The inventor’s trial-and-error approach can be automated by software that mimics natural selectionAND
“I HAVE not failed. I have just found 10,000 ways that won’t work.” So said Thomas Edison, the prolific inventor, speaking of his laborious attempts to perfect the incandescent light bulb.AND
As in biology, most mutations are worse than the original. But a few are better, and these are used to create the next generation.AND
A century and a half after Darwin suggested natural selection as the mechanism of evolution, engineers have proved him right once again.
Genetic algorithms (which have been around for decades; Kosa himself has been hyping the stuff in nearly identical articles since the 1980s) are more toys than practical tools, and any results they generate can be found by more practical machine learning algorithms.
ReplyDeleteSure, they're really neat in principle (and I've coded up a few myself just because I thought they were neat), but selection is a horribly inefficient process, and despite what the article says, faster computers don't really solve the problem, as more efficient algorithms would also benefit from the speed increase as well. The beauty of selection is that it generates "solutions" without design; but while it strongly appears that nature has no design, there's no good reason to remove design from algorithms. It's cool to simulate evolution on a computer to help understand biology (systems like Tierra and Avida), but that's different than trying to use this simulation as a practical tool.
Probably the most telling point about the impracticality of genetic algorithms is that the group of programmers that know the most about biology and evolution -- bioinformaticians -- don't for the most part use them in their own work.
OK - but at least the Economist seems to think that Evolution is a reasonable thing. Even if the details are neither novel nor accurate.
ReplyDeleteOK - but at least the Economist seems to think that Evolution is a reasonable thing.
ReplyDeleteFair enough. Although _The Economist_, despite its reputation as being right-wing due to its "the free market will solve everything" viewpoint, has always traditionally been a supporter of evolution and science in general, so it isn't all that surprising.
I like the economist quite a bit and do not view them as a right wing mag. They are "conservative" the the original meaning of the word. And yes they do tend to support science. and occasionally write nice things about evolution as a science. But it is still rare for any national magazine to write articles in support of evolution and natural selection without some inane comment from intelligent design supporters. Here the Economist has taken the correct approach - there is no need to quote from anti-evolution folks.
ReplyDelete