The German poet Trakl uses a vary limited vocabulary and builds short poems out of rather simple sentence types using rather simple forms which are often unrhymed. I developed an elaborate scheme to feed Trakl's oeuvre into a database, order it in terms of frequency of use, inventory the sentence types and how they match up with lines (often one line = one sentence) and then produce randomized Trakl poems.

Someone could go on and further specify vowel harmony, alliteration, etc. At some point you might end up producing a variant of an actual poem.

I'm willing to sell this idea to a German computer nerd for one dollar.

Sort of the reverse of A Classic Ode.

The Araki Yasusada hoax was great, too, but much more limited in its effects because it was all too easy for editors of other pretentious journals to tell themselves that *they* never would've been taken in. This hoax was much more educational because so many people were caught up in it. It raised some interesting philosophical questions, but even betterm it made quite a few folks look very foolish, including the would-be arbiter of online avant-garde poetry, Ron Silliman. And of course those of us whose names *weren't* used had to contend with a strange mixture of jealousy and relief.

I was one of the (many, many) Issue 1 poets, and I was entirely pleased with the project, and surprised at some of the people who got cranky (or in one notable case, litigious!) at it.

Issue 2 collects the various blog posts and comments relating to Issue 1, and might be worth grabbing. (Some of the material in it has since been taken down.)

No worse or better than the 'real' thing, of course. In fact, you'd only have to stick Wallace Stevens' name on one for it to be hailed as a lost masterpiece.

Conrad, we know you don't like any poetry written since the year one thousand, so why bother?