Scott Aaronson: On Two Blog Posts of Jerry Coyne: "David Gelernter... right-wing commentator... argued that recent work has definitively disproved Darwinism as a mechanism for generating new species, and until something better comes along, Intelligent Design is the best available alternative.... Gelernter’s argument falls flat... because it indulges in bad math and computer science.... Gelernter says that (a) a random change to an amino acid sequence will pretty much always make it worse, (b) the probability of finding a useful new such sequence by picking one at random is at most ~1 in 10^77, and (c) there have only been maybe ~10^40 organisms in earth’s history. Since 10^77 >> 10^40, Darwinism is thereby refuted—not in principle, but as an explanation for life on earth. QED. Gelernter can’t personally see how a path could cut through the exponentially large solution space in a polynomial amount of time, so he asserts that it’s impossible. Many of the would-be P≠NP provers who email me every week do the same. But this particular kind of 'argument from incredulity' has an abysmal track record: it would’ve applied equally well, for example, to problems like maximum matching that turned out to have efficient algorithms. This is why, in CS, we demand better evidence of hardness—like completeness results or black-box lower bounds—neither of which seem however to apply to the case at hand. Surely Gelernter understands all this, but had he not, he could’ve learned it from my lecture at the workshop in France!...
#noted
