The opposite of “exactly”. Too many projects are low potential, because not enough resources are thrown at them, just to satiate peoples’ need to “save money”.
The question is, how do you pick them?
There is probably a good case for funding some ‘blue sky’ projects. Nothing ventured, nothing gained.
But how do you distinguish something that just might pan out from something that is pure crackpot?
2023 appropriations:
NSF $9.877 billion
NASA $25.4 billion
DOE $48 billion
NIH $47.683 billion
NIST $1.65 billion
NOAA $6.9 billion
Maybe $150 billion, total.
I’m not easily coming up with numbers for DoD research programs, but let’s say that doubles the number.
So $300 billion. And those are total agency budgets, where a lot is not research.
That compares to 2023 US Federal spending of $6.1 trillion.
We’re spending less than 3% on research. I don’t think “crackpots” are the problem.
I wouldn’t disagree. Scientific research has almost always paid off eventually in useful applications!
But you do need to somehow find a way to sieve possible ideas versus the tinfoil hat woo-woos…?
You’re spending 3% on government-funded research. A hell of a lot more than that is spent on R&D. Last year Microsoft alone spent more money on R&D (27 billion) than did NASA. SpaceX has spent between $5 billion and $10 billion on Starship, and more on Starlink. IBM spent $6.3 billion on R&D last year. AI research investment in the last couple of years has been in the tens of billions, and is expected to reach $200 billion globally by 2025.
According to the OECD, 60% of all research is funded by private corporations, vs only 10% by government and 20% by universities.
And no, it’s not just funding for some new product development. Plenty of fundamental science is funded by corporations.
The Onion could have an end-of-year article about how fusion researchers met their yearly goal of keeping fusion power no more than twenty years away.
Also in science “news”:
- scientists re-re-re-re-re-re-re-re-re-re-re-re-re-re-re-re- confirm General Relativity.
- search for dark matter particles turns up negative.
- SLS launch delayed.
‘Research’ should be put in air quotes when it comes to corporate interests because that often encompasses work such as ‘market research’, pharmaceutical companies ‘researching’ how to extend their patents through minor changes in formulation, the tobacco and food industries ‘researching’ how to conceal harms and promote broader commercial appeal of their products, et cetera. Of the major technical innovations of the 20th century, only the transistor could be claimed to have actually been developed by purely commercial interests (primarily Bell Labs) although the applications beyond transistor radios and large stored program computers were largely driven by military and space applications.
Even when corporations are on the forefront of research—as in the case of generative AI—they are often so concerned about being ‘first to market’ and maintaining competitive advantage that they don’t publish or share information readily, which results in multiple research efforts going down the same blind alleys and making similar mistakes. One of the frequent criticisms about nuclear fusion startups by scientists working the the field is that they are very secretive about their experimental work, which results in repeating research that was determined to not lead to a viable path years or even decades ago, while the academic labs that have worked on fusion and publish a wealth of information about fusion and plasma research have seen progressively less funding and in some cases have just shut down entirely despite great research and experimental work that they were providing.
Stranger
1899: Lord Kelvin declares “Physics is complete. Just a matter of a few more decimal points on the measurements”.
Don’t listen to those charlatans (Curie, Rontgen, Bequerel) behind the curtain…
According to this page, he didn’t actually say that.
Yes, of course. Butn then, not all government ‘research’ is pure research either. My first job out of college was on an NSERC grant. I was hired to write fast fourier transforms for some chemical analysis stuff. What I actually wound up doing was building a customs database for chemicals for the bean counters. But it got billed to the government as ‘research’. Also, government research is often top heavy in salaries for bureaucrats.
But corporations don’t just do product research. Plenty of basic physics, chemistry and materials science is done commercially. In fact, most of the things changing the world right now, from SpaceX rockets to AI and mRNA techniology, came from private organizations, not government. Small Modular Reactor designs are coming from private industry.
GE scientists have won two Nobel prizes.
John Bardeen won the Nobel for the transistor while working at Bell Labs.
Walter Gilbert won the Nobel prize in Chemistry in 1980 for DNA sequencing while working at Biogen.
The 1993 Nobel in Chemistry went to Kary Mullis, who invented the PCR test while at the Cetus corporation.
The Nobel in chemistry went to a Dow scientist a few years ago.
Carbon Nanotubes were discovered at NEC.
Robotics has been greatly enhanced by Boston Dynamics, Tesla, and others. Noore’s law is really a statement about the market’s ability to continually push the boundaries of physics and chip size. We wouldn’t have cell phones today without the billions in research spent by Qualcomm, Motorola, and others.
Slso, it should be noted that as we get closer to fundamental limits in the design of many products, product research is increasingly looking like basic science research.
Neither “SpaceX rockets” nor “mRNA technology” are basic research; they are applications of prior research and development and utilize many decades of mostly government funded research performed at government and academic labs. Small modular reactors are such a direct descendant of US Navy research applications of nuclear power that it shouldn’t even need to be explicated.
Generative AI can actually be considered a hybric of ‘basic’ and applied research done predominately by private industry (although again leveraging decades of largely academic research) but also exposes the dangers of commercial motivation in research on the cutting edge of technology, to wit being in such a hurry to be first-to-market with a novel technology that no time is taken to consider the impact of potential applications, and anyone who tries to put a pause to consider the ethical implications or establish guidelines for public safety gets run out on a rail, while the “hype men” overpromise and underdeliver, creating expectations that will not be met.
Robotics is definitionally an application, not basic research, and Moore’s Law is an empirical observation about the rate of transistor density on a chip, not any kind of general basic or physical principle. Cellular communications are another application, although one that did involve basic developments in signals processing, but the hardware side of mobile technology has its roots, as does all computing technology, in military and space applications.
No, it doesn’t. In fact, the gulf between ‘basic science research’ and applications is growing as our knowledge in fundamentals in science exceeds our ability to industrialize it. There are a handful of exceptions, such as quantum computing or CRISPR-Cas9 gene editing technology where basic research actually produced a practicable application, but much basic science from the discovery of the Higgs boson and cosmology from JWST and other orbital observatories to neuroevolutionary history and planetary science has little actual application in the near term, and the problem in conflating basic and applied research is that the public ends up expecting that all scientific research of value should produce shiny baubles or else isn’t worth funding or even learning about.
Stranger
In most US institutions of higher learning, the highest-paid faculty member, usually by a large margin, is the head coach of the football team.
Show your work.
the definition of “bureaucrat” can be somewhat flexible
I track Federal R&D spending for a living. @Ruken 's graphic is still about right for 2023. Total R&D funding has grown over the years (currently a tad under $200 B), but Basic Research has been flat at about $40-45 B per year for the past ~20 years. It sound like a lot, but it is divided across just about every university and government lab across every discipline in science.
Corporate R&D funding dwarfs Federal R&D funding, but as @Stranger_On_A_Train points out, it is concentrated on getting things to market or protecting things in the market. While some corporations are notable for funding basic research, it tends to be a tiny amount of “seed” money from their R&D budgets. The majority of corporations that identify R&D funding spend nothing on Basic Research.
SpaceX rockets were based on improvement to existing products. I’m not sure what was basic about it. The fundamental breakthrough in generative AI occurred in 2017. Most of what created today’s ChatBots was development and scaling.
And the Air Force has funded 7 Nobel Prize winners. Counting prizes is a fool’s game (and that includes the AFOSR counts)
GE’s Nobel prizes were awarded 50 and 90 years ago. Bardeen won his transistor Nobel almost 70 years ago and Bell Labs hasn’t existed in any meaningful way for decades. What does this have to do with how research is being funded today?
I thought we were trying to convince @xtenkfarpl that research funding wasn’t all going to “crackpots” and “tinfoil hat woo-woos”
And that’s what you think AI funding is?
It’s safe to conclude there was no work to show.