There are eighteen times more people involved in transistor-related research today than in 1971. So if in 1971 it took 1000 scientists to increase transistor density 35% per year, today it takes 18,000 scientists to do the same task. So apparently the average transistor scientist is eighteen times less productive today than fifty years ago. That should be surprising and scary.
He is citing Bloom, Jones, Reenen & Webb (2018). This paper was discussed at a conference Alexander attended. He writes,
constant growth rates in response to exponentially increasing inputs is the null hypothesis. If it wasn’t, we should be expecting 50% year-on-year GDP growth, easily-discovered-immortality, and the like. Nobody expected that before reading BJRW, so we shouldn’t be surprised when BJRW provide a data-driven model showing it isn’t happening. I realize this in itself isn’t an explanation; it doesn’t tell us why researchers can’t maintain a constant level of output as measured in discoveries. It sounds a little like “God wouldn’t design the universe that way”
My favorite economics professor, Bernie Saffran, was wont to observe that learning takes calendar time as well as studying time. A student cannot master a concept merely by putting in a certain amount of hours studying it. It takes some amount of days or weeks or months for a concept to sink in. You could write L = f(T,t) where L is learning, T is the amount of time you spend studying, and t is the passage of calendar time. Throwing more T at a subject brings diminishing returns, unless you also increase t. We can speculate that some of the brain rewiring that takes place is unconscious, and you cannot artificially speed up this process.
Suppose that there is an analogous factor at work at the level of society. That is, scientific discovery depends on calendar time as well as the time that scientists spend working on a problem. It takes a while for X to sink in, and only after X has sunk in can we go on and discover Y.
Alexander sees no reason to expect that we can speed up scientific progress with simple policy changes or institutional tweaks. I am inclined to agree.
But having said that, I can think of institutional habits that may be holding progress back. I probably will write an essay on those. UPDATE: The essay offers two modest reforms.
I wonder if part of the story here is related to the decline in labor force participation among prime age males as well. Researchers like video games, too, and probably especially transistor researchers.
I suspect the trouble is simply that many fields encounter hard natural limits that require lots more effort, cleverness, and ingenuity to make increasingly marginal gains, and that often-times achieving even feasible improvement is uneconomical. Measuring in terms of productivity in terms of the transformation into outputs is of course important to know in terms of improvement to capabilities, wealth, and welfare, but also somewhat misleading and unfair.
For example supersonic flight has been possible for generations and was operated on a (heavily subsidized) commercial basis over 40 years ago. But the nature of aerodynamics makes loud booms very hard to soften and requires huge amounts of fuel and power (takes over 7 times more fuel per passenger mile for a transatlantic flight, for just a few hours benefit, which anyway is small in comparison to the total door-to-door time from the lengthy modern air-travel process.)
It looks like we’d be lucky to get commercially viable nuclear fusion power plants even an entire century after the first fission plants, and that despite the enormous incentives to do and during the most technologically sosphisticated era in human history. Nature has made doing it really, really difficult.
Both the moonshot and independent study reforms are excellent. They need to be picked up by foundations and championed by a strong leader with deep pockets. Maybe Tyler Cowen would fund such an effort.
I have trouble taking the idea of a science slowdown seriously. I am still awestruck and dumbfounded by all of the progress around us. Amazon can deliver a nice, full size mattress to your door in a medium size box. Elon Musk can fire rockets into space and land them back on earth, he can build tunnels at a fraction of the price it takes governments. Kid’s have robotics competitions for cris’sake. My kindle has more volumes of classic literature than are on the shelves of the country library. My older house is insulated now to an extent that my powerbills are less than half of what it was when I bought it. I could go on an on but I would start to sound like a cheap Don Boudreaux impersonator and he really is the master of this perspective. My detached retinas would have left me blind 40 years ago. I’ve got a combined defibrillator and pacemaker implanted in device smaller than a pack of cards. I don’t know how one can not help but feel overwhelmed by the tidal wave of progress we have experienced.
If this is a slowdown…..
I think maybe we ought to imagine the scene in Moonstruck with Cher slapping Nicholas Cage. “Snap out of it!” Manage your expectations people, and look around you.
FWIW – chewing some more on the “Curious Students and Helpful Mentors” modest proposal, it would appear that this philosophy is embedded in the Summit Learning Plan. They have a nice website and their 2017-2018 annual report provides useful objective metrics that the idea of combing self-directed learning with enhanced mentoring is both efficacious and scalable.
The estimable Joanne Jacobs did a nice post July 23, 2018 on EdSource entitled “Charter school network spreads ‘personalized learning’ model nationwide.” Jacobs writes:
“Students master academic content by choosing from “playlists” made up of a range of learning tools including free online instructional videos from the Khan Academy, animated videos from BrainPOP, guided practice problems and interactive exercises, websites and texts. Students take tests when they feel ready, moving on only when they’ve achieved mastery of the content.”
This is combined with structured and frequent one-on-one mentoring, “In addition, teachers work one-on-one with students, helping them set short-term and long-term goals and develop “habits of success,” such as self-management, responsible decision-making and persistence.”
Too early to tell on null-hypothesis rejection but worth watching.
200 hundred years we go from wooden rotary presses to high speed linux web serves. The AI bots take us the rest. In the next step, it becomes mostly apparent, as it happens, that a new trend is effecting us. I see it in software development, the ability to compile code into complete, and portable interface definitions, automatically. Collectively, these bots find the patterns in our desires and interests, groups of collaborators use them to manage the semantics of their collective project. Bots operate underneath, reading our stuff, finding the patterns, checking our balance sheets, buying our goods, running our transportation.
Because of the factor of 1,000,000x bottlenecks in communication the effort is not additive. Compare the bandwidth in the synapses-neurons brain network to physical interpersonal communications to distance communications via published research.
100 athletes don’t run a marathon 10x faster than 10 athletes.
I used to be in electro-optics research, and I’m pretty sure it’s just the law of diminishing returns. If you take a traditional process and apply a bit of engineering to it, you may go from 20% efficiency to 60% efficiency, and that’s a huge payoff economically. But when engineers have been optimizing a process for decades, it takes a great deal of effort to go from 90% to 91%, and the payoff is minimal. Adding features works the same way; after a while it already has all of the cheap useful features and you’re tinkering with those buttons on the remote that nobody uses.
Going from 1 mile per gallon to 2 miles per gallon means you go from 1,000 gallons to 500 gallons every thousand miles a savings of 500 gallons. Going from there to 100 miles per gallon saves you 490 gallons. You can never save as much as that first improvement.
Hmmm… Back about 1970, the US spent about 2.75% of GNP on research and development, most of it coming from the federal government. The government share began shrinking about that time, but industry picked up much of the slack, maybe spending somewhat more on short term development than blue sky research. Anyhow, near 50 years later, the US is still spending about 2.75% of GNP on R&D.
Population’s doubled, roughly, since 1970, so we might think that if 1000 researchers were working on shrinking transistors then, we’d have 2000 working on the problem now. But instead we have 18.000 and we aren’t happy about their progress.
So where’d those extra 16,000 researchers come from? Well, space programs got cut back since 1970 (we used to have people walking on the Moon. Imagine that!). Oceanography got cut back — there’s no more foolish talk about undersea mines, is there?). I’ll bet we cut back some sorts of medical research (why spend taxpayer dollars on yaws and beri-beri when you’ve got Bill Gates around?). Some other possible R&D programs maybe got a little talk once upon a time but never really moved into the big time (do a Google search on “nanotechnology”).
And on and on and on. And it strikes me that. rather than viewing this as a shortfall or some sort of problem, that modern economists are basically quite content with things. After all, they still have Moore’s Law to point to. Adam Smith and Joseph Schumpeter and Robert Solow and Ayn Rand got it all right!
Not taking away from anything that has been said before, and which reflects the breadth and complexity of the general issue…I would point out that the teaser paragraph’s “claim” of 1,000 scientists versus 18,000 scientists –> same 35% reduction equals 1/18th productivity obviously reduces all progress/effort to increasing chip density. Even if one assumes that all this staff were actually scientists working on science (not likely), a fair bit of their effort could have been aimed at making chips more cheaply, more safely, more sustainably, better documented (among other, not trivial attributes of the product)…and then you get into physical and man-made inefficiencies being discussed.
Did the authors take the time to verify their model?
Sigh. The answer appears to be no.
“That should be surprising and scary.”
What should be surprising and scary is that anyone would think that after the early rush in a new technology that advancements wouldn’t slow as limits start being reached. Similarly, with why productivity increases might go flat as early automation settles out and workers shift to service-jobs where productivity has hard limits of an individual’s time. Or the 20th century run up in the uses and efficiencies in the new power source, oil, might slow.
As the article deals with transistors, let’s consider that just next year will it be 200 years since the discovery that runs our world was made, i.e., the linkage of electricity and magnetism. This discovery allows electrical power generation from machines, electric motors, all radio transmissions, spark generation in ICEs, etc. Seventy years since the viable transistor was discovered. A century since the vacuum tube. Fifty years, 1968, when the Programmable Logic Controller (PLC) was invented making factory automation cheaper, quicker and more reliable.
For the PLC, the next advancement will never be as great as moving from relay logic, with miles of relay cabinets to a small controller, that these days can have its program uploaded remotely after extensive pre-deployment testing and simulation.
The late 19th and early 20th centuries were a human history anomaly. After increases in individual liberty starting in the 17th century, technological discoveries coalesced and their exploitation/implementation advanced apace. Disease theory resulted in far less mortality prompting a shift in population control to fertility management. Technology freed people from drudgery and facilitated even more crowding into cities.
But to think that the economic, productivity, population, etc. growth of the mid-20th century when technology was advancing but so was State efforts to regain control, is the height of folly. Since the 1970s in the West and increasingly elsewhere, government has come to impeded advancement just as technology is in the long tail of discoveries.