sha_metamorphosis

in my first reading of the warner piece, i read the comparison of the jellyfish as a metamorphosis of Medusa, something transfigured physically but remaining true to its essence. what is interesting, then, is that the jellyfish is so insubstantial it becomes a sort of "emanation" of the seas "currents and eddies". in this sense it is almost so essentialized (metamorphosized?) that it is not even conscious -- it is not intentionally poisonous or murderous, it just is poison. the 'unknowability' of the jellyfish is partially its seductiveness, but also its danger.

frazer is excited by this 'resilient strangeness' as a way to bring architecture into new territory, by using computing power as an opportunity to iterate and generate thousands of varieties, morphing buildings to hopefully find an unexpected optimization. there are dynamic systems that have been cropping up -- the unseen video, a dynamic flash video that changes based on your weather and location, is one that pops into my mind -- but frazer seems to be excited specifically about the chance of a better among the dynamic mutations. it reminds me a lot of the idea of cultivating crops, pairing or attempting to isolate a mutation.

my housemate actually found himself in a similar situation last semester. the assignment was to create a pacman game, and code the ai for the pacman. his pacman was driven by a set of priorities that changed based upon the current situation (i have to eat a dot, i have to eat a capsule, i have to avoid a ghost, i have to eat a ghost because i just ate a capsule, i have to turn as little as possible, etc...). frustrated by his intuitive attempts to place priority (avoiding ghosts is a priority highest, eating dots is high, eating capsules is incidental), he wrote a program to evolve his pacman -- gradually changing each parameter, brute force, and logging which pacman did the best. eventually his pacman was extremely streamlined, and its priorities were completely unexpected (turns out his pacman became a hunter with almost absurdly risky behavior, constantly letting ghosts get close, then finding a capsule and eating all the ghosts). what he found, though, to his disappointment, was that his pacman became unintentionally site specific. his pacman had become so optimized for the test map, that when a random map was loaded, his pacman would do poorly.

it is precisely that specificity that would make frazer's idea of cultivating architecture so interesting. it would be very exciting to test and generate a building -- iterating through a 'large number of evolutionary steps', then sifting through the results for unexpected, emergent forms.

(though it seems like you could essentially pass a building through a series of wind, sun, noise, etc... simulators and come up with variations that optimize or maximize building efficiency, i'm not sure if frazer is completely content with using building parameters as metrics for evolutionary vigor. he first describes computers as tireless slaves, then later as an electronic muse -- but he holds that the initial creative spark is still our own. where does the design stop and the algorithm take over? is the design process simply choosing the right parameters and metrics?)

( www.theunseenvideo.com )