Reviving Patronage and Revolutionary Industrial Research

Note: this essay was originally written for the 2019 St. Gallen Symposium competition. (I didn’t win)


To evaluate past work or promise?

In 1933, the first group of Harvard Fellows, “all chosen on the basis of their exceptional promise, not their past work” [1], started their fellowship. The Fellowship was created because Harvard’s then President wanted to “provide an alternative [to Ph.D] path more suited to the encouragement of the rare and independent genius.” [2].

85 years later, 38 out of 39 Junior Fellows either hold a PhD or are enrolled in a PhD program [3] and are explicitly chosen on the basis of past work, rather than promise [4]. It seems clear that Harvard is satisfied with missing out on an occasional late-blooming genius or a brilliant nonconformist who cannot stand 10+ years of higher education in pursuit of a degree, while making sure that all Fellows are “good enough”.

However, the point I want to make is not about Harvard specifically. Any institution, given the opportunity and the time, will learn to prioritize the median quality of its members and projects while trading-off variance — even against its mission; but universities and almost all academic institutions are unique in the extent to which this tendency manifests itself.

Legibility, long-term outlook, and genius

It’s natural that universities want to hire the best and most brilliant scientists. In the past, they would be restricted to indirect and noisy signals. Today, academia can quantify the exact “scientific contribution” of any scientist via their objective h-index and the number of publications in top journals with orderly impact factors. This means that when universities choose whom to hire and whom to promote — and want to be objective about it — they rely heavily on these indicators. What this means when we think about it on a systemic level, is that the modern researcher’s goal is not merely to solve the great mystery of the universe or to discover a new natural phenomenon, but to answer the great mystery and publish their findings in a top journal within the next 5 years to get a postdoc (if they’re a graduate student), within the next 2-3 years to get another postdoc or the position of an Assistant Professor (if they’re a postdoc), or within the next 3-5 years (if they’re an Assistant Professor who wants to get promoted or to get tenure).

There are a few obvious problems with all of this. First, all surrogate indicators can be gamed in some ways. For example, you can salami slice the results of the experiments to maximize the number of publications and citations [5]. But more importantly, if you’re aiming at solving the grand mystery of the universe — however brilliant you are — there’s good chance it will take 10 or 20 years to make discernable progress on that mystery, and when the time to move up the ladder comes, you will have nothing to show, except for years of failures; and Nature is not a journal that regularly publishes negative results.

So Peter Higgs — the discoverer of Higgs Boson — ends up concluding that he wouldn’t be productive enough for today’s academic system [6]; David Deutsch — the inventor of quantum computation — says that his first grant on quantum computers would have “no chance” of being awarded today [7]; and four of the United States’ most accomplished biologists publish a paper titled “Rescuing US biomedical research from its systemic flaws”, calling for science agencies to “significantly increase the numbers and kinds of awards that emphasize originality and risk-taking” [8]; all this, while the average age of receiving first R01 — the most sought-after NIH grant — is now 44 [9].

Competitive displacement of institutions and timescale misalignment

There are two key reasons for why risk aversion grows over time. The first one is that organizations get burned [10]. When you take a chance on a person, they may turn out to be a genius or a trickster. When an organization is built on principles of trust, some crooks will inevitably learn to take advantage of it. Thus, the organization has an incentive to ask for a record of achievement and letters of recommendations, and to create procedures that would prevent crooks from taking advantage of it, which inevitably inconvenience and slow down honest people.

The second reason is misalignment of incentives — in particular, the misalignment of timescales. People saving for their retirement want to maximize their returns over 20 years, but the hired CEO of the company they invest in wants to maximize her impact over 5 years and then move on to the next venture. Society wants scientific breakthroughs that would take 20 years of work, but individual scientists need to publish on schedule to get promoted; likewise, grantmakers want to maximize the number of Nature papers they are responsible for (regardless of whether these papers are extraordinary or merely above average) while avoiding failure-bound projects.

Academia is uniquely afflicted by both of these problems. Firstly, since academia lacks the mechanism for competitive displacement, bloat accumulating over time and inevitably rising risk-aversion can grow without bound [11]. If a firm becomes inefficient, it collapses and is replaced by another one. If science becomes inefficient… it continues to take in money and people and, well, scientific progress slows down. Secondly, as I pointed out above, science is severely afflicted by the problem of misaligned timescales. Grantmakers’ planning horizons (note that I’m talking about specific individuals who make specific decisions, not abstract institutions that theoretically care about the long-term) are severely limited by their own career planning horizons and by their understanding of what it takes to work on fundamental problems with little short-term payoff.

The oft-repeated suggestion is to make grant-makers fund more high-risk projects. This is the choice of the four esteemed biologists I quoted previously, but it is not likely to bring significant change. For example, the European Research Council already does some targeted high-risk research funding. The results for 2017 are out now: 79% of high-risk projects achieved “a major scientific advance” or “a scientific breakthrough” [12, 13]. This seems extraordinary, until you realize that if these projects were high-risk, the vast majority of them would have failed completely and only a few of them would have achieved major advances. Which, in turn, implies that the “high-risk” projects the European Research Council funds are nothing out of the ordinary.

Taking responsibility for progress

My proposal is simple: take responsibility for progress. It is time for corporations and entrepreneurs to recognize that they can no longer rely on governments to fund and on academia to conduct fundamental research. Instead of doing translational research and simply bringing academia’s fruits to market, they have to become bolder and take responsibility for what the future will look like, fund fundamental research, and bring to life a new vision for philanthropy.

Historically, it was often individual patrons who were responsible for supporting genius artists and inventors; it was various princes, dukes, and counts who provided tutelage and financing to scientists and iconoclasts like Spinoza, Kepler, Descartes, Galileo, and endless others; it was the capitalists and financiers who founded universities and who are responsible for some of the grandest collections of art; and finally, it is two giant corporations — Xerox and AT&T — that are responsible for the two laboratories – Xerox PARC and Bell Labs – that spearheaded the computer revolution and produced numerous invaluable contributions to physics, computer science, and engineering. Bell Labs’ researchers developed first transistors and lasers, information theory, Unix operating system, C and C++ programming languages; all in all, 9 Nobel prizes were awarded for work done at it. Xerox PARC’s researchers developed modern desktop computers, laser printers, graphical user interface, the computer mouse, object-oriented programming.

Without these industrial laboratories, the state of technology would have been very different today. But what does modern industrial research look like? Every large corporation seems to have a research division. Yet the research they carry out is almost always short-term and applied [13]. Corporations have ceded the responsibility for fundamental science to government and academia. It is time to take this responsibility back and to create radical new forms of doing and funding science.

What radical ideas do I propose? They are not radical in a sense that they haven’t been tried before. They’re radical for the simple fact of them being out of fashion today, even though they have proven their worth in the past.

Today, philanthropists donate to risk-averse universities that then redistribute the funds. But nothing really prevents them from becoming the patrons of specific scientists they believe in. And yet, it’s unheard of.

Today, philanthropists create foundations that give short-term grants, simply extending the broken academia system, instead of starting long-term focused research labs. And even when corporations and universities do “fundamental” research for many decades, sometimes, a dozen engineers working on their problem for several months can outsmart all of them, as DeepMind recently showed with their protein structure prediction software [14].

My two propositions

Capital needs to again be used to fund individual mavericks and iconoclasts; and to fund them on the promise of greatness, instead of the proof of accomplishment. Capital needs to again be used to create bastions of intellectual freedom, instead of further buttressing the shortcomings of academia. The world needs the revival of the idea of patronage of scientists. And the world needs new Bell Labs. Here’s how to do it.

First of all, philanthropists should realize the value of long-term grants. However generous the funding is, if you need to renew the grant every three years, this effectively means you have to publish on schedule. Granting 10-year or 20-year grants — effectively becoming a patron — should become normal again. The reason this does not happen today is because we don’t have enough trust in people. When you give a 20-year grant, there is always a chance that the person works for the next few years and then decides to just treat it a sinecure for the rest of their career. The prospect of this happening is so painful that today nobody gives long-term grants at all. Tenure — the only thing somewhat resembling them — is given after a months-long process that bears more semblance to the recruitment of a high-level diplomat than the hiring decision for an academic position [15]. And scientists still have to fight for grants — and therefore, work on incremental improvements — after getting tenure.

The natural question is then, how to find the scientists to give 20-year grants to? But this is relatively straightforward: if you already give short-term grants, you can simply pick the scientists with the biggest promise and offer them a 20-year grant; and if you don’t, the search will be more difficult, but in no way is it impossible. After all, princes, dukes, and counts were indeed able to find them.

Recreating the visions of Xerox PARC and Bell Labs will require figuring out the right organizational and incentive structure, securing long-term funding, convincing established scientists that it is a good idea to move out of their universities, discovering young scientists who would blossom in the lab, hiring operational staff… and also finding your own Bob Taylor whose vision to build the lab around. It will be very hard to get all of these right, but again — we know that it’s possible and we’ve seen the potential that lies behind doing it.

So, for now, my main hope is to make the idea itself seem realistic enough for someone to pick up the banner and to finally bring it to life.


Thanks to Katya Chegaeva, Guy Wilson, Anastasia Kuptsova, Brian Timar, Masha Taktasheva


Afterword: ok I don’t really know if 20 year grants are a good thing. There’s a good chance they’re not. Maybe the entire history of patronage is fake. Maybe all the books about Bell Labs are wrong. I do still believe someone should try to build another Bell Labs (Microsoft Research and HHMI Janelia are probably the closest to that existing labs). And I do still believe someone should at least try giving 20 year grants.


Bibliography

[1] Michael W. Miller. An Academic Free Lunch: A Look at the Society of Fellows (perma). The Harvard Crimson

[2] William Foote Whyte. Participant Observer: An Autobiography

[3] Senior & Junior Fellows (current academic year) (perma). Harvard Society of Fellows

[4] David Ottenheimer. A Conversation with Kapil Ramachandran, Johns Hopkins Graduate Student and Incoming Member of the Harvard Society of Fellows (perma). Biomedical Odyssey

[5] Vesna Šupak Smolčić. Salami publication: definitions and examples (perma). Biochemia Medica

[6] Decca Aitkenhead. Peter Higgs: I wouldn’t be productive enough for today’s academic system (perma). The Guardian

[7] David Deutsch. I spoke recently to a member of the panel that awarded me my first research grant on quantum computers (1985). He said it was a close-run thing. I asked if I’d have got it under today’s criteria. He said: no chance; basically I could tick none of the boxes. (perma). Twitter

[8] Bruce Alberts, Marc W. Kirschner, Shirley Tilghman, and Harold Varmus. Rescuing US biomedical research from its systemic flaws (perma). PNAS

[9] Michael Levitt and Jonathan M. Levitt. Future of fundamental discovery in US biomedical research (perma). PNAS

[10] CatCube. I have a hypothesis that a lot of the useless bureaucracy isn’t a function of government/private, or size of organization, it’s time. As an organization gets older, it’s gotten burned more times, and created more rules to deal with it. (perma. Slate Star Codex

[11] Michael Nielsen. New essay from @patrickc and myself, arguing that science has suffered from greatly diminishing returns over the past century (perma). Twitter

[12] Inga Vesper. Europe’s top science funder shows high-risk research pays off (perma). Nature

[13] Michael Nielsen. A European Research Council report suggests 79% of projects they fund “achieved a major scientific advance”, & only 1% make no contribution. Also, that they fund mostly “high risk” work (perma). Twitter

[13] Simon DeDeo. OK: following, my thoughts on Google Brain and similar institutions. Disclaimer: this is my opinion, as someone who doesn’t need their funding, and has built a career without needing to flatter them. (An anti-disclaimer, if you will.) (perma). Twitter

[14] Mohammed AlQuraishi. AlphaFold @ CASP13: “What just happened?” (perma). Some Thoughts on a Mysterious Universe

[15] Jeff Havig. I was explaining the timeline for a typical academic tenure track hire to someone not in academia the other day, and they were completely flabbergasted, so here it is for those that are unfamiliar. This is specifically for an R1 institution. Others may deviate significantly. 1/ (perma). Twitter


You can send your comments to me by

If you enjoy reading my site, you can subscribe by email:

View previous campaigns.