|
shallowdeep -> RE: Making the world a billion times better (4/13/2008 10:14:49 PM)
|
It's nice to see someone optimistic about the near(ish) future. I'm in agreement that technology can solve many of the problems facing us, and I'm optimistic that it will. With that said (and leaving the subjective billion times "better" title aside), some of Kurzweil's comments left me scratching my head.
Take his comments on nanotechnology, apparently in relation to photovoltaics:
quote:
Nanotechnology itself is an information technology and therefore subject to what I call the "law of accelerating returns," a continual doubling of capability about every year.
His extrapolation of Moore's Law to PV doesn't make sense. The closest analogy to processor performance in PV is efficiency. Unlike computational power, efficiency has a very clearly defined upper limit: 100%. Considering that we're already around 40% on the cutting edge, there isn't much room for exponential growth in efficiency. Comparing PV price/efficiency to information technology price/performance similarly doesn't make sense. What's often ignored about integrated circuits is that while the transistor density has increased exponentially, so has the cost of the plants that fabricate them. Since process shrinks allow for more transistors per wafer, price/performance has steadily decreased, but the same argument doesn't hold true for PV where the maximum power you can generate is directly proportional to the surface area of the material.
If he's simply referring to manufacturing capacity growing exponentially, that's possible... although without continuous exponential population growth it's not necessary (or likely) in the long term. Comparison with more traditional manufacturing fields would give a better idea of the potential cost reductions from scale than a flawed association with information technology.
Maybe he only meant the utility of nanotechnology in a general sense will increase exponentially, which is more vague and, therefore, defensible. =) But the context is misleading, as few specific applications (and not PV) are going to be following an exponential curve in a measurable attribute.
And then there's this:
quote:
We are now adding three months every year to human life expectancy, but given the exponential growth of our ability to reprogram biology, this will soon go into high gear. According to my models, 15 years from now we'll be adding more than a year each year to our remaining life expectancy.
While I believe biology is going to be one of the more interesting fields in the next few decades, I can't help but see his model as more a product of his desire for immortality than of facts.
A bit on the computer issue:
His assertion that computers will continue their exponential increase in performance over the next 25 years is possible, but not with the relatively straightforward process shrinks that have driven Moore's Law in the past, which he acknowledges in the article linked by Alumbrado. He has faith in innovations beyond planar MOSFET technology, and uses historical precedent to support that. In the long term I wouldn't bet against him on this, but in the short term there might be a noticeable hiccup.
We haven't yet reached the limits of venerable MOSFET technology, but we're getting there. Commercial manufacturing of processors recently reached the 45 nm node. The next nodes, the 32 and 22 nm ones, are likely to follow with relative ease, and 16 nm is also likely. Much beyond that, things start to get dicey. That means the current rate of progress is likely sustainable for another 8 years, maybe longer, before any really novel approaches or breakthroughs are needed.
This article has an approachable explanation of some of the physical limits facing the technology. A big issue is that, beyond the 16 nm node, quantum tunneling through the gate when the transistor is off starts to reach the point where leakage current becomes indistinguishable from a logic 1 at reasonable voltages. The International Technology Roadmap for Semiconductors has a more technical look at the challenges facing the industry.
|
|
|
|
