Bad CS papers

We talked about it long ago. Somebody finally did it. SCIgen generates papers, and has one accepted at a "conference". I am betting it is possible to beef it up to the point it can generate a paper that can be accepted at a real (but mediocre) systems conference.