BTW, Kurzweil suggests that universe (at least our part of it) has been increasing in intelligence at an exponential rate and that machines are just an extension of that.
Moore's law was suggested by Gordon Moore, one of the founders of Intel. It was an empirical observation that the speed of the processors doubled approximately every 18 months. That's the way it's usually stated - what it actually stated was that the # of transistors could double because the gates would be made smaller and smaller.
Check out http://en.wikipedia.org/wiki/Moore%27s_law
Kurzweil (Age of Spiritual Machines) suggests that this exponential growth is actually an extension back into the time life first arose. Big Leap, of course, but he seems to have made a career of Big Leaps.
http://en.wikipedia.org/wiki/Ray_Kurzweil Moore's Law has a theoretical limit that is fast approaching - our existing computers will only ever go so fast. But there are two things we're doing to improve this - the first is distributed computing (like the internet) - this enables collaboration, but also more specific resource sharing. The other front is that we're considering other kinds of computation - including, chemical and biological computation, as well as quantum computing. These are not digital computing systems, though, they're analog. I'm not sure how the chembio stuff is going, but the most interesting possibility of the QC stuff is that many believe it will be able to solve NP-complete problems in polynomial time.
http://en.wikipedia.org/wiki/NP-complete As has been pointed out, though, wide-spread QC is not a sure thing - primarily because of the problem of decoherence.
People have actually built QCs, though:
http://www.findarticles.com/p/articles/mi_m1511/is_1_20/ai_53501821 They're trivial and useless, but the first steps in digital computing were a long time coming as well.
