A comment here the other day mentioned Scigen, which I hadn’t seen before. Some folks at MIT have whipped up a bit of code and a database of computer science topics, phrases, and graphs, and developed a quick paper generator. The paper will make no sense at all, of course, but it is quick. And what they’ve found is that making no sense isn’t as much of a handicap as you might think when it comes to some conferences and some journals.
Scigen papers have been accepted for presentation at some of the less prestigious meetings, and have been sent to various cheesy journals, which have cheerfully “reviewed” them once details of payment were cleared up. This is not a good sign for your field when total gibberish can be passed off like this, although one assumes that it says more about the sorts of conferences and journals that are accepting these things.
And yes, a comparison to the Sokal hoax comes to mind immediately. That one was even more damning, though, because the gibberish paper that Sokal came up with wasn’t sent to some sleazy fee-generating publication mill, but to what was considered one of the better journals in the field (Social Text). Who (famously) published it anyway. The editors later backtracked by saying that they thought the paper, you know, lacked originality, that it wasn’t well written, that they (ahem!) just accepted it as a favor to a physicist visiting their rigorous area of study, and so on – but the fact remained (and remains) that an editor should be able to distinguish a valid paper from a sticky pile of superglued nonsense.
The reason the Scigen papers aren’t picked up on, clearly, is that no one’s looking at them, at least no one with any knowledge of computer science. The editors and organizers who let them through are interested in collecting the registration and editorial fees first, and after that, well, that’s not really their department. A perfectly analogous example is the utterly crazed “Atlanta Nights” manuscript, whipped up by a loose team of authors to expose the “editing process” of a pay-to-publish operation (PubilshAmerica) for what it really was. The book is a bit hard to follow. Characters change names and/or genders, die and come back to life, and find themselves doing ridiculous things in impossible tangles of verb tenses. But hey, the manuscript was supposedly read through, and accepted without one solecism out of place. If the credit card number is valid, so’s the syntax. (Don't want to take my word for it? Here it is, under the byline "Travis Tea", published by a print-on-demand house after PublishAmerica hastily backtracked.)
No one’s tried (as far as I know) to submit a Scigen paper to a reputable comp-sci venue. I assume (and very much hope) that it would be sent back with a puzzled note attached. The same goes for the chemical literature, or at least it had better. A chemistry-focused version of Scigen would be an interesting experiment, but I think I know what the likely results would be. There are bottom-tier journals and conferences in every field. They’ll bite. As long as that check clears.