Publish and Perish

Coronavirus gain-of-function research comes from playing the "Science Game"

There is a growing consensus in the scientific community that the jury is still out on the origin of COVID-19. I refer to this recent editorial in Science calling for further investigation: “We must take hypotheses about both natural and laboratory spillovers seriously until we have sufficient data.”

In the past week, this shift in scientific consensus has been discussed in MIT Technology Review, and The Washington Post, and many others. In-depth reports by New York Magazine have profiled both the letter in Science and offered a detailed investigation into the conceivability of the lab leak hypothesis. It is immensely unfortunate this issue became politicized.1 I want to be clear from the beginning that the only conclusion here is that of doubt. There is no direct evidence that COVID-19 escaped from the Wuhan Institute of Virology. There is no direct evidence for a zoonotic jump from animal to human either.

Of course, one can argue about indirect evidence. Lab leaks have happened before: the evidence is pretty convincing that the 1977-78 flu was not natural, but an escape from a lab (it also started in China) and there are a number of other less-well-known lab leaks from other countries.

COVID-19 appeared with an already high human-transmissibility rate mysteriously in the middle of a city far from any bats. A city that contains, just a few miles away from the first major outbreaks, a virology institute known precisely for gain-of-function research on bat coronaviruses that make them more transmissible in humans. Nature takes many tries to get something right, which often leaves evidence of animal-to-human transmission, and no one has been able to trace the zoonotic jump despite massive efforts. The closest known virus in genetic sequence to COVID-19, RaTG13, was actively being worked on at the institute. It is very possible this work was in terms of gain-of-function to make the virus more human transmissible (by scientists who had been trained to splice spike proteins in a way which leaves no trace, called “no-see-um” techniques).

Yet these indirect matters are all debatable due to the iron dome the Chinese government has laid down. So the jury is still out. This universe does have its coincidences. It’s still entirely possible that either the wildlife trade or some other simmering but unfound contact between humans and bats is to blame. The only consensus is no consensus. We may never know.

Yet regardless of which origin hypothesis is true, one result has to be a re-examination of gain-of-function research. Because COVID-19 very much could have come from a lab.

So it’s worth asking: Why did these scientists even do gain-of-function research? Why try to engineer a zoonotic jump?

What you’re going to see over the next few weeks is the myth of the mad scientist rearing its head. Baseless conspiracy theories will run wild, like the American government did it on purpose, since the Wuhan Institute of Virology was part of a 3.7-million-dollar NIH grant to study bat coronaviruses, or that the Chinese government released it on their own citizens. These aren’t true.

But people will indeed say that scientists did it because they “were so preoccupied with whether or not they could they didn't stop to think if they should!” Biologists will be accused of playing God. And there is some truth to that. Frankly there is hubris in storing something in a freezer that could kill millions. Yet what commentators and media will miss is the hidden reason why biologists do gain-of-function research, which has to do with the broken incentive structures of science.

If you ask a gain-of-function proponent, they will say that by creating viruses that might emerge in nature, you get to understand zoonotic jumps from animals to humans better and possibly prevent them. Specifically, you get a head-start on developing vaccines for them. This possibility of curing future diseases might be true in some cases. But Pfizer’s COVID-19 vaccine was developed in a few hours back in January 2020. So, uh, what was the reason again?

Oh right, it’s a monetary and career reason, not a scientific one. Scientists are economically incentivized to switch to gain-of-function from merely studying and preventing existing viruses. It’s driven by the “publish or perish” nature of academia (with gain-of-function research this edges closer to “publish and perish”).

Imagine science as an advancing filopodia, searching around for resources. When you have a lab, you need grant money. Not just for yourself, but for the postdoctoral researchers and PhDs who depend on you for their livelihoods. The biology labs creating new viruses are often bringing in multi-million dollar grants. They are big machines that need to keep moving.

So lab leaders are forced to play the Science Game™. It’s a thing subtly different from science. Frankly, much of what what goes on in academia is really the Science Game™. It’s a term stolen from my recently-published novel, The Revelations, which takes place in the world of consciousness research. In the novel, a scientist thinks:

Of late he feels like all the activity of himself and his peers is just playing the Science Game: varying some variable with infinite degrees of freedom and then throwing statistics at it until you get that reportable p-value and write up a narrative short story around it.

Think of it like grasping a dial, and each time you turn it slightly you produce a unique scientific publication. Such repeatable mechanisms for scientific papers are the dials everyone wants. Playing the Science Game™ means asking a question with a slightly different methodology each time, maybe throwing in a slightly different statistical analysis. When you’re done with all those variations, just go back and vary the original question a little bit. Publications galore.

Examples abound. Why is Karl Friston the most-cited working neuroscientist? Because he came up with the most popular way to analyze fMRI data, providing a neat mathematical package to do so. And when most of us neuroscientists look at a hulking fMRI machine, we don’t see a multi-million dollar medical device. We barely see something physical. To most of us it’s mainly a handy dial for cranking out papers with. Stick people in for [insert task] and publish each new pretty image. This time it was this area! At each crank of the dial Friston gets cited, which is why he has over 250,000 citations. It’s a pattern you’ll find in many fields; frankly, there is more reward in science for coming up with new methodologies than new discoveries.

Every scientist personally knows about the Science Game™. Even if they don’t say anything publicly, they show it in their whispered criticism about some colleague’s latest cranked out paper. Or they show it in memes like:

It may not be an exaggeration to say a scientific career is determined by your initial choice of what dials you’re going to vary. Perhaps the strongest incentive is to develop clever tricks that let you play the Science Game™ forever. This means finding a dial that never stops turning. As James Carse writes in Finite and Infinite Games:

There are at least two kinds of games: finite and infinite. A finite game is played for the purpose of winning, an infinite game for the purpose of continuing the play.

Science should at least strive to be a finite game. Pick a problem and try to solve it. If you’re hitting consistently diminishing returns, move on to something new. But all the incentive structures go the other way.

Due to these incentive structures, what you really want as a scientist is a perpetual science machine. And there is indeed a way to engineer such a thing. Simply invent the very entities you’re studying. For the filopodia this is like identifying a source of infinite calories. A lot of research falls into this category, especially in fields like computational modeling wherein you can develop a model, study it, change the model, study it again, and so on.

Most of the time the outputs of perpetual science machines are harmless and merely spawn a lot of scientific papers. This is good for the owners of the science machines, although it can sometimes have the negative externality of decreasing the signal-to-noise ratio for everyone else.2

New advances in biology, like the ability to splice, change, and engineer, have made creating the very entities you study much more tempting as a biologist. Gain-of-function research3 is merely one subfield of this. But it’s actually truly dangerous.

Let’s examine the incentives. In virology, there are only so many dials—only so many natural viruses. And each is a source of competition, as famous labs make claims to various viruses to study and monopolize them by beating others to publication. The big excitement is in finding a new virus, mapping the genome, figuring out its function and transmissibility, comparing to other viruses, etc. But that’s all a finite game. Viruses are hard to find, and eventually a virus is pretty well-understood. Or at least, as understood as it’s going to get with contemporary methods. The bang for your scientific buck in terms of new papers in high-impact journals and new funding begins decreasing.

In comparison, gain-of-function research allows you to create new viruses to develop new vaccines for. Need a paper? Create a virus! Your scope of research has made the jump from actual evolved viruses to the space of all biologically-possible viruses. What a nice airy space to multiply in. Do you see the temptation?

If you don’t believe me, let’s examine that 2019 3.7 million dollar grant from the NIH, some of which went to the Wuhan Institute of Virology. Its innocuous title “Understanding the Risk of Bat Coronavirus Emergencefirst states that:

In a previous R01 we found that bats in southern China harbor an extraordinary diversity of SARSr-CoVs [coronaviruses]. . . We found that people living close to bat habitats are the primary risk groups for spillover, that at one site diverse SARSr-CoVs exist that contain every genetic element of the SARS-CoV genome, and identified serological evidence of human exposure among people living nearby. These findings have led to 18 published peer-reviewed papers, including two papers in Nature, and a review in Cell. Yet salient questions remain on the origin, diversity, capacity to cause illness, and risk of spillover of these viruses. In this R01 renewal we will address these issues through 3 specific aims…

What is Aim #3?

We will sequence receptor binding domains (spike proteins) to identify viruses with the highest potential for spillover which we will include in our experimental investigations. . . In vitro and in vivo characterization of SARSr-CoV spillover risk, coupled with spatial and phylogenetic analyses to identify the regions and viruses of public health concern. . .

It’s tame language, but what this is saying is that we have 18 citations, there are no more actual zoonotic jumps to study, but, you know, “salient questions remain.” So let’s just take the viruses we have and start studying ways to manipulate them, turning that dial, maybe passing it back and forth between human cells and animal ones, and just see what happens.4

Oh play me that Science Game™! Play it as the ship goes down.

All to say: scientists create dangerous synthetic viruses to achieve “high-impact” scientific output. I’m not saying this is the only reason, but it is a significant reason. In my mind, likely the main reason. Ethical and technical arguments are the ego of science, but the id is always funding and prestige.

In many fields this behavior isn’t much of a problem, but in virology playing the Science Game™ turns out to have negative externalities like potential mass death at a global scale. Other fields like artificial intelligence, high-energy particle collision experiments, attempts at alien communication, and more, may one day pose similar risks due to the run-away prestige games scientists play. Science doesn’t naturally have breaks. It’s up to people outside of it to give it some.

None of this is an excuse. But it does offer a reason different than hubris or outright malevolence for gain-of-function research. In this alternative view scientists are merely humans, pushed and pulled by the incentive structures laid down by governments, academic bureaucracies, and their own peers. When you feel like a little matchstick doll strung up by wires far beyond you, it’s easy to forget that people like you split the atom. Science has never been a game. But treating it like one is a risk. Maybe even, one unlucky day, an existential one.


You can read more about the politics of this debate here.


A perpetual science machine where you get to invent what you study looks a lot like the scientific equivalent of Dan Dennett’s criticism of much analytic philosophy as “Chmess.”


Note that some people use “gain of function” to mean solely the creation of new viruses via serial passage between animal and human cells. I mean it as broadly as possible: whenever you fiddle around with creating new viruses or new biological diseases, then studying or developing vaccines for them.


Since the publication of this essay, the NIH has officially gone on record claiming they never funded any gain-of-function research on coronaviruses. I feel like they are quibbling over what counts as “gain of function” since they clearly say in this grant that they will do things like “We will use S protein sequence data, infectious clone technology, in vitro and in vivo infection experiments and analysis of receptor binding to test the hypothesis that % divergence thresholds in S protein sequences predict spillover potential.” S protein sequences are spike proteins. It seems to me very unclear how you can predict spillover potential from doing in vivo experiments to vary S protein sequences, without, you know, changing the virus. I mean maybe there’s something I’m not understanding? How could that not count as gain-of-function? I’m open to explanations to someone who has been at a molecular lab bench more recently than myself.