When America endeavors to tackle an ambitious project, we speak in terms of moonshots or a “Manhattan Project for X.” The assumption is that vast government resources, directed toward some objective, can yield results on the scale of the Moon landing or the atom bomb. But federal research funding is more complicated than throwing dollars at our problems. And with Congress poised to inject American science policy with an adrenaline shot of funding, I’ve brought Tony Mills back on Political Economy to discuss the bills working their way through the House and Senate.

Tony is a senior fellow at the American Enterprise Institute, where he studies the federal government’s role in scientific research and innovation, as well as how to integrate scientific expertise into our governing institutions.

What follows is a lightly edited transcript of our conversation, including brief portions that were cut from the original podcast. You can download the episode here, and don’t forget to subscribe to my podcast on iTunes or Stitcher. Tell your friends, leave a review.

Pethokoukis: I was under the impression that the United States was about to embark upon a great period of federal science research. That because of, secondarily, years of subpar productivity growth, and maybe more primarily the China threat, we were just going to start throwing money at R&D. Is that going to happen? It doesn’t seem like it’s happened yet, but is that going to happen?

Mills: Well, that’s a good question. To a certain extent, it requires predicting political events, which is a hazardous game. I think if we back up a little bit, what we can say is there was this push—it really picked up in 2020, interestingly before COVID really took off. The best example of this was the Endless Frontier Act, which was Chuck Schumer’s bill. It was a bipartisan bill, introduced in the Senate in 2020. It was very ambitious, and among its ambitious proposals was a $100 billion investment in the National Science Foundation and a considerable reorganization of that agency. Fast forward to today, that bill passed the Senate, but it is currently still being negotiated with the House whether or not anything like that will come to pass.

What happened in the meantime is the House introduced its own version of this bill. There were some important differences between them, and there’s been kind of a back-and-forth debate about what we need to do over the past few years. That’s kind of still where we are. It looks like Congress is hoping to pass something by this summer. It could turn out that a version of this bill will in fact pass, but it’s a very different kind of bill than what was originally introduced. It’s also considerably less ambitious in terms of its budgetary targets.

So, just to recap because it can be confusing: We have two bills and both would spend $80 billion or so on research. (You can correct me on all these.) The House bill would spend it a little bit more like how we traditionally spend money in this country: mostly on kind of basic research, less on more directed or applied research—maybe you want to explain what that even means. Then the Senate bill is different: It would spend money on basic research, but it would also have this sort of technology directorate, and it would focus especially on certain key sectors as selected by government—AI or biotech, high-performance computing. We have two bills, maybe about the same size, but a little different in focus.

Correct. Although, I would complicate it a little bit further.

Good. Please complicate it.

The Senate bill, you’re correct, in its original form had two principle focuses. One was this $100 billion investment in the National Science Foundation (NSF) to create a tech directorate within the NSF. The NSF traditionally is the agency that funds basic scientific research. It’s the only agency in the federal government whose sole mission really is to do that. A lot of other agencies do fund basic scientific research—which is a controversial term but roughly means research that is directed at advancing our understanding rather than building technology, inventing medical devices, or something like that; more sort of curiosity-driven research is one way people will describe it. That’s what the NSF does. It was created to do that in 1950. It does more than that, but that’s still its primary purpose. The vast majority of its research budget goes to basic research.

I should add one more thing about that from the history. One of the issues that was hotly debated in the lead up to the creation of the NSF was how to allocate federal funding. Should it be allocated meritocratically—give the federal spending to the best science, the best institutions—or should it be distributed geographically? There was some precedent for doing that in the federal government. The Department of Agriculture’s experiment stations, for example, are in each state. The thought was, “Should we distribute federal research geographically for political reasons, or should we just give it to the best scientists?” The meritocratic view more or less won out in 1950, and the NSF has essentially operated in some version of that. There are exceptions and so forth.

What’s interesting about the Senate bill, the Endless Frontier Act that came out in 2020, is that its two central pillars were the contradiction of both of those things. It was a geography-based innovation program in which they proposed a regional tech hub program to try to build up innovation in areas that are traditionally less competitive in science and technology. The other was this tech directorate, which as you note would be keyed into strategic areas of technology that would be updated every five or 10 years. I don’t remember specifically. Areas like artificial intelligence, quantum computing, green technology, and so on and so forth. What’s notable about this is that this would be quite a change for the NSF, right? Not only in terms of its budget, which is under $10 billion. (The original proposal was $100 billion and it would all go to this new directorate.)

Is that $100 billion per year or over a number of years?

Over five years. Now that was the original plan. It has been considerably whittled down since then. Now the directorate would receive something more like $29 billion—still substantial.

Wouldn’t that increase its budget per year?

In the original proposal, over five years this investment would be divvied out. In the current version, I don’t recall the specifics of how it would break down. But through the course of the political debate over this bill, the ambitions for the budget were kind of whittled down because funding went over here, DOE got into the mix and it became kind of a political mess in a lot of ways.

What was interesting about that bill was, it was called the Endless Frontier Act, which is a reference to the report The Endless Frontier, written by Vannevar Bush, FDR’s science advisor and the head of federal research during World War II. That sort of was the blueprint, the constitution for the NSF. I just note the irony that it was called this, but it was really in contradiction to sort of the spirit of Bush’s vision. I think it took Congress by surprise that the scientific community was not altogether happy with this bill. It’s a huge amount of money, but because of the emphasis on this sort of directed research, the way in which this could radically change the institution of NSF, there was actually some pushback. It turned out to be more controversial, I think, than anybody was expecting.

Meanwhile, the House introduced a rival bill. This was the NSF for the Future Act, which was more responsive to the scientific community’s view of what should happen. As you say, there would be a new directorate with a different name. It would focus on science and engineering. Now, it would also be geared toward technological competitiveness, so use-inspired basic research that is trying to bridge the gap between scientific research and technological innovation. But it would be kind of within the traditional operations of the NSF. There are other differences that we can talk about. That’s kind of where things were a few years ago. What’s become even more complicated is that the Endless Frontier Act was rolled into a larger, much more sprawling bill: the US Innovation and Competition Act, USICA as it’s known, which does a lot of things, not just science policy. Meanwhile, the House rolled its various science agency authorizations into its own large sprawling bill, the COMPETES Act, which also does a lot more than science policy.

There is some trade stuff in these bills, too. I’ve heard these bills called the chips bills, because we’re going to try to promote domestic semiconductor manufacturing. So “sprawling” is probably the best word, right?

Sprawling, yeah. I would say that’s the common denominator. Competition with China is a big issue for both, maybe a little more on the Senate bill, and trying to beef up strategic areas of domestic industry, areas of technology, particularly semiconductors. They both do that in different ways, but they also do a lot of other things, too. The Senate bill was passed as bipartisan. A lot of support. The House bill was passed almost entirely on partisan lines because there were objections to all these ride-alongs that got in there that had really nothing to do with science or even technology policy. Now Congress is in a negotiation to try to reconcile these two things. Meanwhile, the administration proposed a number of its own R&D, science and technology policy proposals. Of course, we’ve had budgetary preparations debates happening separately. In the meantime, NSF funding has actually increased and the agency decided to—

It increased through the basic budgetary process?

Precisely, and not by nearly as much as these bills would’ve done or still could do. Also, the NSF, sort of through its own initiatives, set up a new technology directorate. It’s in the process of doing that right now. What that will look like is to be determined, partly depending on what happens with the reconciliation process between these two bills. It’s a complicated situation.

Support for basic, curiosity-driven research that’s not meant to solve a defined problem is fairly uncontroversial. But how far beyond that basic level of research should we go? How much direction from Washington is a good idea?

It’s a good question. I would answer it, hopefully not by dodging it. I think there are two ways to understand the question, and it really turns on the ambiguity of these terms. On the one hand, you could say basic research, curiosity-driven research, is a sort of a commonsensical-sounding thing, right? Let scientists do what they do. In practice, it’s complicated to really define what we mean by that. It’s also rarely the case that scientists are ever doing whatever they want. I mean, they’re highly dependent on external sources of funding, federal and other. They propose projects and they try to get funding for the project. What kinds of projects get funded are obviously partly determined by what the agencies are interested in, what Congress is interested in, what other sources providing funding are interested in.

I think there’s a genuine question about what it is we mean by these terms. It’s not merely academic to try to figure that out. Part of the problem is that when you use that term, it means different things to different people. Some people hear basic research and they think astronomy, just trying to study the chemical composition of a star.

You’re just adding generally to the stock of human knowledge.

Exactly. But you might mean, “Basic research. Oh, early-stage research in a given technology area.” I mean, in different sectors, different research communities will use these terms in different ways. Basic research means one thing in the medical community, means something else in the physical sciences, and means something else in private industry. The complications around that, I think, are actually not trivial.

Let me try to answer the question. One way to do it is to say, “Well, what fields should we be funding? Should we be funding the physical sciences? Should we be funding particular areas of engineering or whatever?” There’s another way to hear the question which is, what is the government’s proper role in the broader science and technology ecosystem? I think often the debate about basic versus other kinds of research is kind of a proxy for that larger set of issues.

One of the problems, to my mind, is that when we talk about science and technology funding, we often talk as though the government’s involvement is purely monetary, as though the sort of infrastructure that’s been created to fund science by the federal government (we can talk about private industry, too) has no impact on the scientific community. I think that’s really quite wrong. There are a lot of problems within the scientific community that are arguably traceable to this sort of massive, sprawling enterprise that we have. The bureaucratization of research, the project-driven nature of science is something that a lot of scholars have pointed out as being problematic. There’s the replication crisis, which we could talk about, which I would argue is related to these things, where you have a kind of drive to publish findings that may or may not really be that robust. There are a lot of issues.

I think one of the ways that this debate plays out is in terms of how much government control of science we have or want. That’s separate from the question of, “Which fields we should be funding?” I mean, it could be that the most promising developments are going to be in biomedical science versus material science. That’s a hard question, to know which field we should prioritize.

Let me ask the original question slightly differently. I often hear things like, “Well, we should do a super DARPA. That’s how America should do its science R&D.” Tell me a bit about the Defense Advanced Research Projects Agency. What do they do? And what can we can learn from it?

It’s a great question. This kind of embodies the ambiguity I was talking about, actually. DARPA is an agency within the Defense Department, the Defense Advanced Research Projects Agency. It was created in 1957 or ’58, ’58 I think, in response to Sputnik. It’s changed over time and there are different eras in DARPA’s history that some people think have been more productive than others, but I would say by all accounts, it’s been very successful. It’s successful at seeding breakthrough technologies. It does that. One of the ways it does this is it gives a lot of discretion to the people within the agency who are in charge of the research programs to really just experiment and do kind of whatever they want. They have a lot of freedom, which is different than the way a traditional science agency might operate.

Because of that, a lot of people have, exactly as you said, drawn the conclusion, “Well, why don’t we imitate DARPA in other areas?” We actually have done that. There is an ARPA-E in the Department of Energy, for example, modeled on DARPA. One of the motivations behind the tech directorate in the Endless Frontier Act—I believe it says this in the text of the bill—was to create a DARPA-like entity within NSF. There have been proposals for a health ARPA, ARPA-H, which has recently become reality. The Biden administration essentially stood one up within NIH. It has its own weird structure.

One of the debates about creating these other ARPAs is, should they be standalone or should they be housed within other agencies? If you house it within another agency, you take on a lot of the freight of that agency for better or worse. One of the things that makes DARPA unique is its kind of independence. A lot of people have criticized the Biden administration’s decision to put ARPA-H in NIH. Although the director of ARPA-H would report directly to the HHS secretary to sort of bypass NIH in some way. ARPA-E is in DOE. There are different kinds of models for thinking about that.

Now, is it a good idea? I mean, I think it depends on what we’re trying to do, right? DARPA is often used as a good example of government funding basic research effectively. Really if you look at what DARPA does and especially what ARPA-E does, they don’t really fund that much basic research. If you think about research on a pipeline (which I think is not really the right way to do it, but people do all the time), you kind of have basic research here and then you have commercial development on the other side, and then in between is a spectrum. DARPA is certainly funding early-stage experimental research, but the idea is to build technologies, right? It’s a military research agency whose purpose is to develop technologies that are useful. ARPA-E is similar. The purpose is to develop technologies that are useful for green technology, for instance, or other areas of energy. If you look at where they spend most of their money, it’s kind of in the middle of the pipeline. It’s sort of on the early-stage commercial research part of the spectrum.

It’s a successful model in a lot of ways, but it shouldn’t be confused with scientific research. It’s really more of a way of organizing technology research and to do it, I think, in ways that have been successful. Now, is it applicable to every domain? I think that you have to go case by case, right? I mean, it really depends. One of the things I would say though is, we have DARPA. People like DARPA and they want to imitate it, but DARPA exists, it’s still doing what it does, right? A lot of the breakthrough technologies that it has helped give rise to—from GPS to the internet—that wasn’t really the intention. The intention was to develop technologies that were useful for the military. In a sense, by trying to imitate DARPA to do a predetermined outcome, you’re kind of not doing what DARPA does, in a way. I think that there’s a lot of excitement about that idea, but in practice I’m a little bit more lukewarm about whether it would actually be effective. But again, I think it depends.

How can we get more out of our science spending? There’s a field called metascience about how we do science in this country and the incentives. Can we be doing that substantially differently and getting a better end result, or have we sort of optimized it in this country?

I do not think we’ve optimized it. The short answer is, yes, I think we could do things a lot better. Precisely how to do that is where things get complicated. Maybe it’s worth stepping back. This is what I was getting at in answer to your question about basic research.

I guess I’m asking, should we change the incentives? Should we be having more lotteries? Are there other ways that are different than the way we currently distribute money?

Yes. The established way of funding science research, if we take, for instance, NIH and NSF as examples, is a peer review system where you evaluate the merits of a proposal using a peer review system. You get people in the relevant fields to look at it and decide, “Does this look promising or not?” and then you ultimately decide whether or not to fund it. I think there’s a genuine question of whether that model is the most appropriate way or at least should be the only way that we fund science. Peer review is extremely important. There’s a lot of discussion about the limits of peer review. It’s overly conservative and so forth. I think peer review is actually very, very important. It’s hugely important for science. Whether it’s the most effective way for the federal government to allocate its research funding is a different question.

In that one area, I think there is a lot of room for experimentation, for other ways of distributing funding. I think lotteries are interesting. I’m on record in favor of a modified lottery system, at least as a tool in a toolkit. Certainly that’s not a silver bullet and they have their own problems, but I think that what we do need is better evidence about what works for what purposes. We can only get that if we inject some more experimentation into the system. We currently do things—I wouldn’t say one way, but there are certainly received ways of doing it. Agencies like NIH and NSF do study these metascience questions, but they have an interest in a certain outcome, obviously. I think that having more boundary organizations (as a political scientist would call them), institutions that are not just the NIH and NSF looking at how we’re doing these things and whether it’s working, is very important so that we have a better evidence base for making decisions about how we do this.

I think if we step back a little bit, the important question (this is what I was getting earlier) is, what has been the effect on science of all of these infrastructures that we’ve built? There’s a kind of widespread assumption that if we just crank up the spigot with money, we’re going to get more outputs. We’re going to get more good science and technology. But it doesn’t really work that way. There are a lot of issues that sort of plague the research system which really need to be looked at. I think the funding models is one. The other thing that I alluded to earlier is the bureaucratization of science.

Ghastly percentages of researchers’ time go to essentially doing paperwork to comply with various aspects of regulations and rules that come from the funding sources that they have. Some of this is on the federal government side, some of it is within the universities themselves. That’s a real problem. I mentioned the replication crisis. I think that we need to be thinking more carefully about how to improve, get rid of the kinks in the system, before we just flow more money through it. I’m not opposed to increasing federal spending on science by any stretch, but there’s too little attention given to that.

The two bills we were talking about are an example. The Senate bill mentions almost nothing about any of these metascience questions. The geographic diversity issue is the only kind of metascience issue in there, if you want to include it. The House’s is a little better. There’s some discussion of replication in the House bill. It’s pretty limited. It’s mostly in the computer science field. To me, this is very unfortunate because there are longstanding debates and lots of scholarship on these questions that we could be drawing on in thinking about ways to reform our scientific institutions.

Do you think the current push for expanding federal investment is a singular moment and we’ll be on cruise control after this moment? Or do you think, because of concerns about long-term economic growth and geopolitical concerns about China, that this is the beginning of a period of ramping up science?

Yeah, that’s a good question. I don’t know. Coming out of COVID and with a lot of the other factors that you mentioned, there’s competition with China, there is kind of a perfect storm of forces that sort of incentivize people in Congress to want to increase federal R&D spending. Even now, I think what happens, we’ll have to wait and see. Whether this is going to recur, I don’t know the answer. Again, step back a little bit. Federal R&D spending always goes up. Scientists are always complaining about insufficient funding. I’m sympathetic. But on the other hand, if you do look at overall trends, science funding, it goes up, the question is where is it going up how much? There are periods in recent history around the turn of the millennium when spending has gone up quite a bit, doubling of the NIH budget, for example. Certainly it’s not out of the realm of the possible that we could have really significant federal investments maybe now or in the future.

What I don’t think is likely is a fundamental shift whereby the federal government becomes the principal funder of R&D. That was the case in the postwar decades. The high point was 1964 when I believe it was close to 70 percent of all US R&D spending was federal government. It’s basically flipped now, with the private sector taking the lead. It’s worth noting that overall US R&D spending is extremely high. It’s comparable of that high point in the ‘60s. It’s just that it’s flipped.

The private sector tends to be more interested in applied research and development by a lot. That has skewed the overall priorities of the US R&D system. We have lots of funding, and federal funding is also pretty significant, which a lot of sort of science technology policy scholars and advocates alike. Will it flip back the other way? That seems very unlikely because the numbers would be really quite staggering. But it does seem to be a moment when there is bipartisan interest in increasing funding. Again, I don’t want to exaggerate how exceptional that is because doubling of the NIH budget—I mean, imagine if we did that today. That would be a $100 billion budget for the NIH. That would be quite significant. I think we are in a moment that’s like that, where there is this bipartisan appetite. I don’t know how long it will last, but I’m skeptical that it will fundamentally change the current R&D system.