I agree, there seems very little merit in this argument, whether this is from Kurzweil or someone else. Suppose someone with intelligence x can build something else with the intelligence f(x). First, the argument seems to assume that f is increasing. This is reasonable, although I think this is hardly a golden rule. Success doesn't seem to be perfectly correlated with intelligence. Next, in order for intelligence explosion to occur we must know f is an unbounded function. I see no reason to believe that. Quite the opposite, the intuition and observation suggests that f is rather asymptotic. Perhaps it even stabilizes at a certain level. In order to do great science one only has to be smart enough, and then one has to also be hard-working, stubborn, lucky, persistent, passionate, and a hundred other different things that have nothing to do with intelligence.