Posted by Jeroen on September 29, 2003 at 07:10:58:
In Reply to: Re: metagenes posted by David on September 26, 2003 at 23:54:39:
I agree. I also can see David's problems with coding lexical data. That's why most of Markh's pseudowords are nice: independent of the kind of words that follow.
I'd like a coding for 'repeat the next word as the first word of the next line', or the line after that. (I'm a sucker for repetition, as you might have guessed by now.)
: This is a really interesting idea. The line breaks are already handled this way, but I like the idea of embedding more meta-data in the genome. I had been toying with the idea of evolving structure separately from the actual words, but I think interleaving them (like real DNA) makes more sense. Of course, part of the implementation would require lexical data to be included with the words. Right now the words are merely text strings.
: --David
:
: : One of the things "real" genomes have is genes that control the expression of other genes. This makes it possible for small changes in the genotype to have large effects on the phenotype.
: : Maybe something similar can be done.
: : I was thinking of two "psuedowords" (hideline) and (hideword). When the (hideword) gene appears the next gene is not expressed. When the (hideline) gene appears, no genes up to the next(newline) is expressed.
: : Other possibilities include (swapwords) (swap the next two words), (swaplines), (nounonly) (the next word gets expressed only if it is a noun), etc.