Limits of Technology?

It's a shame that Jonathan Huebner isn't more famous. If he were, we could place him alongside Thomas J. Watson, the head of IBM who famously predicted a world market of maybe five computers, and eminent scientist Lord Kelvin, who stated that heavier-than-air flying machines are impossible.

Huebner, a physicist, thinks he's spotted a slowdown in innovation. That is somewhat interesting, and John Smart at Accelerating Times wrote quite a detailed review of the idea. Smart suggests that perhaps technical innovation is not declining -- he identifies some methodological problems. Or perhaps it is declining because technology is capable of supplying the basic human needs, and we've become too comfortable to be motivated to invent at the same breakneck pace. Or, perhaps innovation is moving away from "hard" technologies into more psychological and sociological realms. I'd suggest that in the past few decades, massive innovation has occurred in computer software, but most of it is "under the hood" and we only see the user interface.

But what I find ridiculous is Huebner's theory, as quoted in a New Scientist article about his work: "Perhaps there is a limit to what technology can achieve."

A few years ago, I attended a talk by a physicist who spoke about the physical limits of computation. He derived some astronomically vast number of operations per cubic centimeter per second. Then someone in the audience announced that Moore's Law predicted that we'd achieve that in five hundred years. Perhaps that represents a real limit to what technology can achieve. But with the computer invented a mere half-century ago, we're only one-tenth of the way there.

Molecular manufacturing relies on existing technological theories, and doesn't even begin to strain today's fundamental scientific theories. But even such a mundane technology will be able to build computers at least a billion times as efficient and powerful as today's. (Actually, even nanoscale technologies will probably be able to do that, eventually.) Molecular manufacturing will be able to build motors at least a million times as powerful, and materials a hundred times as strong, and integrate these powers into complete products at all scales. And it will enable automated general-purpose manufacturing, a technology that does not exist in the world today, and that will do for manufacturing what computers have done for information processing. (What, you mean information processing isn't synonymous with computers? Think about it... there used to be other ways to process information, but now computers are used almost universally.)

All I can say is that anyone who thinks technology is nearing its limits must not have thought through the possibilities. It's an attractive idea, if you're afraid of change -- as many people are. But it just doesn't fit the facts.

Chris

Comments

[michael vassar]

Actually, although Huber's speculations about the limits to possibility are grossly wrong, his empirical observations are essentially inevitable if one uses his research methodology. I am aware of four other people who have analyzed the rate of technological change using the same methodology and all have reached strongly convergent conclusions. Drexler's preferred metric of charting exponential trends ignores the fact that the same trend can be seen as exponential, linear, or diminishing depending on the exact definitions of the metrics used, as was convincingly argued at the 2004 WTA meeting by Phillip Goetz.
Different people have different preferred methodologies, but it is clear to me that the problem preconceptions based on "folk theory of progress" considerations are the dominant determinant of Smart's and Kurzweil's conclusions. Hanson provides a third class of technological conclusion by the way, not a sub-set of Kurzweil's, and his conclusion also has many independent backers. It is also worth noting that Kurzweil isn't careful with his logic. In the same paragraph he glibly states that the rate of technological progress doubles every decade (20 years?), and that the progress over the next 10 years there will be as much progress as over the last 20, while the logical implication of doubling rates every decade is that Every decade observes more progress than all of previous history, an utterly laughable conclusion.

[michael vassar]

per capita GDP % growth provides another proxy for tech progress, while Nobel prizes are an important data set. The number per year is fixed, so there's no metric, but "accelerating change" implies that they should be for increasingly important discoveries. For instance, a 4-fold increase in speed of change implies that the average year after the increase awards a discovery as important as that associated with the best 1 year in 4 before the increase. Casual observations of scientifically knowledgable people unbiased by the desire to confirm theories could be checked to test this if necessary. This test also checks the "low hanging fruit hypothesis" as the depletion of low hanging fruit would suggest that the average discovery after a 4-fold depletion would require as much cleverness as the best discovery in a four year period before that depletion.

[Chris Phoenix, CRN]

What I was complaining about is not that he says innovation is slowing down, but that he says it could be because we're reaching the limits of technology.

If innovation is in fact slowing down, and not just moving to another area where we'll only notice it in hindsight (software!), there are several possible reasons:
1) we've gotten too complacent and comfortable.
2) we're focusing too much on driving consumerism rather than advancing technology. Too much focus on advertising and user interface.
3) It's just a temporary gap, while we re-group to tackle complexity and the nanoscale (which are not the same thing).

Chris

[Chris Phoenix, CRN]

Michael, what I was complaining about was the theory that we're reaching the limits of technology.

If innovation is in fact slowing, there are several possible reasons:
1) We're too comfortable.
2) We're spending too much effort on consumerism.
3) We're just regrouping to tackle complexity and the nanoscale (which are not the same thing).

Chris