Rethinking the science of science funding with Sam Arbesman

Danny Crichton:
All right, I think we've got it. The old board did not work, but we have the new board or we have have what looks like a Geiger counter.

Sam Arbesman:
Yeah. I'm not sure what's going on.

Danny Crichton:
I don't know what's going on, but we are ready to go and it sounds good. 3, 2, 1. What do I say in this podcast? I don't even know what I say anymore on this podcast. 3, 2, 1. Welcome to The Securities by Lux Capital Podcast. I'm your host, Danny Crichton, and here we're talking about science, technology, finance, and the human condition. Today my guest is Sam Arbesman, Lux Capital's scientist in residence, who's visiting us here in New York for a conference on what, Sam?

Sam Arbesman:
About new types of scientific research organizations.

Danny Crichton:
This is a subject near and dear to my heart because when I think of the mission of what we're trying to do with Securities, improving the pipeline of great science, that's just so fundamental to progressing human knowledge, accelerating economic growth and more. And what's interesting to me is that this is a subject that you've written about a lot on. Even going so far as compiling a compendium of these organizations and what you dubbed The Overedge Catalog. Why your interest and why are more science entrepreneurs thinking about new institutions these days?

Sam Arbesman:
Over the past few years, there's been this explosion of kind of new types of research organizations, things outside of traditional academic universities or corporate industry labs or even deep tech startups. There's just been this explosion of organizations saying, "We want to do research, but we want it to maybe be more interdisciplinary, maybe in a for-profit or nonprofit structure." Either the research output is different, the structure of the organization is different, the types of people who are involved in the research are a little bit different.

Danny Crichton:
And why are there so many people suddenly interested in funding science? Because the government's been doing this since the 40s with the advent of the National Science Foundation, National Institutes of Health. But all of a sudden in the last two years, we've seen basically a [inaudible 00:01:42] explosion of new science foundations. Why all of a sudden?

Sam Arbesman:
Part of it is just I think there are more people with money who want to actually fund these things, which is exciting. So now there's actually the potential for people who want to build these organizations to actually now get funding to actually continue to actually make them. But I do think there is kind of this evolutionary process where people are just trying lots of different things and some of these organizations, universities or [inaudible 00:02:05], they're in some high dimensional space of potential organizations and we've really not explored it and we need to explore it. And then we're going to find out some of them are actually quite fit. Some of them are things maybe we shouldn't try. But they're all kind of interesting mutants and we need to just try lots of different organizations. And luckily now people are saying, let's actually give this a shot.

Danny Crichton:
When you think about the academic research complex, which obviously is massive, tens of billions of dollars through the major science research foundations, what are people responding to? What's going wrong with that system? Because obviously many innovations in the 40s, 50s and 60s and 70s, the internet being one of them came out of these sorts of research organizations. Why are people responding today?

Sam Arbesman:
A number of different factors, but I mean certainly if you look at, I think the age at which researchers get their first major grant, that has increased over time. It takes a longer and longer amount of time to get to the point in your academic career when you can begin to do potentially game-changing research on your own. And people have just noticed that. So within biomedical fields, rather than just getting a PhD and then becoming a faculty member, you might get a PhD and then do postdoc after postdoc after postdoc, and then realize there are no actual faculty positions. And you mentioned about the 40s, 50s and 60s and 70s, that period, in addition to having a lot of research funding, that was when a lot of universities were both being started and universities were expanding, and so there were a huge number of faculty positions available.

And that window, for the most part, I think it's closed. At this point now, and maybe pyramid scheme is a little too strong. But at the same time though, if you look at traditional academia within science, you might have a single PI, a single faculty member and they have postdocs and lots of grad students below and a whole bunch of research techs. And even though the kind of traditional mindset within academia is you get your PhD, maybe do a postdoc or two, and then you become a faculty member, there just aren't the same number of positions as there used to be in previous years. And so I think as a result of that, people are casting about for new types of models and saying, "Okay, if I want to do research but I can't do it in the traditional structures, is it worth creating my own thing? Is it worth trying something new and something different?" And so I think that's kind of another reason why people are beginning to explore this.

Danny Crichton:
Well, it's really good timing because I just wrote about science and aging in the newsletter last week. And for the National Institutes of Health, their RO1 grant, which is the core kind of mainstay grant of the NIH, the average age of scientists has increased from 38 in 1995 to 42 today, which means that if you want to do an independent research project, you're basically 20 years post undergrad before you're even allowed to operate your own experiments and run your own lab.

I've always had this struggle though of, but science has also gotten harder. If you go back to the 40s, not to say that quantum physics was not complicated, but I do think there's this category of if you go into the biological sciences, there's a reason that some folks are going into postdocs. You need more techniques, you need more knowledge, you almost need a second PhD to be able to connect bio and AI or bio and biochemistry or whatever the case may be. Isn't there an argument that this is just a natural evolution as we get closer to the edges of science that you just need to know more to be able to do original research?

Sam Arbesman:
There's a research paper where in the title there's this idea of the burden of knowledge, this idea that as a field becomes more mature, to make advances at the frontier, you just need to learn a lot more to get to that frontier. And so there's an increasing burden of knowledge, and so it becomes very tough to get to that point. And there's definitely some truth that it's just harder to make advances, but I think some of that is also artificially placed on top of the system where ... And you can get up to speed in things fairly quickly, but because of the structure of the guild system of you have to train an apprentice, it just takes longer.

Danny Crichton:
Well, I guess when we're talking artificial constraints placed on top of the system, a lot of the complaints are around grant making, no?

Sam Arbesman:
Oftentimes to get a grant, not only maybe you need to be a little bit more advanced in years, but you also need to make sure that everyone who's reviewing the grant thinks it's a really good idea. And so people have come in and said, "Well, maybe we should actually try different methods." And saying, "Maybe a grant where half the reviewers think it's a really good idea and half the reviewers think it's a terrible idea, those are the grant proposals that should be funded because those are going to be the most polarizing and weird things." You don't want to fund grants where no one thinks it's a good idea, but if there are some really well-known people and credentialed people who think this is interesting, but other people who think it's a terrible idea, then maybe there's something there. And so I think people are also trying to figure out new ways of, I guess, highlighting grant proposals and research ideas that may be in this kind of more bureaucratized world are not getting funded.

Danny Crichton:
I think one of the concerns with the big science grant bureaucracy though is how to handle fashions in science. The counterargument is that this National Science Foundation and other government institutions aren't trying to avoid funding all the exciting research on the frontier, but rather are trying to stay above the daily fashions of what's hot right now in science. With these new science organizations, are they going to be more fashionable or Sam, do you think that they're going to actually protect science from some of the vagaries of the interests of grant makers?

Sam Arbesman:
I can see certain organizations becoming very fashionable because there are some organizations that are actually funded and founded specifically for the reason of working on something that is very hot right now. But because of the pace and timescale at which universities operate, they can't really do much of that. If you want to, I mean, have an organization that kind of, I guess, stands the test of time even for 5 to 10 years, even just can actually do anything, it needs to find funders.

And so there's still is the argument to be made of why you have to talk to funders, whether it's private individuals, private foundations, public grant making agencies. You still have to create something of value, even if it's a public good in terms of knowledge. What people think is valuable is still subject to those kind of fashion pressures. So it might be the kind of thing where the things that the NIH or NSF thinks are the hot things right now are different than what private funders think are the hot things right now. You still have to ultimately say, "Okay, I want to build something that is generating public good" and you have to convince someone to fund you in some way. I think some will be more immune to others, but ultimately there are different groups that are thinking about different trends, and it really kind of depends who you align yourself with in terms of what the trends are.

Danny Crichton:
If you had to give the optimistic, the median and the cynical case over the next decade, the cynical case is probably nothing changes. Maybe that one's easy, but when you think about the optimistic case over the next 10 years, what does that look like for research funding of institutions?

Sam Arbesman:
Yeah. And I think an optimistic case, which I mean might actually ... And I'll include a little bit, like a hint of the cynical even in the optimistic case, which is that a decent fraction of these organizations actually end up failing. But the reason I'm optimistic about that is because that means that there is, to a certain degree, an effective evolutionary process happening here where we're trying a lot of things and learning actually what works long term and what doesn't.

But the optimistic case would be that as a result of these kinds of organizations, there is now greater movement between academia and these other organizations. It would be considered acceptable to move to one of these and move back to academia, as well as there're now being this kind of a rich ecosystem of other organizations where people can move within that without having to worry about having academia in the background.

In addition, government agencies are much more open to funding these kinds of things. They're now willing to change the decision making process behind how they make grants. So whether it's more random or doing these kind of things with high variance grant proposals, those are the ones that get funded, as well as potentially even funding some of these organizations. And so I've seen a proposal around the idea of, or at least talk of almost using these organizations as beta versions of like, oh, let's try these things out, and if they seem really good, the government rather than adopts these things is like, "Oh, this is a really cool new research institute. We will now just give it much more stable funding and allow it to continue doing its kind of thing." And so the equivalent would be back in the day and realizing that Xerox PARC is doing something very exciting and the government, like the NSF kind of steps in and says, "We want to just fund this thing." And I don't know, not necessarily make it a government research agency, but at least give it some sort of stable funding.

The less successful one would be that nothing changes, and a lot of these organizations fail where we've learned some things, but we actually haven't learned from it and done anything based on it. And so we just realize in retrospect, this was an interesting moment in time when a lot of money was poured into interesting things and the organizations didn't end up doing more innovative research than people expected and nothing was learned from it, so that would be very negative and sad. But I already can tell that I think that's not going to happen. I think some interesting research is going to come out of these.

Danny Crichton:
Well, that's what I was thinking about because for the first time, you're going to have a statistical set of right research funding organizations, an actual good sample of different approaches, different evolutions of these organizations. I'm curious, one, how do you get the best lessons out of that? As well as how do you evaluate work in a context in which research can oftentimes take years, decades in order to come to fruition, to know which ones are actually working, and which ones are not?

Sam Arbesman:
We're still kind of on that border, I think, between anecdote and data in terms of the number of organizations. The best we can do right now is collect a lot of data. Don't use the data for how these organizations are doing, to drive the organizational decisions in the moment. Over focus on metrics can often end up meaning people focus too much on the wrong thing. If you focus too much on citations of papers, then that's all you're going to get. You're going to get lots of highly cited papers, but not necessarily highly influential or game-changing or impactful papers. It is a long term game like science is slow, it's messy, it's a very human endeavor, which is very exciting. You want to collect the data, you want to kind of go into this whole process with open eyes, but you don't necessarily want to quantify it too early or too much, and then use that to make decisions too quickly about whether or not something's succeeding or failure.

Danny Crichton:
I think that's the most optimistic take in all these new developments. We're finally seeing investment in what you might dub meta science, the science of creating science. We're going to get data, we're going to evolve some of these institutions and hopefully some of the best practices will migrate to the government where the big budgets are and will have a lot of impact. But I know you have a conference up the street to get to, so I want to thank Sam for joining us. You have a great rest of your trip.

Sam Arbesman:
Thank you very much. It's good to be in New York.

Danny Crichton:
President Joe Biden recently announced his proposed budget for 2023, which includes huge increases for science research organizations. One major beneficiary is the National Institutes of Health, which would potentially see an increase of 46% in its funds compared to 2021, much of it devoted to pandemic preparedness and vaccine development. Meanwhile, the Defense Department's budget for RDTE, Research, Development, Test and Evaluation, reached a new high of $130.1 billion, a 9.5% potential increase. These are all headline numbers, and of course, all these budget line items will be negotiated in Congress where they're all but certain to be scaled back. In the 2022 budget last year, which was just passed last week, ambitious increases in science funding were scaled back, and the pressures on Congress this year will be even more keen as inflation spikes and the economic fallout from Russia's invasion of Ukraine continues to spread. Nonetheless, it's a healthy time for science budgets, but the question is, will it matter?

As Sam was just discussing, there are corrosive issues in the American R&D system, issues that remain completely unaddressed, despite decades of concerns from scientists and research leaders. First and most painfully, it's harder than ever to be a scientist. Graduate education can take up to a decade in the biological sciences, while multiple one-year terms as a postdoc have increasingly become the norm as faculty slots remain few and fleeting on university campuses. Salaries are poultry, particularly given the alternatives in technology and finance where the STEM education background of these scientists are highly prized. Physicists heading to Wall Street has been a theme since at least the 1980s, and yet salaries have still not equalized between the 2 fields. Then there's the absolute just incredible overhead that comes from running a research lab. Even once you get your first independent research grant, maybe at the age of 38 or even 42, that's not when you can devote yourself full-time to science.

Quite the opposite. Once you start on the grant train, you can't get off. Labs have to expand and grow, which means more and more time has to be devoted to the latest grant proposal in order to keep the lights on. It's not uncommon for some principal investigators to spend half their time or more on applying for grads and managing their labs rather than doing science. We have somehow decided to direct our brightest scientific minds to filling out paperwork and minutiae, and then we wonder why progress seems to have stalled in many areas of science and economics. The status quo has to change. One part of that is making grants less competitive, both by increasing funding, but also building an earlier sorting hat into science careers. People shouldn't find out whether they made it or not in science in their late 30s. Far too many scientists burnout far too late in their careers and earlier guidance on success is crucial for building a saner, healthier talent pipeline.

But then there's the trust that has eroded between science and administrative bureaucrats. Instead of trusting scientists to do their best work, we have increasingly burdened them with accounting requirements, which ensured that instead of thinking up experiments or analyzing data, they spend most of their time on management overhead. That's a perilous place to be. Yes, accountability matters, but when accountability is snuffing out the very activity we're trying to fund, something is very broken. Government funding orgs need to scale back and design simpler administrative systems that ensure that bureaucracy never eats up more than 10% of a scientist's productive hours.
Finally, we need to continually recommit to the notion that fundamental science research leads directly to America's prosperity. Every year, science funding budgets are scaled back in Congress because they don't affect "bread and butter issues." A quick look though at the history of the past two centuries would indicate just how completely asinine such a view is. The rapid improvement to quality of life came directly from science experiments and innovations that started in laboratories and garages. The government needs to improve the lives of people today for sure, but it can also simultaneously ensure that we are flooding the fountain of knowledge and ensuring that our country's future is well secured.

Maybe the government's budget this year will get approved, but I do know that many other countries aren't waiting. India, China, South Korea and others are heavily investing in the future of science and innovation, securing their future growth. The United States has a sclerotic and numbing research bureaucracy that increasingly doesn't fund the most compelling ideas. Competition is a flight away and global research leaders are seeing the mistakes we are making and are avoiding them. We've got to refuel our juggernaut and fast. The good news is that we can do it, and maybe with some more episodes of Securities, we can map the route forward.