[pmetzger]

In addition to Sokal's hack on "Social Text", I believe that Rob Pike did a test in which random texts produced by running a Markov chain over Derrida were found to be indistinguishable by informed readers from actual Derrida. (I'm probably misremembering at least some details of the actual incident but the gist is, I'm pretty sure, true.)

[pg]

I did the same thing with Paradise Lost to produce the "poem" at the start of ANSI Common Lisp. When read aloud it passed as poetry to a sample of 1 English Lit grad student.

[dfranke]

Another of life's mysteries solved. I've been wondering about that poem every time I open that book.

[pg]

It's explained in section 8.8. The source code is there.

[pmetzger]

I've done a bit of research -- the Pike incident involved Baudrillard, not Derrida.

[kcl]

All that proves is that Baudrillard's message is invariant under Markov chains. I'd like see a mathematician produce an argument of such fine structure.

[jey]

In other words, contains 0 bits of information?

[sketerpot]

Pike's comment at the beginning of the Markovified text is brilliant:

> For those unfamiliar with M. Baudrillard's work, the thesis of the essay is, in a nutshell, that reality is indistinguishable from good simulation.

[dfranke]

That is hilarious. I plan to repeat this experiment.