I liked his idea, but it seems that he has chosen particularly convenient items to use in his algorithm. And it falls apart for me on the sonnets (the one where he uses individual words from lines--working from memory, so I can't quite describe the process), where he freely rearranges and the output from his algorithm and then invents new contexts for the arrangement.

It seems like a convenient way to permutate (?--the word is in M-W Unabridged to which I do not subscribe and not at all on dictionary.com) the arrangements for compatible pieces. Ultimately, howver, I think given all random samples of data it would churn out poetry like my haiku machine:

Waving daffodils
Feed only on Beccah's mice
Death on stalagmites

A 'permutation' is a rearrangement of the individual parts of a group - like the new order you get when shuffling a deck of cards. To 'permutate' is the verb form of this- 'to shuffle'.

Poor Beccah! Hopefully the daffodils aren't too hungry.

i think mathew's idea is kind of neat. i don't think it's as easily applied to already written literature as he makes it out to be - that is, i doubt all the results of the algorithms would come out so readable as his examples, but it'd be a fun way of putting together story lines/poems/etc.
This reminded me of Mad Libs - but instead of simply limiting it to verbs, nouns and adjectives, you can open it up to sections of a story. i have a feeling that of all the various outcomes, the humorous or nonsensical ones would be the most interesting - as in the haikus and the bot conversations. aw man, now i'm linking everything together... *shakes head*

I agree, the concept is interesting, but I think he makes to big a deal out of the algorithm he proposes. There is nothing particularly unique to his algorithm it is merely a way to take a group of sets and make different orderings from them.

Later on he does mention "Oulipian techniques in general" can achieve similar goals. but in the article's introduction he states the algorithm "is a new means of tracking down this otherness hidden in language". I almost get the feeling like this algorithm is noteworthy, but really any way of switching up the words could do the same.

Not really a big deal, but I thought his insistance on his algorithm was silly.