

Everybody knows spam. And everybody hates spam. Right?

For a long time, I was largely ignorant of the topic since I simply didn't get any. Spam, I mean. One per week, perhaps. That's just not enough to merit attention.

However, the frequency gradually increased. In 2003, it was about one per day, which motivated me to finally install a spamfilter (namely, Spamassassin). In the following years, the amount of spam increased further. I installed Razor. The spammers increased the frequency. I increased the detection rate.

That was the time when I made a virtue of necessity. I wrote scripts to automate the learning of my spamfilters (plural, since spamassassin was soon complemented upstream by bogofilter, a configuration I much recommend). I also wrote scripts enabling me to analyze the spam I was receiving. These were the times of 100 spams per day, and 99.95% detection rate. A thousand spams, and not even one slipped through. Ha!

And nowadays? I'm back to "one spam a day":

And the detection rate is a mediocre 86%. What happened?

Well, now all of my mail providers have their own spam filters. Shouldn't I be happy thus? Yes, I should. Am I? No. I miss my spam...

At least it's an interesting mathematical problem: having two sequential filters, and knowing the detection rate of one of them, one should be able to calculate the detection rate of the other. If one knows the degree of correlation between the two filters, that is. 😉