Lab leaks and why expertise matters
Of late, I’ve been trying to understand the spike in enthusiasm for the so-called “lab leak” claim about the origin of SARS-CoV‑2, the virus that causes COVID-19. To me, and as far as I can tell, to the majority of people with the expertise and training to judge these things, a lab escape is far less plausible than a zoonotic spillover from some wild bat population, or some intermediate species which picked up the virus from bats and then spread it to humans.
Zoonotic spillover is how SARS and MERS became epidemics, and so its pretty reasonable to expect the same thing happened with COVID-19. It’s also how we get Ebola, Lyme disease, rabies, West Nile, zika, and a bunch of other diseases that remain relevant to modern life. We can’t always trace the exact site or wild population where that spillover happened. So it’s possible we might never find the source of SARS-CoV‑2, or that it might take a long time.
One of the things that seems to plague these discussions is that non-scientists’ expectations about how science works are often not how science actually works. So people looking for signs of something nefarious at the Wuhan Institute for Virology will latch onto some nubbin, from which they spin up a grand theory of how a lab escape could’ve happened. But it turns out to be built on sand, because of things most people don’t know and usually have no reason at all to know.

For instance, today I got into an interesting discussion with the New Yorker’s James Surowiecki. We were discussing a passage from Lindsay Beyerstein’s excellent New Republic piece. She points out that most people envisioning a “lab leak” picture researchers holding giant tubes full of virus, any one of which might slip and unleash the Apocalypse. Beyerstein writes:
the popular version of the lab leak theory that posits that Covid was taken from nature and escaped in its wild form. The problem with that scenario, [Dr. Angela Rasmussen] told me, is that a swab from a bat contains very little infectious virus. Each bat weighs less than half an ounce, and each sample is basically a Q‑tip swiped briefly over a bat’s mouth or anus. These samples are stored in vials in the freezer; they’re not likely to spill or leak, the way disaster movies have primed us to suppose.
Surowiecki replied asking about claims that WIV may have kept wild bats on site, which might have more readily spread viruses than an errant cotton swab. It’s a claim that’s circulated in the more conspiratorial parts of the lab leak discourse, but is hindered by the fact that there’s no evidence beyond a video of unclear provenance and some ambiguous quotes in popular press accounts to suggest that there ever were bats held at the facility.
Now, I’ve never done research in China, nor have I tried to keep bats in a lab, but I have been involved in academic animal research, catching mice in the field and releasing them after brief measurements. I know that the jump from that sort of research to caring for captive animals is an absolutely enormous leap from a practical standpoint and in terms of administrative burden. Is it impossible that WIV did that? No, not impossible. But it would be a heavy lift, and should leave some pretty clear trails to follow, none of which exist.
Other than a few seconds of footage of a bat (from the wrong family) being handled in a lab somewhere (not necessarily at WIV), advocates point to a pair of quotes, which Surowiecki cited, from research collaborators from the US who are claimed to have waffled on whether WIV might have worked with captive bats. Here he cites Peter Daszak, a US researcher who co-authored grants with WIV staff, receiving funding from US agencies.
Also, I don’t understand why Daszak tweeted of WIV “They DO NOT have live or dead bats in them,” only to later totally back off that statement.
— James Surowiecki (@JamesSurowiecki) June 29, 2021
Surowiecki notes that Daszak later declined to rule out the possibility that WIV had any bats, suggesting that Daszak was backtracking.
I looked around, and found a fuller version of Daszak’s first quote, which limits his claim significantly. He said, “This piece describes work I’m the lead on and labs I’ve collaborated with for 15 years. They DO NOT have live or dead bats in them. There is no evidence anywhere that this happened.”
To me, that seemed significantly different. But Surowiecki answered, “I don’t get it. The WIV is one of those labs that he’s collaborated with, and he issued a blanket denial that it had live or dead bats.”
And that makes sense! It seems like Daszak first says the lab isn’t doing this work, then says “we didn’t ask them if they had bats.”
But to me, and to an academic researcher like Daszak, WIV is not “a lab.” It contains probably dozens of different labs, most doing nothing of interest to Daszak. It’d be like suggesting that because Daszak works with NIH researchers, he must know all about every animal colony NIH maintains. Daszak commented on the labs he collaborates with, which would also be the labs most focused on sarbecoviruses like SARS-CoV‑2. Rather than being self-contradicting, Daszak’s comments make total sense in this light: he’s saying that the work he knows about doesn’t involve live bats, but he can’t say there are no live bats anywhere in the entire, large research center. Contextual knowledge changes our interpretation, and thus shapes how we weight the likelihood of a secret bat colony.
A similar thing happened when we talked about the practicalities of actually keeping such a colony, and keeping it secret. I could certainly imagine it being practical for a field researcher to grab a mouse and bring it home for the weekend, but I also know it would go against their training and all sorts of standards for animal care and use. It would be against all sorts of rules, some of which may be different in China, but others are enforced by US funding agencies (which would hold foreign researchers to similar standards as US facilities), and are enforced by publishers as well. There’s no reason for a non-scientists to know those details, but it’s a firm barrier to some sort of ad hoc bat zoo at WIV. They want to publish in Science, Cell, and Nature, and they can’t (regardless of any regulatory differences between the US and China) if they don’t get ethical review for that research, and document that review and the conditions of the animals’ care at the time of publication. If they were working with captive bats, we’d see evidence in their published research, and likely in regulatory paperwork submitted with NIH and other funding agencies in the US, if not in China.
On top of the bureaucratic barriers to this work, Beyerstein tweeted a link to this interesting paper on growing interest in bat research, and the challenges of keeping a colony, especially of the insect-eating bats that are host to SARS-like viruses. Unlike a rat that you catch and bring home in a cage, a bat that you bring to the lab needs a lot of room to fly, and often requires elaborate work to get it to feed. If they maintained any sort of colony of horseshoe bats, that would probably have merited its own research publication. Maybe multiple papers. And it’d require maintaining a substantial facility and staff to care for the bats, maybe the size of a netted-in soccer field or more. Not something easy to hide. They’d probably consult with other people outside of China who have done similar work. There would be a trail of evidence in research papers, construction, and conversations with colleagues outside of China.
The few quotes that people cite as evidence some captive bats may have been kept at WIV all suggest something extremely casual, and there’s just no sign that you could run such an operation at all casually. Here’s what it takes to feed a colony of insect-eating bats:
Again, no one who isn’t a bat biologist has any reason to know all of that. How different countries, funders, and publishers regulate animal care and use, and what biologists have to do to maintain a research colony, are miles from most people’s daily experience. Exactly what academics consider “a lab” is irrelevant to most people’s lives. There’s absolutely no reason for anyone outside the field to know these details. But each of them shapes our intuitions, and the upshot is that ideas which seem implausible to the scientists closest to the question can seem much more plausible, indeed may seem most likely, to those farther away.
Lab leak advocate Alina Chan inadvertently made this point a few days ago when she wrote this:
One thing I’ve noticed is that experts in the hard sciences (mathematics, CS, physics) afaik are more likely to think that SARS-CoV‑2 has lab origins.
— Alina Chan (@Ayjchan) June 27, 2021
But more experts in the life sciences afaik seem to think that a natural origin is more likely.
Could it be a matter of priors?
It is about priors. It’s incredibly telling that people with informed priors think a lab leak is extremely unlikely and that zoonosis is most likely, while those who know less about the murky day-to-day of biological research find the lab leak intuitively likely. Even at a different level of abstraction, biologists see these issues through different lenses. Biologists know that there are a lot of bats (about 1200–1500 species in the world, many undescribed), and that bats are good at hiding but travel all over. They also know that bats have a lot of viruses and that those viruses are good at hiding, and that there are a ton of ways a virus could leap to farmed animals, or animals that are caught live in the wild and brought to markets in big cities like Wuhan. That’s a causal chain that makes sense if you’ve studied biodiversity or tried sampling these things in the wild. At the same time, it all can easily seem incredibly tenuous to people who rarely see bats and don’t realize just how little we really know about the biodiversity all around us. But human error? We all see that every day.