There's an angle to this that I see on the ground as a graduate student in neuroscience that I've never seen anyone talk about, including Greenspun. It's that there are way more training positions for graduate students than there are jobs waiting for them.
Some back-of-the-envelope calculations reveal the problem.
In 2001, there were 57,639 grad students in the biological sciences [1].
If they want to labs of their own, how stiff is the competition?
9100/98,700 scientists in the "Educational Services" sector are life scientists (9.2%) [2]; 46,300/98,700 scientists in this sector are in the "Colleges, universities, and professional" category (46.9%) [3]; so, 0.092 * .469 = 4.3% of 98,700 = 4258 scientists are life scientists in the "College, university, or professional" category.
Let's say that the average graduate student takes 5.5 years to graduate; that means that there are 10,480 newly minted Ph.D.s in biology each year. Let's say that the average PI has a lab over a 30 year period; that means that there are ~142 jobs a year in biology that open up for those 10,000 graduates.
This means that 1/73 of the newly minted Ph.D.s should reasonably go on to start a lab of their own. The actual number will be higher, but they will be winnowed out once they fail to get grants at some point early in their careers.
I don't know how real these numbers are, but they are in agreement with the sense of struggle I see day-to-day.
The government should fund fewer training slots at graduate schools, and more R01s (the grants that keep independent labs running), to get this ratio in better balance. Right now, it is a pyramid scheme.
It also should come as no surprise that domestic students would stay away from grad school. It's an economically irrational pursuit when other options are open. On the other hand, this is the best country in the world in which to train; students can go home and be big fish in their local ponds. For them, it is economically much more rational to spend some time here.
Great post. I wrote a long comment in this thread, where I went off on a bit of a tangent on how the US should institute a moratorium on funding for graduate student training, but I deleted it, because it felt overly negative and conspiratorial. You did a much better job of saying what I was trying to say, and backing it with numbers. Bravo.
Suffice to say that I agree that we're training far too many scientists in this country, and that somehow, we need to find a way to wean Universities (and industry) off the government dole. The problem is, after decades of subsidy for advanced research, the US is addicted to cheap intellectual labor in the form of under-employed graduate students and post-doctoral researchers. If the government were to drastically cut funding for these "training" positions, the university system would collapse.
My opinion is that the only way out of this conundrum, is to force industry to re-invest in R&D over the long term. That means no more licensing patents from university research programs that were developed using federal funding; no more "tech transfer" that allows corporate research to be done on campus for slave wages; and a slow dial-down in agency funding for applied research at universities (as opposed to basic science). Research and development has to be driven by market demand, if science is to be a stable career choice for intelligent students.
I was just going to post this exact link -- I'm not sure that science jobs have had much luster relative to their counterparts over the last 30 - 40 years in general because equivalent kinds of professions like finance, law, and business have vastly outstripped science pay and job security.
Greenspun's article seems a bit hyperbolic, but its main contours are accurate. If you're curious about the issue and haven't read it, you absolute should.
As I proud libertarian I find it hard to justify the case for EVERYTHING to be privately funded.
Take things like railroads which pay back the investment in them in 100+ years.
Usually the cost of borrowing is a function of how much risk there is.
Here's where things get interesting, how many private companies can you name that have been around for over 100 years?
Yes, I now there are dozen of companies older then that, but all those dozens world wide are still just exceptions.
In general almost no companies last that long.
So the risk for private borrowing over 100 years would make the interest way too high for anything.
But governments can borrow at very reduced rates, and that's why they can build railroads profitably.
The same argument can be made with basic science. We know most of the research will not yield anything, while a small part of it will in the long term create thing we can't even imagine.
But those things my not necessarily be profitable, think airlines which are a terrible business to be in.
So the government is like a basic science VC that knows 999999999999 out of 999999999999+2 investments will fail, but 1 will change the world.
And private VCs can't do the same because the risk is too high and the payback takes too long.
I think bridges and roads should be privatized, just look at France.
But I'm not so sure about rails and basics science.
I'm not sure it's a good idea to be making 100 year investments: many such investments will turn out to be investments in a 100 year buggy whip factory. Consider, for instance, passenger rail to Detroit, Gary or Buffalo (in 1900).
Note: I'm not expressing an opinion on whether basic science qualifies, I'm too biased (my salary comes from an NSF grant). I'm just questioning the assumptions.
In art the patron knows the artist. In the government funding of research, the government picks people (heads of NIH, NSF, etc.) who then pick people (division heads) who then pick people from the community (study sections) who then choose from among their peers who gets funding...
...I'd say it's probably the least politicized thing the government does.
I'd say it's probably the least politicized thing the government does
I don't know if I'd describe the grant approval process as a-political, especially after hearing my professors describe how it operates on a nuts and bolts level.
There is probably a salient difference between Congress, the grant committee reviewing AI research proposals, and a gang of chimpanzees doing grooming and dominance games. An alien race unfamiliar with primates might conclude that the difference is how many fleas are killed in each activity.
There is probably a salient difference between Congress, the grant committee reviewing AI research proposals, and a gang of chimpanzees doing grooming and dominance games. An alien race unfamiliar with primates might conclude that the difference is how many fleas are killed in each activity.
That's one of the best lines I've ever read. Your real name is PJ O'Rourke, isn't it? :)
