Yes, this is funny. The existence of random paper generators for a discipline doesn't say much about the rigor of that discipline. You might be able to make some kind of argument based on how long it takes for someone trained in the discipline to detect it as randomly generated.
Of course, you could imagine building a program that creates perfectly compileable, yet randomly generated programs. These guys actually did it to test gcc's handling of the volatile keyword: http://www.cs.utah.edu/~regehr/papers/emsoft08-preprint.pdf
This is fantastic. And there are some cool stories and links on that page. Thanks for the link :D
http://pdos.csail.mit.edu/scigen/
..much more funny and even pdf direct output :D