3 ms·
A Googleplex for $1000 by 2040?
The Wikipedia donation page says: "Google might have close to a million servers."
Assuming as Kurzweil suggests that the limits of computing hardware are really far out, i.e., we can expect Moore's "Law" to continue unabated until we hit computoronium, today's Googleplex should be available for $1000 in 2040 (2011 + 1.5 [years per doubling] * log2(1e6)). Somehow it sounds too good to be true.
- nextparadigms 15y agoI don't think it's inconceivable for today's computing performance to seem very trivial 30 years from now. But just like now, when we get there we'll take it for granted that whatever service is available then uses 1 million times more computational power than today's Google. People don't seem to think that much of the fact that a simple smartphone that you keep in your pocket today is more powerful than a computer powering a rocket 50 years ago.
- tluyben2 15y agoNot even a smartphone; your cellphone is a lot more powerful as well.
- ntkachov 15y agoPresumably, with quantum computers this isn't, exactly, far out.
- nostromo 15y agoA Googleplex at any price in 2010 probably seemed too good to be true in 1980, no?
- bfrs 15y agoAs Tyler Cowen points out in "The Great Stagnation", maybe these first 20 doublings or so were the low hanging fruit. Kurzweil keeps saying that people underestimate exponential processes (its also the topic of "The most important video on math" [1]), but I think it cuts both ways. It is by no means clear that the next 20 doublings will proceed at the same rate as the last 20. People might just say, "Siri via the cloud is good enough for me, I don't need Siri running directly on my phone". [1] http://www.youtube.com/watch?v=F-QA2rkpBSY http://www.youtube.com/watch?v=F-QA2rkpBSY
- jacknews 15y agoI think moores law is a given until 2020, according to a couple of articles I've read linked here. That gives us 7nm chips, I guess about 16x the density of today. So your dual core laptop will have 32 cores. I think that might be enough for most people, until we figure out some uses for that kind of power for the everyday person. Moores law is bound to be an S curve, and I think we'll start to get diminishing returns economically, as well as demand dropping off for more power sometime in the 20's.