Who wouldn’t like their computer to warn them about their language? After all, computers aren’t intrusive or annoying enough yet. Laura from lanceandeskimo.com has noticed that Eudora warns you if your emails (coming and going) are a little spicy.

For one thing, Eudora assigns a “chili pepper rating” from one to three. Actually, two to three. No one has ever gotten one chili pepper. I attempted this feat several times by sending myself emails with various levels of swearing in them, but finally I gave up. On the way, though, I did make a few interesting observations.

They are interesting — and hilarious. (But not work-safe. You are warned.)

But surely it wouldn’t be hard to surpass the performance of Eudora’s classifier. From these examples, it seems to examine a very narrow n-gram window (2 or 3 max), doesn’t attempt word-sense disambiguation, and leaves out some really juicy lexical items. Does Eudora do this poorly on spam detection? This makes me want to trawl for filthy Usenet messages and build a better classifier. Anything for science.

While my mind’s in the gutter, I should mention that ‘Good Reason blog’ scores a slatternly 23.81% on the Slut-o-Meter.