Department of Scientific Efficiency

Can we make science better for scientists?

When an old growth tree falls in a forest, it’s fair to be sad about the loss of a giant, an ancient mass of trillions of cells that survived for hundreds to even thousands of years. Below is a kapok tree I encountered during field work in the Peruvian amazon, a giant I love, admire, and will mourn when it falls.

An old kapok tree deep in the Peruvian amazon

Yet, all living things must one day die, and every time a tree falls in the forest it clears the canopy for the next generation of trees to grow. That kapok tree itself grew in the gap of the canopy created by another old tree yielding the sky.

Our country’s institutions have many such old growth trees. Many folk on the left look on in horror as the incoming administration sharpens the blades of their axes, turns on the chainsaws, and proposes a Department of Government Efficiency. There are deep divides in our country and those deep divisions overlap naturally with how people feel about Elon Musk and Vivek Ramaswamy’s DOGE. I can’t heal those divisions, or at least I don’t feel like I can. Instead of refusing to accept the things I can’t change, I’ve spent my time and forecasting brain of late wondering what could happen as the axe of government efficiency strikes trees close to home.

What could government efficiency mean for science?

There’s ample attention paid to the potential harms of cutting funding/support for vaccines, peddling false or unproven claims, and more. I won’t comment on those (I may or may not agree, it’s irrelevant). The warnings of harm are the cries of many eager to save the canopy trees.

A lot of the conversation surrounding science, especially health science, in the upcoming Trump administration swirls around RFK Jr. and how much people love some of his views and how much other people hate his other views. I’m not seeing anybody getting convinced by those words, so I am not going to spend words defending or opposing RFK Jr.’s views, because views and actions are not the same. I’m eager to, even if just for an exercise, brainstorm how to build new growth in the inevitable gaps in the canopy, whether by this administration’s axe or another’s.

Can change & efficiency-gains catalyze improvements in science?

When I bemoan “science”, I’m resonating with leagues of current and former graduate students who entered our scientific enterprise wide-eyed with optimism and left (or stayed) jaded and cynical about the crusty, canopy trees of scientific institutions that make being a scientist painfully inefficient, nepotistic, and sometimes wrong. Our scientific institutions funded dangerous research because it answered to the lobbying of scientists, not the will of the people, and many scientists wrongfully excluded competing views on everything from public health policy to the origins of SARS-CoV-2. From the historical pains of being a scientist to contemporary harms during a likely science-caused and science-exacerbated (and, to a lesser extent, science-alleviated) pandemic, there are a lot of problems with science, and it’s okay to recognize that.

It’s from a place of love - a deep love of science and what it can be - that I hate the inefficiencies of science. Many of us, even those who voted for other presidential candidates, can’t help but feel a strange sense of satisfaction in our gut when somebody says they will take an axe to those old institutions. We feel this way because there are saplings of alternative ideas that might have a shot at growth.

The ecosystem of modern science is vast and full of anachronistic old growth that may be towering and impressive, yet unproductive if our goal is to maximize the net primary productivity of our crop. In order to strategize improvements to scientific efficiency, we have to study the physiology of our scientific ecosystem, understand the functions we want science to maximize, and trace the day-to-day enzymatic activities of scientists to find out where in this large metabolic pathway converting money to new tech we build up waste or experience rate-limiting steps.

The function of federal investment in science is to make new tech and boost US competitiveness. Scientific investment helps us make tech by funding scientists’ exploration and discovery of new methods, compositions of matter, algorithms, and more. These, in turn, help US competitiveness by giving our military an edge, boosting our economic activity by fostering the growth of companies that generate revenues and employ people, and develop a robust base of scientific human capital capable of not just future R&D wherever they go but also consulting policymakers on critical issues such as defense, public health, education, economics, and more.

Now, let’s track the life of a scientist, identify the inefficiencies and frustrations, and think concretely about what we’d need to change to make things more efficient.

SCIENCE FUNDING

Scientists are extremely dignified beggars. Our research is costly and most of us are not independently wealthy, so we have to beg people to fund our ideas.

Venture Capital

We can follow the orders of corporate R&D, but that R&D can often be slow and focused on expanding the moat of the corporation you work for when sometimes that corporation is, itself, an old growth tree that may not be as efficient as the next generation of company that could grow with disruptive innovations.

To fund disruptive innovations, we can beg private investors and tap into our impressive private capital markets. Most scientists don’t know how to do this, in part because most scientists aren’t rich, don’t know how to start companies or make pitches, and don’t have the right connections in business to set up the best teams and get their feet in the door. In some places like Silicon Valley, the connection between scientists and venture capital is stronger, but this is often due to legacies of extremely rich people who’ve stayed in the area and funded local scientists. If you grew up in Albuquerque, like me, you probably never interacted with a venture capital investor until your 30’s, if at all.

One way to improve scientific efficiency can focus on increasing the liquidity and upside potential of scientific ideas in venture capital marketplaces by creating programs to help scientists build startup teams, compress their big ideas into business models that can grow, and investing in scientists even if they live outside our VC hubs.

Elon Musk is buddies with Marc Andreesen of Andreesen & Horowitz, one of the biggest venture capital companies in the world. Accelerators like a16z have been doing the work of taking foundlings and converting them into disruptive industry, there are just a few inefficiencies in how well we point scientists in our educational pipeline towards these career options, and how much we support people to pursue these dreams, especially people who have good ideas but lack the wealth, health insurance, and other safety nets that can help them make the leap.

Industrial R&D

Our portfolio of federal science funding shouldn’t go all-in on private equity. While early-stage funding is key, late-stage funding and the R&D of established industry is also important for our competitiveness as sometimes we don’t need the Levy flights of disruptions but instead need the gradient ascent of optimization (sidebar: I know a lot of people like to use gradient descent to minimize loss, but I’m a glass-half-full kind of optimist who likes to think about improving function instead, so I buck convention and use gradient ascent).

We can support industrial R&D in several ways. First, we can teach young scientists about industrial R&D. Current academic pipelines focus most of grad students’ career attention & ambitions on the academy, and this career myopia underserves the majority of PhD students like myself who end up working outside of the academy.

Our federal science funding can incorporate industrial science opportunities in several ways, including (i) grad programs that include at least some time spent in industry, (ii) grants that provide matching funds for academic-industry collaborations, (iii) resources that help grad students (and not universities) have majority ownership over their IP so the innovators themselves can be incentivized to transfer it, (iv) specific support for scientific R&D that advances trade secrets, and so much more.

While our science funding does support commercial enterprise, it typically does so in an either/or way that funds industry or others, making non-industry research centers appear to be competitors of industry. We can remove this competitive inefficiency and find ways to better connect academia with industry. Not only will this increase the impact of our funding where industry will match taxpayer funds, but it will also increase the human capital of scientists by giving them more career options and helping them feel like part of a larger ecosystem.

Grants

Scientists hate grants. We depend on grants, but we absolutely hate them because most of our grants aren’t accepted so we waste a lot of time writing voluminous, carefully referenced, extravagantly detailed grants that could’ve been a phone call or email saying

“hey, yeah, I want to do the same things I’ve been doing all along and kind of follow my feet a bit. LMK if you wanna hear more. Science!!!”

When a grant gets rejected, scientists get dispirited. Making scientists depressed through repeated rejections after all their wasted effort is bad for scientific stoke & the human capital of scientists. When reviewers from another field poo-poo your idea without demonstrating an understanding of it, or when they make elitist claims about your irrelevance or your institution’s lack of infrastructure as reasons they can’t fund you, and more, it hurts science by hurting scientists. Our science review process in general is not very good, and while science is supposed to be an epistemological warzone with ground rules it’s still good to separate training and nurturing from live-ammo epistemological warfare that we will inevitably face.

Let me give you one example. I submitted a grant years ago, back when I was a postdoc, and I spent months preparing a perfect proposal, thinking through all the scientific technicalities and preparing a long list of documents from budgets to project management plans. I submitted the grant and prayed, but within a week I got a rejection notice: I forgot to put my project management plan in the “Supplemental Information” folder on their submission site (I put the project management plan in another folder! they had it!!!), and so they wouldn’t proceed with my grant.

This is the sort of experience that makes scientists hate grants, and this is just one of many I could share with you. Instead of being reflexively partisan and hoping Trump fails at everything, leaving our country and institutions in disarray, I encourage considering ways this catalyst of change bearing a great big axe might actually improve our lives. Maybe, if we put in the work and help, we can make things better.

We can do a lot of good for grants by having lotteries.

Don’t invite highly educated and busy people to waste their time until they have a very high probability of success. We can ensure a high probability of success if everybody can submit their name into a fixed number of Mission Buckets (e.g. Mission: Cure Cancer! Mission: Develop new Vaccines! Mission: Solve H5N1 Crisis!”). We’ll draw a bunch of names out at random and those selected from the Goblet of Scientific Fire/Stoke are invited write a grant supporting the mission of said bucket. Rather than rejecting grants due to project management plans not found in the Si folder, government officials acting like VC investors - trying to help them succeed knowing we will be investing. Just tell them where the SI folder is, and I guarantee they’ll politely move the file. Tell them what budgetary things they need to consider, and I’m sure they’ll budge. Two cruxes will be making reasonable budgets across projects as random draws have unknown budgetary needs. Two possible solutions: fix budgets and/or use a DARPA/ARPA-H/SBIR funding model where we start small, track progress, and allocate more to performers who are striking gold.

Is this perfect? Nope. Random draws might not pick the best candidates, but at least they avoid the wasted time and crushed souls.

Can this be part of a more perfect system? Yup. We can use the random-draw funding pool as a null model, and all alternative funding efforts can be contrasted to the random draw on many benchmarks. This can turn science-funding more explicitly into a rolling A/B testing apparatus for meta-science hypotheses, allowing us to continually evolve new ways of funding science instead of getting so horribly stuck in a meta-scientific funding rut as we have with our current institutions.

PUBLICATIONS

There are few things scientists love to hate more than journals.

We should not allow federal funds to pay for scientific ads like journal publications.

Will this destroy the journal system? Yes.

Will many of us either publicly or privately rejoice? Also yes.

Let me explain how horrible our journal system is. A scientist does science, often using taxpayer dollars, and writes up their result. They can optionally upload their result as a pre-print but ultimately our scientific institutions value journals, so most scientists are coerced by career incentives to submit their papers to journals.

Journals have editors who typically don’t know what they’re talking about but feel like they do, and those editors can block a paper for dumb reasons. Maybe you get by the editor because the editor is part of your old-boy’s network, maybe you got lucky and the editor thinks your paper’s title could generate revenues for the journal (not the best reason to advance), or maybe the head of the Wellcome Trust Jeremy Farrar pings the editor and kindly encourages they publish your work saying a lab origin of a virus from a lab funded by the Wellcome Trust is implausible.

In any case, if you get past the slimy editor’s desk, then you get peer reviewers. If you’re part of an old boys’ club, you just list the old boys as your reviewers. If, however, you are proposing something disruptive, such as the possibility that SARS-CoV-2 has a synthetic origin from a lab funded by the Wellcome Trust, good lord you are in for a ride and your paper will get rejected in the most hideous way possible, and not because the science is necessarily bad.

For example, our paper proposing a synthetic origin of SARS-CoV-2 was first rejected by an editor who didn’t know what they were talking about, then we made an appeal because we did know what we were talking about, then the editor in chief wisely stepped in and said this should be sent to peer review. The editor in chief sent the paper in front of a mixed bag of scientists, received some fair reviews from unbiased scientists and then received flak on social media and beyond from scientists who are invested in risky virological research. Eventually, the board of directors, which includes some people from the Wellcome Trust (as the Wellcome Trust funds the journal itself), unceremoniously fired the editor in chief for freely speaking his long-held political views, and then replaced the editor in chief with a guy who claimed they couldn’t find “unbiased” reviews, made some claims demonstrating he too didn’t know what he was talking about, and then rejected the paper.

That, folks, is our journal system. Don’t believe me? Subpoena eLife and find out.

That, folks, is why we should not spend taxpayer dollars on scientific ads like journals. We wouldn’t allow taxpayer funds to support boosting tweets of a preprint, why should we support journals? Do we have any federal oversight over journals we’re paying to ensure they’re complying with federal laws and not vulnerable to foreign influence like having the Wellcome Trust or possibly people associated with the Chinese government block American scientists from publishing evidence that might undermine these foreign interests?

We have no such oversight over journals and can never have assurances that anonymous peer-reviewers will be reviewing papers without conflicts of interest.

We should bring all scientific publications in-house. NIH, for example, oversees the National Library of Medicine. One could require all NIH-funded works be placed in federal repositories such as these (if nothing else, to ensure our taxpayers and others within the federal government can read the papers without a paywall). If we need review, we can have a comments section, or better yet someone can just write their own paper critiquing a paper they disagree with and other scientists can sit back with popcorn and think for themselves who they agree with. If industry needs somebody to provide unbiased perspectives on this tangle of diverse opinions accurately reflecting the diversity of scientific beliefs, they can hire a consultant.

Thankfully, we already have such a policy… kind of. Researchers are required to make federally-funded research publicly available. However, academic journals still cost $1000-$10,000 per article and, as I understand it, people can still use taxpayer funds to advertise their papers in these journals.

That won’t be necessary. We can improve scientific efficiency by not allowing taxpayers to fund journals, especially journals that do significant amount of business with our adversaries and may thus be incentivized to avoid publishing controversial articles that undermine their relations with authoritarians who may have created a virus that killed 1 million Americans.

Good lord. End journals once and for all, and scientists will rejoice. If need be, redirect the money from journals to mentorship or grad students.

MENTORSHIP & CAREER DEVELOPMENT

Becoming a scientist takes time. Academia does a decent job training our workforce to do science, and this training culminates in degrees like PhDs.

I was lucky enough to be a National Science Foundation Graduate Research Fellow, meaning NSF paid for the first few years of my graduate career. This made it easier for me to choose where I wanted to go and study what I wanted to study. I really like those programs.

We can even improve these programs by expanding ideas above. I’ve mentioned the concept of scientific human capital a few times because I think it’s important. Scientists are valuable people, they can do things others cannot, and the things scientists can do can revolutionize our world. A scientist is like a tree, where if you nurture it for long enough, it will bear fruit. If, however, you crush its soul, grow it in rooms with inadequate light, or fail to water it, then all the money you put into fertilizer won’t pay off.

We need to think holistically about scientific human capital, and that requires explicit attention to mentorship and career development. Our current system treats mentorship as an accidental byproduct of university business - professors get big grants, but because professors are so busy writing big grants they ask for extra funding to hire graduate students to do their work. Sometimes, we support entire PhD programs, and Princeton’s program for Quantitative and Computational Biology was one such program, funded by NIH (NIGMS, I believe?) to explicitly support mentorship.

In PhD programs, if you choose one advisor, you will be doing what the professor says and only allowed to use resources on projects the professor approves. If you choose another advisor, you may have independence to discover your own scientific voice and gain the independence you need to make it in the world. The former makes science (or science publications), the latter makes scientists. The former coddles someone along an enclosed path to an academic job where suddenly they may not know how to do truly original things. The latter throws a young scout into the wilds exposed to the elements and they must learn to hunt and gather or they will die. Neither is perfect, and also people are different - some people are hardy scouts, others are dutiful aides. What we really need is a recognition that different people want different things and we should design graduate funding and mentorship as a vehicle to help people find the things they want and help them become well-utilized, fully-actualized scientists.

Mentorship should not be an accident. If we’re going to mentor the next generation of scientists, let’s dedicate time and energy to learn how to do it right. Let’s not just learn about the people, let’s show them the many doors to alternative careers. The funding models above connecting scientists to VC opportunities, industry R&D, as well as clinicians, public health experts, government scientists, and more, can show young blood the many adventures of science and help them find their calling. Rather than hope they stumble upon their dream job, we should at least give them exposure to opportunities - and give these many employers exposure to students - to connect our marketplace of scientific human capital, ensure a more liquid exchange of opportunity, and accelerate our scientific innovation across all sectors of our society.

Concretely, this means investing in grad students not just with graduate research fellowships, but with explicit funding for young scientists to pursue scientific opportunities that appeal to them. Grad programs should have summer internships outside the academy to expose scientists to (1) government, (2) industry, (3) non-profits. These sectors of our society should anticipate a rotating cast of scientific talent and those that desire talent should make their workplaces awesome to recruit PhDs the day they graduate. Instead of having PhD students throwing their sad CVs around LinkedIn, coach them on how to make resumes for different jobs and let employers start to compete over grad students.

Like the lottery system as a null model for grants, we shouldn’t just lean into our skis with mentorship, we should study it closely, track the successes and failures, and continually maximize our ability to turn bright-eyed young scientists into dignified career professionals instead of the current inbred pipeline of coddled scientists who did their professors’ bidding and became professors themselves beside the mass of disillusioned PhDs thrown into the fields of employment without direction or mentorship. We reap what we sow in science, so let’s sow some damn good oats, help them grow to become damn good scientists, and keep our finger on the pulse of careers to provide them the most opportunity we can.

Needless to say, I also want this attitude of vocational cultivation, care and maintenance of human capital to extend to other sectors of society, including people without college degrees. The United States doesn’t have the most people in the world, so we stay competitive by having the best, brightest, most productive, happiest, and most utilized people in the world. Don’t just burn all the GPUs to accelerate AI, consider the humans and accelerate them, too.

DOGE: Tend to your Garden

I’ve put a lot of thought into how I’d improve science. As a scientist, I’m always looking for ways to improve things around me and excited to test new ideas.

I understand that many of my colleagues are afraid of the horrors the incoming administration could bring. As always, I really hope we end up with the most positive outcome despite the high uncertainty. I can’t control the world, I can’t control most of the US government, but if there will be a Department of Government Efficiency then as a scientist I hope I can have some impact on scientific efficiency. If we set aside our legitimate disagreements and focus on our core human, economic, and governmental needs, if we examine the functions of science and how these functions are frustrated by inefficiencies, we might be able to make our world just a little bit better.

If you’re excited about the incoming administration, you don’t need my words of encouragement.

If you’re afraid of the incoming administration, I offer you these words of purposeful consolation:

We can’t control the whole world, but science is our garden. Voltaire’s Candide lived in a world of absolute horror, causing him to eventually abandon Panglossian ideals, hopes and dreams about how it’s all good (or, more accurately, how it all keeps being more good even if there’s some bad somewhere). Candide, at the end of his odyssey of horrors, turns his focus to tending to his garden. Whatever uncertainty lies ahead, if someone does take an axe to old growth of scientific institutions, would you rather sit back, despair, and let it all happen, or, if you know it’s going to happen, would you rather work the soils, plant the seeds, and help the next generation of canopy trees be a little bit better, a little bit more efficient?

There’s a chance that change could be a good thing, especially if we help it.