Several people have noticed an uptick in spam that includes strings of words unrelated to the topic of the spam or to one another, apparently in an attempt to fool Bayesian spam filters into thinking they’re legitimate. (Popfile is nonetheless capturing most of them.)

I don’t have any idea what program these randospamos are using, but there are plenty of generators that go opposite the way, putting together words that (based on a particular corpus) are likely to go together. See Fun with Markov Chains where you’ll find Alice in Elsinore, gibberish generated by intersecting Hamlet and Alice in Wonderland.

Categories: Uncategorized dw