Posted by Shawn Burgess on July 27, 2003 at 14:31:01:
In Reply to: Re: two suggestions for version 2.0 posted by David on July 27, 2003 at 06:45:42:
: I'll explore Markov chaining more, but I'm not sure I want to completely disallow repetition. There was an early poem in which the phrase "he he" could be read as "hee hee" and it worked really well. Serendipitous.
: I do plan to release a package at some point; I hadn't considered the cross breeding. I'll have to think about that. I do, however, hope to maintain a central directory of projects so that people can compare the results of different initial conditions.
: --David
:
: : I came over in the slashdot, and I was impressed by how much the poems have improved in the last 24 hours. All hail the power of the 'dot!
: : Two suggestions for version 2.0:
: : Use Markov chaining, aka "Disassociated Press". This was mentioned in the old message board. I'm not enough of a programmer to tell you the details, for a definition google for "Hackers Jargon File". I understanded it enough to know that it will tend to conserve good phrases and cut down on repitition. (the "the the" problem)
: : Secondly, to package the entire shebang so others can put up their own sites, with their own weighted word lists, and to set it up so that the resulting programs can be crossbred across the different sites.
As an opinion, I am a bit against adding rules to make the poems behave more "properly". As David said, serendipidous events could be expunged. I return to the example of e.e. cummings breaking all sense of verse and violating most of the "rules" of poetry. More modern poets draw from his (her?) poems today than the Shakespearian sonnets (for better or worse). Evolution is whimsical and chancy, but that is also the beauty of it. The art here is how you get to the poems, not the poems themselves per se.