Discover more from ChinaTalk
Endless Frontier, The Most Important Bill No One's Heard Of + Essay Contest
Tweets of the Week Featuring Chad Laozi
Two announcements at the top. First, this month I’m launching the ChinaTalk-Rhodium Student Research Symposium! Having your final year be zoom school was not ideal to say the least, but here at ChinaTalk and Rhodium, we’re looking forward to celebrating your academic success! If you’re graduating this semester from an undergrad or masters program, and have done any research incorporating Chinese-language sources, submit your paper or thesis for consideration here! Authors of the best five papers will win $200 each and have the opportunity to appear on the ChinaTalk podcast talking with me about your findings.
And second, I’m hiring! Rhodium is looking for a research assistant to support our growing technology and industry research practice. If you have confident Chinese and strong English-language writing skills, a passion for technology policy and industry analysis, and maybe some Python or R, please apply here.
The Endless Frontier Act is the most important bill no one is talking about.
The legislation as it stands would create a Directorate for Technology within the National Science Foundation (NSF) that aims to $100bn over five years into applied research on ten key technologies:
(i) artificial intelligence, machine learning, and other software advances;
“(ii) high performance computing, semiconductors, and advanced computer hardware;
“(iii) quantum computing and information systems;
“(iv) robotics, automation, and advanced manufacturing;
“(v) natural and anthropogenic disaster prevention or mitigation;
“(vi) advanced communications technology;
“(vii) biotechnology, medical technology, genomics, and synthetic biology;
“(viii) cybersecurity, data storage, and data management technologies;
“(ix) advanced energy, batteries, and industrial efficiency; and
“(x) advanced materials science, engineering, and exploration relevant to the other key technology focus areas described in this subparagraph.
Here’s the way the NSF defines the difference between different types of research.
The term R&D covers three types of activity: basic research, applied research and experimental development. Basic research is experimental or theoretical work undertaken primarily to acquire new knowledge of the underlying foundations of phenomena and observable facts, without any particular application or use in view. Applied research is…directed primarily towards a specific, practical aim or objective. Experimental development is systematic work, drawing on knowledge gained from research and practical experience and producing additional knowledge, which is directed to producing new products or processes or to improving existing products or processes.
As I’ve harped on in the past in this newsletter, China has largely eschewed basic research in favor of applied and experimental R&D. Thanks largely to fear of China’s technological progress, after decades of ho-hum funding federal R&D as a share of GDP is set to rise.
Endless Frontier as it currently stands would be an enormous infusion of money into America’s research ecosystem, doubling the NSF’s funding over five years. My sense is that America’s comparative de-emphasis on government-funded applied research represents some serious low-hanging fruit.
Some GOP critics point out that a focus on applied research would be a foolhardy attempt to “out-China China,” a frustration that most NSF money goes to blue coastal states, and a general concern with the pricetag. But overall, given the investment that both Schumer and Biden have in this bill, it just feels like something big’s going to move.
However, Lots of people from within NSF seem to be pretty upset about the prospect of tens of billions to put to work on applied reserach. In particular, they worry that this new technology and applications-focused directorate will detract from the institution’s core focus on basic research.
Getting this right is profoundly important both from a US-China competitive framework as well as if you care about technological progress more broadly (and you should!). However, from conversations I’ve had, it seems like most of the information flow into Congress seems to consist of American universities set to gain billions in funding and former and current NSFers who seem set against this change.
If the NSF doesn’t want to do this work, and Congress believes that applied research needs a plus-up, it makes more sense to me to put the work to stand up a new independent organization as opposed stick it inside one that doesn’t believe in the mission. From my reading of bureaucratic history, new organizations with defined missions and fresh employees recruited committed to that mission (as opposed to seconded from other agencies) seem to have the most juice in them.
Science policy has been a real think tank blind spot in recent years. For the dozens of think tank papers I’ve read calling for boosts to R&D, I can’t recall one that has really dived deep into the (though I count this paper and accompanying presentation from In-Q-Tel on how the US should spend to boost its microelectronics ecosystem as a pleasant exception). I have also yet to come across any relevant quantitative work which would help right-size this funding, both from a “how much can a bureaucracy absorb” perspective as well as looking more broadly at what optimal levels of R&D may be for a developed economy. The argumentation I’ve found so far around the bill’s spending levels doesn’t go beyond “well, China’s doing it and that’s scary” and “we had higher R&D spending as a percentage of GDP back in the 60s and that seemed to work pretty well.”
Endless Frontier pushes NSF to adopt a DARPA-inspired model for hiring and empowering program managers (see here for some fantastic write ups on what the DARPA model is and how it can be applied more broadly). But this does not go far enough. With the amount of money Congress is talking about, the US should be wise to spread its bets across different science funding schemes instead of hoping a reluctant NSF gets the job done.
See Ben Reinhardt’s critiques on the bill as well as his suggestions on how to improve it.
Long term funding for people, not projects. The Howard Hughes Medical Institute selects fellows to receive ~$1m/year over seven years to be spent on whatever they see fit as opposed to constantly writing grant applications and focusing on project deliverables. This allows researchers to focus on long-term projects and to quickly discard dead-ends. HHMI’s entire budget is ~1/20th of the proposed NSF expansion but has yielded 17 Nobel prizes and created massive respect among the scientific community.
Funding medium science. There is a gap in science funding for ‘medium science’ - projects that cost dozens, but not hundreds of millions of dollars to complete. There are many grants for projects that cost <$10m. Big Science’ projects that cost >$100m like the LHC and ITER do tend to happen albeit slowly. Scientist Adam Marblestone has pointed out that there are many projects in this gap that could potentially have huge payoffs, but do not get funded because there aren’t many systems in place to fund projects of that size besides maybe DARPA programs.
Give riffing on DARPA a real shot..The proposed ‘DARPA-like program managers’ won’t work unless they are embedded in an organization set up to empower them to work on high risk, non consensus programs. The NSF is not set up this way, but it could create a ‘wholly owned subsidiary’ from scratch to riff on DARPA. DARPA’s entire budget is $3B: 3/20th of the proposed expansion.
Try faster grants. The grant application process can take months and approaches a full-time job. Formed in the face of the COVID crisis, the philanthropic Fast Grants showed that this doesn’t need to be the case. The Infinite Frontier Bill could run an experiment with an initially small pool of money for grants that require short applications and make decisions in days. You can literally science the science by rigorously measure the science quality that comes out of them compared to a control group.
Don’t shackle research to universities. The university is at the core of the system Bush set up after WWII. The Infinite Frontier Bill continues that tradition by planning to channel the money through Universities and University-led consortia. However, America has a long tradition of research done outside of universities, whether it is at corporate labs or in people’s garages. What would happen if the bill decoupled research from the incentives in academia? [Jordan: See here for another critique from former Congressional science staffers who are also skeptical of the focus on academia]
Play around with prizes. Several economists have suggested that prizes could provide an alternative to grants with much less overhead on both sides of the table. Prizes have a storied history in spurring innovation - from transatlantic airplane travel to navigational instruments.
Poke at IP. Vannevar Bush realized that scientists would be incentivized to commercialize their inventions if they could make money off of them. This (correct) sense was codified in the Bayh-Dole act in 1980, which gave universities default ownership of any IP created by federal research. Good-intentioned results can lead to bad results. Anybody who wants to commercialize research has to negotiate with dreaded tech transfer offices. The litany of problems is long, but consider the example of a complex product that depends on inventions at two different universities. Each will want a large enough chunk that it would be totally unprofitable. There are possible alternatives to the current system that preserve the profit motive while addressing the poor incentives that have become baked into the system. These include consulting-based IP transfer in which professors make more money when multiple companies use their invention and government patent purchases.
José Luis Ricón Fernández, who appeared on ChinaTalk a few weeks back, recently put together a post with some creative ideas on how to innovate on how national science organizations give out their money. A selection of some of my favorites below:
The "Yes, and" Funding Mechanism
You find a paper you love, but the author is not working on that anymore because the grant ran out or some other reason. A team of "Yes, and" officers would scour the reams of old interesting papers that came out years ago and find if their authors are interested in revisiting their own work. A "Yes, and" paper might involve revisiting an old paper with more statistical power, or with more precise, modern techniques, measuring different endpoints, or trying an altogether different approach that perhaps was even planned back then but that there was no time to actually take to its conclusion.
Science Funding Tokens
Science Funding Tokens (SFTs) would be issued to all participating scientists. Each year you can allocate tokens to other scientists you think are going to produce "good work". You can't allocate them to yourself. Funding is allocated then based on the distribution of tokens (e.g. perhaps according to quadratic funding). Every 5 years, papers funded by this mechanism are assessed and scientists earn cash prizes if they voted for the work later judged as impactful.
The Young Researchers Research Institute
Current science funding institutions typically recognize a need to incentivize younger scientists to contribute and stay in the system. NIH for example has various Early Stage/New Investigator policies. While the case for gerontocratic pessimism is not fully warranted (Older scientists are not generally at a disadvantage relative to younger ones), it is unarguably true that younger scientists have it harder than ever in steering their own investigations: They must do so under the "adult supervision" of an established PI and only in their early forties do they get to apply for grants themselves.
But what if we radically empower the young?
The Young Researchers Research Institute (YRRI) would take in students that have just finished their PhDs and would put them in an environment with other younger scientists to pursue any kind of research they want. Given that it would be a weird career move, it might dissuade some researchers, but if you truly believe in your idea, and your PI doesn't want you to do that, then the will to try might be stronger than the uncertainties involved in joining the YRRI.
The Institute for Scientific Roadmapping
What does the field of longevity research look like? Well, you can look at my longevity FAQ for that, but there is more to a field than an introductory FAQ. What are the fine-grained research questions that are roadblocks for the field? What progress has been made in recent decades? What are promising avenues forward? What high-level problem is even the field trying to solve, if any?
The Institute for Scientific Roadmapping would produce roadmaps for various fields; a roadmap might include
A map of the field (subareas, state of progress, history, key names)
Key questions the field is working on
Key questions that the field is not working on but should, and why
Key constraints the field faces
These roadmaps would make legible to scientists, policymakers, or philanthropists a lot of the implicit knowledge that exists within a field, with the idea of using that roadmap to then channel resources to the appropriate questions or bottlenecks. These could be addressed with policy, research projects, or more general grant funding. [Jordan: note that this happens for semiconductors]
Endless Frontier is a rare bill with bipartisan support that is big enough to really impact the humankind’s technological trajectory. We’re in a rare moment where good ideas have a decent chance of bubbling up into the legislation as staffers I’ve talked to on both sides of the aisle seem very receptive to suggestions. If you have thoughtful ideas on how Congress should best direct its momentum towards funding R&D, please do get in touch.
Tweets of the Week
Been using this all week, it’s fantastic.
And a rebuttal:
Classic bamboo barbell