The Lab Origin Menu

The COVID lab leak is back with a vengeance, but the story doesn't hold together when you try to lay out the recipe for a pandemic

The case for a laboratory origin of SARS-CoV-2 rests, at bottom, on a single coincidence. A novel coronavirus surfaced in the same city as a major coronavirus laboratory, and chance alone seems too thin to explain it. Everything else in the lab-leak case is an attempt to give that coincidence a mechanism. So, I want to take the mechanism seriously, more seriously than its proponents usually do, and ask what a laboratory origin would actually require.

I recently posted a series on the probability of a Lab Leak and the discussion that followed. That discussion has continued, offering a fuller view of the assertions being made to support a lab origin for COVID. What emerged was not one scenario. It was a menu.

The diner at the Lab Origin Restaurant must choose one item from each category. The challenge is ensuring the combinations are compatible.

At the Lab Origin Restaurant you build your own dish: a substrate, a method, a funding path, a location, and a mode of release. But the stations are not independent. What you pick at one constrains what can work at the next. Consider each category in turn.

1. Substrate

By substrate I mean the virus, or combination of viruses, from which SARS-CoV-2 is proposed to have been built. If the virus was made by inserting a furin cleavage site, the obvious question is: inserted into what?

The DEFUSE backbones are the usual answer. But the relevant question is whether any of the described backbones were close enough to SARS-CoV-2 to be transformed into the pandemic virus by the proposed modifications. They were not. DEFUSE may supply a suggestive research concept. It does not supply the starting virus.

RaTG13 is offered as the closest known relative. It is related to SARS-CoV-2, but it is not SARS-CoV-2 waiting for a cleavage site. Adding a short insertion at one location does not erase the genome-wide distance between the two viruses. That distance is the whole problem, and a single edit does not close it.

So the menu offers a third option, the secret ingredient: an undisclosed, unreported wildlife sample, closer to SARS-CoV-2 than anything in the public record, held quietly in a collection. This one cannot be dismissed by definition. But it is not evidence. It is the assertion that the missing substrate exists precisely where no one can check. It carries a different and heavier burden. It requires an account of why that virus was chosen, what was known about it, where it was held, who worked with it, and how it reached the first outbreak. None of that has been supplied. The secret ingredient is not a starting virus. It is a placeholder shaped like one.

And the supposed secrecy around it is even harder to explain. Investigators routinely report newly identified sarbecoviruses. Here, there is no record despite what would need to have been a major research commitment by WIV. What made this viral backbone so intriguing that WIV chose to build a major research effort around it and, given that, what about this fascinating new virus prompted investigators to keep it a secret?

The problem with these substrates becomes clear when we ask the lab to turn them into SARS-CoV-2.

2. Method

The pandemic itself provides a cautionary tale for any assertion that a substrate can be readily modified to create SARS-CoV-2 in a lab. The entire pandemic, 14 billion human infections, simply moved us from one variant of the virus to another, separated by 120 substitutions and still classified as SARS-CoV-2. A laboratory manipulation scenario would need to replicate something comparable in a lab.

Direct bioengineering can explain the deliberate addition of a feature such as a furin cleavage site, but only if the necessary substrate already exists. Laboratory passage can explain adaptation under selection, but only if the starting virus is already close enough and the selection environment is plausible. Recombination can explain why different parts of a genome have different histories, but it still requires real donor viruses.

Notice what every method has in common. None of them eliminates the substrate problem. They relocate it. Engineering and recombination both need something close enough to SARS-CoV-2 that it can get there with a handful of artful modifications. Passage requires something so close it is essentially just a variant of SARS-CoV-2.

No such precursors have been identified. The closest relative in WIV’s records, RaTG13, was never held as a live virus at all. It exists only as a genome sequence. So, the intended outcome of the recipe defines the ingredients. No identified substrate could possibly have served as the precursor for SARS-CoV-2. That means there was a secret ingredient.

How was it selected? Why was it chosen? Why was it kept secret from the moment of discovery? Proponents of a lab origin must answer all of those questions.

Then they need to explain who paid for it.

3. Funding

Funding has been at the center of the lab origins controversy. We know that NIH funded the EcoHealth Alliance for bat coronavirus surveillance, and some of that money reached WIV through a subaward. But money flowing from the US to China is tracked in fine detail. NIH had meticulous records of where it went. DEFUSE was something else entirely: a 2018 proposal to DARPA, not NIH, for a spillover-risk project whose logic was to identify and test viruses that nature might plausibly produce and that might cross into humans. DARPA declined it. So, one item in this column supplies money without the proposed modifications, and the other supplies the modifications without any money. Neither supplies both, and the story of a Fauci-approved DEFUSE requires both.

That creates problems for the secret funding of secret research on a secret virus scenario necessitated by items one and two on our menu. It is hard to square US subsidized funding scenario with clandestine research. But then who paid for it.

A Chinese military program implies a different logic entirely. A weapons-oriented program would favor a substrate with demonstrated human infectivity, known virulence, and operational predictability. That does not point to an unreported wildlife virus with no known human infectivity, unless that virus had already been characterized as unusually dangerous, in which case the missing characterization becomes one more thing the scenario needs to explain and cannot.

The mystery-money option, an unnamed secret source, is not an answer. It is the funding-column version of the secret ingredient: the required element asserted to exist exactly where it cannot be examined.

Then there is the interesting suggestion by some that the lab leak didn’t even come from WIV.

4. Location

Most Lab Leak scenarios focus on the Wuhan Institute of Virology. Until epidemiology suggests a location seven kilometers away. Then, miraculously, a new menu item appears. The Wuhan CDC. The record suggests this lab was actively collecting samples from wildlife for testing and it is near the center of the market cluster. But it introduces some serious problems for dining at the lab leak restaurant.

First, it requires discarding the rest of the order. The entire lab leak argument up to this point is based on events at the lab and infected lab workers. None of that can be included in the meal if we move to CDC where no capacity for bioengineering existed and viruses were not cultured. A whole new meal needs to be created from scratch, and no one has laid out this alternative menu.

The location column is also where a familiar objection lives, so it is worth addressing here. Ascertainment bias has become the standard explanation for why early cases clustered around the Huanan market. The argument runs like this: once authorities suspected the market, they looked for market-linked cases, so the cluster near the market merely reflects where investigators were looking. In some versions the claim is that a market link was built into the case definition. In others, that early cases with no market link were located closer to WIV.

This argument betrays a misunderstanding of ascertainment bias. Patients were not identified by wandering around the market neighborhood knocking on doors. They were found when they became sick, sought care, were evaluated by clinicians, and reported through medical channels. Geography shapes exposure, and it shapes investigation after the fact, but the initial identification of a novel pneumonia begins with illness and medical contact, not with investigators sampling neighborhoods or asking about shopping habits. In fact, the December case maps include both market-linked cases and cases with no known market link.

So, within Wuhan, the market cluster is real, and it points to the market. Ascertainment bias does not dissolve it. Hold that finding. There is a second, larger ascertainment question, but it belongs at the end, not here, because it is not about which building in Wuhan. It is about why any of this surfaced in Wuhan at all.

5. Mode of Release

An infected laboratory worker implies that a virus already capable of infecting humans was being grown and handled in a way that allowed exposure. Released samples imply a different pathway, the movement of biological material from a collection into the setting of the first outbreak. These are not interchangeable. They leave different evidentiary footprints, and in the versions on offer, neither footprint has been found. Lab-leak proponents have suggested as many as three infected workers, yet there was no identified community cluster around any of them, and the virus is said to have spread unnoticed until it surfaced at the market. A release with no trace, producing no cluster, is not a mechanism. It is the absence of one, asserted as if it were present.

The Meal

A lab origin scenario must have an item from each category. More importantly they must hang together.

You can’t pick and choose items simply because you don’t have proof they didn’t happen, while ignoring their logical inconsistency. And you can’t hide everything behind the lack of transparency of the Chinese government, insisting they seamlessly suppressed evidence of the substrate collection, the engineering records, the funding trail, the personnel medical files, and the sequence databases, across at least two institutions, sustained for years, and kept mutually consistent under adversarial international scrutiny.

The question is not whether a government would suppress information. Of course it would. Governments, including the Chinese government, do it constantly. But this would have required a concealment that identified in advance every record that would ever matter, in every institution, and scrubbed each into a state that stays consistent with all the others, for years, with no defector, no leaked draft, no inconsistent redaction, no seam. That is not a government hiding something. That is anticipatory, frictionless, essentially omniscient control over a distributed evidentiary record, and no state has ever demonstrated it. The tell is the frictionlessness. Real coverups leak, contradict themselves, and leave seams, because they are run by people under time pressure who do not know in advance what will turn out to matter. The coverup this case requires shows none of those signatures. Its very seamlessness is evidence about how likely it is, and the answer is not very.

And presumably, this impeccable cover-up comes from the same government that funded research on a bioweapon that was difficult to control, slow and inconsistent in its effects, readily transmitted back into the attacker’s own population, and with a case fatality rate of roughly one percent, killing primarily the elderly. Research done at a BSL-2 lab with sloppiness that allowed it to leak. It doesn’t add up.

In other words, there is no way to generate a logically cohesive lab origin scenario from this menu.

Maybe We Don’t Need the Menu

Recall the one fact the whole case is built on: the coincidence that a novel coronavirus first appeared in the same city as the laboratory. Set the menu aside and look at that fact directly. It was not a coincidence. It was close to inevitable.

The first detected cases of a new disease are almost never the first actual cases. They are the first cases to pass through the machinery of detection. For a novel infection to become visible, someone has to fall sick enough to seek care, the illness has to look unusual enough to trigger concern, clinicians have to collect the right specimens, known pathogens have to be ruled out, the samples have to reach a laboratory able to identify an unknown virus, and the result has to be recognized, reported, and connected to other cases. That chain is far more likely to be completed in a major city with advanced virology than in a rural county where bat coronaviruses actually circulate, or in the towns and transport routes between.

This is why the earlier ascertainment finding had to be split in two. Within Wuhan, the market cluster is real, because there the search was keyed to unexplained pneumonia and it still found the market. But the reason the outbreak surfaced in Wuhan rather than somewhere upstream is a different matter entirely, and it is genuine ascertainment bias. Wuhan is one of the few places in China where a novel coronavirus could be caught, sequenced, and recognized as novel.

A handful of severe respiratory infections in a rural county in late 2019 would have been logged, if logged at all, as ordinary winter pneumonia. In a relatively young, healthy, rural population, most cases would have looked more like the cold or flu. No one would have sampled them, sequenced them, or thought them unusual, because nothing in that setting could tell a new betacoronavirus from the seasonal background.

The proverbial lamp light used in the search for the viral car keys was not the Huanan market. It was WIV. Among the best equipped labs in China to rapidly identify a novel coronavirus.

We have seen this before, including with the two other coronaviruses that caused recent human outbreaks, SARS-CoV-1 and the MERS virus. The virus responsible for SARS was first identified in Hanoi even though the first cases were in China. Evidence indicates that MERS was circulating in camels for decades and caused a cluster of cases in Jordan months before it was detected in a patient in Jeddah, Saudi Arabia. But the most dramatic example of the disconnect between origin and detection is an entirely different zoonotic virus, HIV.

The pandemic strain of HIV now appears to have crossed from chimpanzees to humans in a remote corner of Cameroon around 1908. More than 70 years would pass before an explosion of immune deficiency among gay men halfway around the world drove virologists to find its cause.

This pattern holds for almost all zoonotic diseases, but lab origin proponents insist that we find a close match in wildlife to prove it. Meanwhile, they dismiss the market data with a wave of the hand as an artifact, an invention, or evidence of an entirely new laboratory source.

SARS was not recognized in Guangdong. MERS was reconstructed backward from Jeddah. Nobody concludes that HIV started in Los Angeles. In each case the place of first recognition was set by where the disease met the capacity to identify it, and the true origin was found later, somewhere upstream. The same restraint, applied honestly to Wuhan, is all the zoonotic account asks. Wuhan is where SARS-CoV-2 became visible. That is not the same as where it entered the human population, and treating the two as identical is the unexamined assumption on which the entire lab-leak edifice is built.

The Wuhan coincidence justified scrutiny. It did not supply a mechanism. Once the proposed mechanisms are separated and required to form a single causal chain, the public laboratory-origin case repeatedly substitutes an unknown precursor, an undocumented program, an untraced release, or an all-encompassing concealment for the evidence it lacks. Wuhan is where SARS-CoV-2 became visible. Until evidence connects that visibility to a laboratory process, it cannot be treated as proof that the virus began there.

As I have discussed elsewhere, all previous novel human infections have been zoonotic. Proponents of a lab origin for SARS-CoV-2 must submit their theories to a level of skepticism equal to that the demand of zoonotic origins. That means they must propose a scenario for generating and releasing the virus that fits all available evidence. They cannot simply pick and choose elements off the menu without regard to their ability to be assembled into a logically coherent version of reality.

References

Substrate

Zhou, P., et al. “A pneumonia outbreak associated with a new coronavirus of probable bat origin.” Nature 579 (2020): 270–273. https://www.nature.com/articles/s41586-020-2012-7 (RaTG13 as closest known relative, ~96% genome identity, described from sequence.)

Method

Zhou, P., et al. “A pneumonia outbreak associated with a new coronavirus of probable bat origin.” Nature 579 (2020): 270–273. https://www.nature.com/articles/s41586-020-2012-7 (Genome-wide distance between RaTG13 and SARS-CoV-2.)

Wu, F., et al. “A new coronavirus associated with human respiratory disease in China.” Nature 579 (2020): 265–269. https://www.nature.com/articles/s41586-020-2008-3 (Original SARS-CoV-2 genome.)

Andersen, K.G., et al. “The proximal origin of SARS-CoV-2.” Nature Medicine 26 (2020): 450–455. https://www.nature.com/articles/s41591-020-0820-9 (Furin cleavage site and the genomic features at issue.)

Morris, R. Pandemic-scale evolutionary constraints on the divergence of SARS-CoV-2. Preprint at https://doi.org/10.5281/zenodo.18353531 (2026).

Funding

Lerner, S., and M. Hvistendahl. “Leaked Grant Proposal Details High-Risk Coronavirus Research.” The Intercept, September 23, 2021. https://theintercept.com/2021/09/23/coronavirus-research-grant-darpa/ (DEFUSE proposal to DARPA, 2018, declined; plans to insert cleavage sites.)

Regalado, A. “Inside the risky bat-virus engineering that links America to Wuhan.” MIT Technology Review, June 29, 2021. https://www.technologyreview.com/2021/06/29/1027290/ (NIH/EcoHealth Alliance subaward to WIV.)

Location

Regalado, A. “Inside the risky bat-virus engineering that links America to Wuhan.” MIT Technology Review, June 29, 2021. https://www.technologyreview.com/2021/06/29/1027290/ (Shi Zhengli’s committee recommended BSL-2 for engineering bat coronaviruses; Baric and Ebright on the inadequacy of that containment.)

Worobey, M., et al. “The Huanan Seafood Wholesale Market in Wuhan was the early epicenter of the COVID-19 pandemic.” Science 377 (2022): 951–959. https://www.science.org/doi/10.1126/science.abp8715 (December 2019 cases, both linked and unlinked to the market, geographically centered on it.)

Maybe We Don’t Need the Menu

Centers for Disease Control and Prevention. “Pneumocystis Pneumonia — Los Angeles.” Morbidity and Mortality Weekly Report 30 (June 5, 1981): 250–252. (First recognized AIDS cases.)

Faria, N.R., et al. “The early spread and epidemic ignition of HIV-1 in human populations.” Science 346 (2014): 56–61. https://www.science.org/doi/10.1126/science.1256739 (HIV-1 pandemic origin, Cameroon, early 20th century.)

World Health Organization. “Severe Acute Respiratory Syndrome (SARS).” Disease Outbreak News, 2003. https://www.who.int/emergencies/disease-outbreak-news (SARS first identified in Hanoi by Carlo Urbani; origin in Guangdong.)

Zaki, A.M., et al. “Isolation of a Novel Coronavirus from a Man with Pneumonia in Saudi Arabia.” New England Journal of Medicine 367 (2012): 1814–1820. https://www.nejm.org/doi/full/10.1056/NEJMoa1211721 (MERS first identified in Jeddah; camel reservoir and Jordan cluster established subsequently.)