LOCAL MAXIMUM


if hill climbing [of evolution in the fitness landscape] was indeed the method used to achieve higher fitness, the discovery of the highest ground could not always be assumed. once the population scaled a medium-size peak, it tended to get stuck. this was due to the built-in reluctance of a population to decrease its fitness, which would be necessary in order to search the landscape for an even higher peak. the population would remain f(l)at(ish) and happy on its hill but miss out on the mountains that lay somewhere else on the chart. the population was then stuck on a: local maximum with no incentive to make the giant evolution leap that pushes life toward more complexity.

or, if you prefer to put thinkgs on their head: inverse/imagine/idea the peaks as valleys:

so now the 'sticking'-point is a: trough. now: jiggling around [>vibrating momentum] in a random way, for example, will create a chance to escape from shallow holes and find deeper ones nearby. there must not be too much jiggling, however, or the whole process will cease to work. as we have seen in a variety of connexions, a complex adaptive system functions best in a situation intermediate between order and disorder.


thus, the thinkg that does the trick (to get un'stuck from a local maxmimum) is :

"NOISE"











the pit

thinkg.net