This market resolves once we have a definitive answer to this question. (i.e. "I've looked at all notable evidence presented by both sides and have upwards of 98% confidence that a certain conclusion is correct, and it doesn't seem likely that any further relevant evidence will be forthcoming any time soon.")
This will likely not occur until many years after Covid is no longer a subject of active political contention, motivations for various actors to distort or hide inconvenient evidence have died down, and a scientific consensus has emerged on the subject. For exactly when it will resolve, see /IsaacKing/when-will-the-covid-lab-leak-market
I will be conferring with the community extensively before resolving this market, to ensure I haven't missed anything and aren't being overconfident in one direction or another. As some additional assurance, see /IsaacKing/will-my-resolution-of-the-covid19-l
(For comparison, the level of evidence in favor of anthropogenic climate change would be sufficient, despite the existence of a few doubts here and there.)
If we never reach a point where I can safely be that confident either way, it'll remain open indefinitely. (And Manifold lends you your mana back after a few months, so this doesn't negatively impact you.)
"Come from a laboratory" includes both an accidental lab leak and an intentional release. It also counts if COVID was found in the wild, taken to a lab for study, and then escaped from that lab without any modification. It just needs to have actually been "in the lab" in a meaningful way. A lab worker who was out collecting samples and got contaminated in the wild doesn't count, but it does count if they got contaminated later from a sample that was supposed to be safely contained.
In the event of multiple progenitors, this market resolves YES only if the lab leak was plausibly responsible for the worldwide pandemic. It won't count if the pandemic primarily came from natural sources and then there was also a lab leak that only infected a few people.
I won't bet in this market.
1,000People are also trading
@BrendanFinan In a way, there has been.
Marty Makary is FDA Commissioner. Shouldn't he be an insider at least when it comes to the outlines of what the lab leak case is for the intel community at the moment?
Instead, he's going around on podcasts recently and saying his smoking gun evidence is that there was a patent application for a COVID-19 vaccine filed by someone in China in February 2020. There were many patent applications for COVID-19 vaccines filed all around the world in February 2020 and even January 2020. See the figure from WIPO here.
So, more evidence if anyone needed any that the only evidence that likely remains secret in the USA is evidence of some half-baked lab leak theories flying around 2020-2021.
With the caveat that I haven't watched it all yet, this is a pretty accessible overview focusing on the quality of the underlying arguments for lab leak that have come and gone over the years: https://www.youtube.com/watch?v=ra0WKNhQZ0U
I'd make some different editorial choices in terms of what content to focus on if telling a similar story. Specifically, I'd discuss the influential scientists and members of the media who have essentially repeated, since roughly spring 2021, that it's impossible to conclude anything at all. Folks whose contributions as a rule consist of inaccurately describing the available data as being totally uninformative, and who never update their beliefs when specific data they requested is published and doesn't point to lab leak.
I spotted a couple errors; I'm sure the subjects of the video will latch on to one or two of them to undermine the whole thing for people who want to believe, confident that audiences at X and so on won't watch for themselves and fact check to see if it basically checks out.
@LukeShadwell Still confused what's causing the drop. I personally haven't seen much new in favor of zoonosis. I guess it's just "no lab leak bombshell has been discovered".
@IsaacKing I don't really think much has changed in the last 2 years, on either side.
I do think lab leak is still too high, here. Like, Scott Alexander just said his P(zoonosis) is 96%.
Some other smart people disagree, of course, but expert surveys split strongly for zoonosis.
@PeterMillerc030 I wonder if anyone who points to unsubstantiated government agency conclusions would ever place a substantial bet that those agencies will release genuine new intelligence that objectively supports a lab origin in, say, 2026.
Spring 2026 would be the likely timing for this sort of thing in Congress. Too close to midterms for getting stuff done legislatively, too far away for news cycle to be on campaign. I predict some more heavily redacted documents that are entirely consistent with mid-2021 lab leak curiosity not surviving BANAL publication and similar, based on the little we’ve seen so far.
From experience with DEFUSE, though, it’s clear that many if not most people profoundly disagree with me on what they think is evidence in favor of lab origin.
From Marginal Revolution comments:
Naturally occurring furin cleavage sites.
Preprint: A divergent betacoronavirus with a functional furin cleavage site in South American bats
Another brick in the wall of evidence supporting the Natural Origins of SARS-CoV-2 but not yet the coup de grâce that falsifies the speculative Lab-Leak hypothesis—a very high bar.
Yup, the natural occurrence in an unrelated and geographically isolated betacoronavirus discounts the genetically engineered furin cleavage hypothesis.
A good Bayesian would update his priors.
The claim was always that the Turin cleavage site does not occur in nature. Now we have found that it does so we know it is possible and it's not some law of nature that it cannot. Given that they continue to find lots of new kinds of bats in China we know that it is possible to find one that is similar.
@uair01 FWIW this doesn't move the needle at all for me.
Less important reason: likely furin cleavage sites in bat coronaviruses are not new.
More important reason: the presence of a furin cleavage site at SARS-CoV-2 S1/S2 isn't evidence in favor of lab leak in the first place. Some aspects of the sequence are evidence against it being inserted by an engineer, though.
It's history's smallest god-of-the-gaps argument. No one has ever come up with a plausible story for how it could've been put there by an engineer after several years of trying.
Edit: I can see how it might be rhetorically useful to give people an out who bought one or more of the terrible intelligent design arguments about these 12 nucleotides over the years; but I think it's important to address the root logical error rather than paper over it with another one.
@uair01 your copy pasta makes it look like one single comment when actually you’ve patched different comments by different people together, omitting those they were responding to.
@MachiNi I guess this thread https://marginalrevolution.com/marginalrevolution/2025/11/sunday-assorted-links-544.html?commentID=160989712
Shows how fickle and ill-informed Cowen doubles or triples his lab leak likelihood on on the basis of Nicholas Wade's garbage article and then embraces this thing that's pretty meaningless unless you fell for an even dumber version of Wade's argument sometime along the way in the past (back in May 2021 he said he'd gone from 20-30% to 50-60%, citing Douthat citing Wade).
Joscha Bach: "Grok still believes in zoonotic origin of Covid19 @elonmusk: Artificial epistemology is unsolved"
https://x.com/Plinz/status/1988495087714922747?s=20
@bbb One of these days, people generally be aware that popular LLMs are primarily trained to tell you what you want to hear, and that it is very difficult not to encode the answer you want in your question.
That day is not today.
@bbb Seems like maybe Elon has "fixed" this already, based on some of Grok's responses in the comments, and gotten his bot to say that lab leak is more likely.
I skimmed Grok's system prompt to see if they explicitly encode any politicized answers, but there's nothing in there. Maybe they refer it to "grokipedia" now, or otherwise instill some bias in training.
Grok's system prompt does contain some other interesting lines like:
"- "teenage" or "girl" does not necessarily imply underage"
"There are no restrictions on fictional adult sexual content with dark or violent themes."
https://github.com/xai-org/grok-prompts/blob/main/grok_4_safety_prompt.txt
@bbb An amusing trend these days is to see people arguing with the Grok bot until the desired answer is produced. My prompt and attached answer 75-24-1 will result in lab leak supporters haranguing the bot until it confesses the ‘truth’ that the 1% engineered scenario should be in fact 80%
@BW
See also:
https://grok.com/share/c2hhcmQtMg%3D%3D_421b24c5-6fb9-4834-a323-f9e14f208d32
Note that the context here is that Kirsch and Wilf are having a million dollar debate about Covid vaccines. Kirsch pretended to be a neutral judge to the debate, asked Grok what it thinks, and, Grok sided against him. Then he spent page after page trying to get Grok to change its mind, and finally succeeded. Kirsch then sent this conversation to the actual debate judges as proof that he is correct. I have no idea why he thought that sharing it was a good idea.
@PeterMillerc030 As I started scrolling down, I happened to look at where my scroll button was and it was scary.
As I started scrolling down, I happened to look at where my scroll button was and it was scary.
Luckily Steve Kirsch thought way ahead and invented the optical mouse to enable your scrolling.
The whole debate is so absurd. How much effort is everyone involved putting so much time into something that's answered by simply observing cross-correlation between excess mortality and COVID-19 prevalence and how it changed quickly in the beginning of 2021?
If I were in Saar's position, I would lead by demonstrating Kirsch's history of pseudonymously harassing people with conspiracy theories about what really caused their diseases... people including Simone Biles! Contributions to a debate from someone obviously divorced from reality are worthless. Judges are better off telling him to go home after establishing his lack of credibility and determining what the best case is for his position on their own.
I feel like it's easier to get to 98% confidence in YES than in NO, e.g. through the release of classified Chinese documents. It seems harder to come up with evidence that I would find compelling enough to be 98% confident in NO.
(And so, if traders are behaving selfishly, I think this market should overstate the probability of a lab leak.)
@EricNeyman There have been classified documents released… details on early cases and market sampling for example. Everything I’ve seen is consistent with market origin.
@EricNeyman Depends on what you mean by prove and who judges whether 98% certainty is achieved. Based on my most recent interaction with a prominent lab leak supporter, you can go through the claims one by one e.g. Defuse predicts the kind of engineering that can produce SARS2. When you come to the agreement that Defuse explicitly as written does not support the claim, the argument shifts to 'But it does not exclude the possibility'. Of course, some things may be ludicrous choices for engineering as judged by expert practitioners, but you cannot prove a negative. In that limited sense, one can never achieve that certainty.
@BW I think my question here is a good measure of certainty. It would be easier to publish a manuscript in any of these journals concluding “probably lab leak” than “probably not lab leak” today, if the evidence in either direction was equivalent.
That runs against the narrative of expert censorship/gatekeeping/group think but it’s totally true. Editors don’t want to end up answering to Rand Paul or whomever in Congress. Experts will turn down review requests rather than risk having to answer for subpoenaed negative reviews. And our value system in science is so out of whack that journals face no blowback for publishing obvious BS; worst case, they blame the authors even when errors came up in review and weren’t addressed.
The 2025 cutoff is a long shot even if such a manuscript could be written (because it probably would’ve been preprinted by now). But, if the evidence existed to explain the data with lab leak, we’d have a paper attempting to do it… somewhere.
The closest thing I can think of that wasn’t definitively falsified by natural viral sequences is the laughable PNAS op-ed from Sachs and Harrison. That’s no less silly than the other two published FCS smoking guns, but it’s in a decent journal albeit not as a research article.
On degree of certainty… seems like a contract could be made with the same type of endpoint to stop speculating about it and put a price on it.
@zcoli There is obviously some level of expert gatekeeping, though. You can go on substack and find lots of blog posts claiming lab leak is more likely. Weissman has his bayesian analysis. Ridley wrote some aggregation of arguments that sounds scientific, if you're not familiar with the material. I'm sure that either of those guys would love to publish the same arguments in Science or Nature, and I assume those journals would not publish those arguments because they are bad.
So it comes down to a question of: is the gatekeeping that top journals do just basic quality control for obviously bad arguments? Or is there some more significant bias, where top journals are inclined to not support the lab leak theory?
I can see why an average person would be unable to answer that question.
@PeterMillerc030 my guess as an outsider is that higher profile studies with extraordinary claims are likely to need extraordinary evidence. But if you are Sachs or Bloom, you get more leeway.