I don't buy that explanation, there's clearly a ton more human involvement here than just rearranging verses. The output is far too grammatical and maintains t

I don't buy that explanation, there's clearly a ton more human involvement here than just rearranging verses. The output is far too grammatical and maintains too much thematic consistency across filler word distance to be output by a low-order Markov chain, but is too unique compared to the (very small) corpus to be output by a high-order one.