Despite a popular misconception, Moore is adamant that he did not predict a doubling "every 18 months." Rather, David House, an Intel colleague, had factored in the increasing performance of transistors to conclude that integrated circuits would double in performance every 18 months.
This is what he actually wrote
The complexity for minimum component costs has increased at a rate of roughly a factor of two per year (see graph on next page). Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000.
In simple terms, he observed that the cost of a given circuit halved over time. Since the structure of such circuits obey the laws of physics, which don't change (they are further understood at times, but none have been discarded outright in a very long time), their implementation (generally, TTL) doesn't "improve" per se. But with more transistors available per mm2 to produce such a circuit, that was the easy surrogate for the cost curve. So, Moore's became that. It never was. These days, we've seen that ever smaller nodes have been at escalating cost of R&D and the machines to make such nodes.
A Skylake transistor is around 100 atoms across, and the fewer atoms you have, the harder it becomes to store and manipulate electronic 1s and 0s. Smaller transistors now need trickier designs and extra materials. And as chips get harder to make, fabs get ever more expensive. Handel Jones, the CEO of International Business Strategies, reckons that a fab for state-of-the-art microprocessors now costs around $7 billion.
In all, innovation doesn't necessarily follow the accepted wisdom. These endeavors yapped about the notion of growth for some time, but impelled by Gordon's book, have yapped ever more, and what steps governments can take to produce more of it. And see that it gets to the majority, rather than the 1%. Solow has been mentioned as the standard issue economist who ignores distribution issues in growth. There has been an exception, and from Solow's generation, William Baumol. He just passed away. I read him in grad school, but haven't paid any attention to him in decades. Too bad. I just rechecked my Gordon, and he does mention Baumol on page 173. It was Baumol's later writing that leads to that reference (from the NYT obit):
For example, he said, it takes exactly the same number of people and the same amount of time to play a Beethoven string quartet today as it did in, say, 1817. Yet the musicians who spent years studying and practicing -- and still have to eat and live somewhere while doing that -- cannot be paid the same as their 19th-century counterparts. Their wages, too, will rise, even though they are no more productive than their predecessors were. As a result, their work eventually becomes increasingly expensive compared with more efficiently produced goods.
So, what happens when Moore and Baumol meet? Ever slowing productivity, which means that there's ever less increase to worry about distributing:
"What this says is that the quality of life 30 years from now could deteriorate," Professor Baumol said in 1983, "because many of the services that we associate with quality of life will become relatively more expensive while mass-produced things become cheaper and cheaper."
And it gets worser:
"The real danger is that the nation, mistakenly thinking it must rein in runaway costs, will curtail valuable health services and render them inaccessible for the less affluent. Well-meaning reformers may take the same misstep in education, law enforcement and other handicraft services."
What is striking, to me for sure, is that the notion of service sectors not being on a productivity curve was obvious for some years. Yet I hadn't been aware of Baumol's writings from those years. At least, not consciously. I do remember that "priming the pump" isn't my idea.
The Moore's Law side of things devolves from the observation that much of PC software (the three primary ones being wordprocessing, spreadsheets, and wordprocessing) hasn't led to increases in service sectors' productivity since their original release. The primary reason, which I expect Baumol would agree with, is that such computing doesn't actually maker the user smarter (i.e., productive): I offer up Alt-A loans and London Whales as proof.
His insight about the low productivity growth in services also helped explain why overall growth in an economy increasingly dominated by services can stagnate.
One of the prime notions, again which I think is mine but who really knows, is that FIRE has been the private sectors' way to absorb the college educated in non-productive overhead labor. It may be that Baumol had figured that out too 30 years ago, although perhaps not in such inflammatory words. My bad.
Ya think Orange Julius Caesar, Laffer, or Bannon can figure that out? Don't bet on it. Just dig more coal and make our city air just as bad China cities'. Wonderful.
No comments:
Post a Comment