Will Peter Miller win the "Rootclaim Challenge" debate with Saar Wilf on the origins of COVID-19?
resolved Feb 18

Important: in the case of a draw (both judges undecided, or a split decision), this market resolves N/A, not NO.

Rootclaim is an organisation, founded by Saar Wilf, that performs analysis and makes truth claims on various, possibly controversial matters. One current such claim is that COVID-19 originated in a lab leak, with high probability.

Rootclaim has a standing offer called the Rootclaim challange, in which someone who disagrees with their analysis may challenge them to a debate judged by professional judges, with each party having $100,000 at stake.

No such debate has yet taken place. (Edit: It has now!)

AstralCodexTen commentor Peter Miller (who goes by the handle BSP9000) is currently considering taking them up on the offer, on the topic of whether COVID-19 originate in a lab leak. Peter Miller will argue that COVID-19 did not originate in a lab leak, and RootClaim will argue that it did. Peter and Saar Wilf are currently attempting to find judges.

Conditional on such a debate going ahead, will Peter Miller win?

This market will resolve YES if a debate largely matching the spirit of the Rootclaim challange, with approximately $100,000 at stake, goes ahead this year, and Peter Miller is judged to be the winner. Reduced stakes will not count, other than as needed to pay for the costs of the judges and other costs relating to the debate itself, which is stipulated as part of the challange.

If Saar Wilf is judged to be the winner, this market will resolve NO.

If the debate does not go ahead or if there is not a clear winner, it resolves N/A.

Edit Oct 22: in light of the rules of the debate prescribing different degrees of victory, as discussed by Peter Miller here, It seems neccesary to clarify what counts as "winning". Peter Miller writes:

The winner will receive $100,000 for a unanimous decision. If one judge is undecided, the payout will be $50,000. There's also a possibility that neither side wins anything, if both judges are undecided or it's a split decision with one judge voting natural and the other voting for gain of function.

Assuming the above is accurate, I'll count a participant that receives either a $100,000 payout or a $50,000 payout as the winner. That is, a participant declared the winner unanimously by both judges, or by one judge with the other undecided, is the winner for the purposes of this market.

Update Dec 30th: the debate has taken place and we now await the judges' results.

Edit Jan 22nd: here is Peter's latest comment with links to videos and documents from the debate, included here in full for convenience (since it will get buried in the comment thread otherwise):

I've posted all the videos and all the slides. Now we're just waiting on the judges to announce.

The original rules said they'd just write something, but Saar and I recently asked judges to make some videos as well, for something more authoritative.

We're still figuring out the timing... maybe another week, give or take?

Get your bets in now before it's too late.

For anyone tuning in late, the videos are here:


And the written documents were posted in a few different threads:






Get Ṁ600 play money

🏅 Top traders

#NameTotal profit
Sort by:
bought Ṁ10,000 YES from 99.7% to 99.9%

@PeterMillerc030 Congrats on the 100k! Loved the video series, one of the few gems out there w.r.t in-depth research

@PeterMillerc030 what will you do with the money?

@NicoDelon Gain of function research?

(it's just a joke)

@NicoDelon 'I'm going to Disney World!'

Tony VandongenboughtṀ10NO

@TonyVandongen Why not go all in? Potential of over a 20:1 return in only a few days.

@zcoli As Warren Buffett says, if you liked it at 5% odds, you should love it at 4%!

bought Ṁ25 YES

@PeterMillerc030 I’m genuinely interested because there’s so rarely any criticism of quality of arguments made internally by lab leak supporters. Your job would’ve been harder if there was more reflection and editing. Presumably people making token bets on this with their free money think Yuri did a bad job, but can’t say so?

@zcoli It's possible that my job would have been harder, had a larger slice of the lab leak community been involved in preparing the arguments. But it's also possible that a larger group would have just prepared inconsistent arguments, given how various lab leak theories contradict each other.

If Rootclaim loses, the lab leak community might say that Saar/Yuri just did not argue the best lab leak theory.

The community could also try to release some new evidence or "smoking gun" in the next few days to overshadow the debate announcement.

The most likely outcome, though, is they'll just ignore the debate, pretend it didn't happen, and keep complaining that no one will debate them. That seems to be the strategy so far, for the biggest lab leak accounts.

@PeterMillerc030 True, except for some like @GillesDemaneuf who know better yet also know they need influential allies, so declare that Rootclaim's contention that hospitals aren't well suited for superspreading "is an excellent thread with some great analysis."

@PeterMillerc030 Seriously, I think the debate that would move opinion on this most significantly is booking someone such as yourself on the Daily Show now that Stewart is back. He dared everyone to openly discuss it and now he has a platform to host the discussion.

@zcoli Stewart's old show would have been good for this, I think The Daily Show is way too short format, I don't know how to get this down to 5 minutes of good sound bites.

I think I'd enjoy talking to Sam Harris, if he'd have me on. Not sure how to arrange that, though.

(though his show is audio only and I'd have to think how well I can present this without slides)

sold Ṁ46 YES

Figured it was worth selling a bit to see what people think the judges will conclude if they choose to give their estimated likelihoods following the debate.

Here's how the market will resolve depending on the median price over the 24-hour period prior to judges' likelihoods (likelihood of "lab leak of engineered virus") being public and confirmed by @PeterMillerc030 and/or Rootclaim.

To simplify: Look at the picture below and see what "Judges' likelihood" corresponds to your beliefs and find the corresponding market likelihood. If the current price is less, then you should buy "YES" --- place a limit order for "NO" to cover you if the price gets too high or check back in often enough to pay attention to the 24-hour median.

bought Ṁ25 YES

@zcoli Trading at 45% now with a “fair” value above 60% per the price of this market. But low liquidity so can’t bet too much safely if you think judges will decide “51% lab leak”.

If the structure of the market is too complicated or just not enough liquidity to be worth spending time on, I’m interested in what people think the likelihoods will be and why. Since I’m judging that question I won’t give a guess, though.

A version of the market where draw doesnt resolves to NA but resolves to NO

Isaac KingsoldṀ907YES

@IsaacKing happy to buy more of your position if you're looking to sell

predicted YES

@Joshua The delay is the reason I'm certain this will resolve Yes, but I actually became very confident two weeks ago when I saw Saar basically concede they expected to lose in debate three part three in this slide where he talks about what would happen if they kept doing more debates:

"Unlike Peter, we did not do a thorough review of studies and data" 💀

predicted YES

@Joshua Imagine what would happen if anyone arguing for likelihood of zoonosis cited in their top 4 bits of new evidence something they read in the Daily Mail about someone’s unverified, but totally verifiable, claim of a not very plausible November 2019 onset COVID-19 diagnosis.

predicted YES

@Joshua If one side researches extensively and finds lots of evidence, that means that there must be an equal amount of evidence supporting the other side that just hasn't been found yet.

Therefore, the side that researched less probably won. That's just how probability works.

predicted NO

@Joshua That’s gotta go in some sort of hall of fame for debate arguments. It’s so beautiful

  • Let’s imagine what will happen if we spend more time on this.

  • We’ll find more evidence…

predicted YES

@NateWatson Rootclaim fundamentally does not exhibit any understanding that (1) low likelihood of changing your mind means it’s likely not worth looking for more evidence, but (2) if someone gifts you new evidence at zero cost on your part, you should happily accept it and update your beliefs accordingly.

Instead, from the very beginning the behavior at Rootclaim is seeing new evidence in terms of the cost of adding a new epicycle to their theory to explain how it’s uninformative. I don’t blame them, since this is how Rootclaim’s muses who occasionally publish on this topic in the scientific literature also behave. Rumors and motivated reading between the lines premised on what’s worth reporting in tabloids or worse is data that demands further investigation. Whenever there’s actually new, relevant data (e.g. retrospective serology in Wuhan or a new related genome) it’s rare that it will even be mentioned by lab leak folks. If there is some need to acknowledge it, it will be in terms of how it’s negligible.

predicted YES

@zcoli One thing of importance is that there is data that is relevant, known to exist, and is unpublished. I’ve seen “zoonati” call for its publication. I’ve seen lab leak proponents call for more fishing expeditions into lab notebooks in North Carolina.

The frustrating part is that it’s totally rational for data to remain unpublished when the data China does publish is ignored when it doesn’t point to lab leak or misrepresented to support some bank shot theory of lab leak. Look at the various reports drawn from interim snapshots of the COVID-19 case database in China. Is this used to inform the picture of the early pandemic? Not really, the primary international impact was that the Washington Post had a conspiracy theorist ghost write a major OpEd based on motivated misrepresentation of the data

This means that we are less likely than we otherwise would be to get additional data.

predicted NO

@zcoli What is the data China has published which has been misrepresented? Which WaPo OpEd are you referring to? Seems to me that the Chinese have been reluctant to reveal information even before lab leak was serious consideration (Chinese academic publications on Covid-19 have required Ministry of Education approval since April 7 2020). It is also hard to take the Chinese position seriously when they have had state media assert that Covid originated in Italy, Australia, and the US. But perhaps I have been too primed to assume they are acting in bad faith. So which data has been misrepresented?

@brp I think this is what's being referred to:

Which turned into this WaPo article:

Gilles' theory is wrong, that was discussed briefly in the 3rd debate. Also, AK and I debated this in an old thread and he even admitted in the end that Gilles was wrong on this particular point:

Discussion also continued in this market:

Or, for a more concise debunking, see:

The market itself never resolved because it was never clear what the question being bet on was. I.e. if Gilles is wrong, then AK admitted that, but if the question is if the WHO report has any errors in it, then the answer is yes.

There is still uncertainty as to how exactly how many covid cases there were in December 2019, but there's no proof of hidden November 2019 covid cases.

@PeterMillerc030 That's not what we agreed on. I agreed Gilles' interpretation of one of the study is wrong.

But the broader point of an incomplete retrospective search, as laid out by Gilles, still holds and is confirmed by multiple sources like early WHO reports, early China CDC report and early peer reviewed papers from China.

An incomplete retrospective search is evidence for a lab leak.

@Akzzz123 IIRC we agreed that there is one study that goes as high as 227 December cases? But Gilles was clearly in error when he "proved" there were 257 cases in December and 9 in November.

I described some of the more specific differences between the case counts in this comment:

(i.e. there's no clear pattern of additions/deletions between the reports, it's not like the one with more cases had a bunch more at the beginning of December that got removed)

@brp I'll take your nonsense from state media in China and raise you one White House leaking a stupidly detailed lab leak fantasy to the press so that it would be asked at a press conference mid-April 2020 and all of the conspiracy theory bait pushed out by the State Department at the last minute. This garbage flows both ways on this and it's hardly the only issue where US/China rhetoric has a loose connection with reality (despite being related to a real issue). The Wuhan Military Games is a particularly good example with propaganda from China saying it meant that SARS-CoV-2 was imported and propaganda from abroad saying that it meant that SARS-CoV-2 was covered up in October 2019. The truth is that everyone's just making things up. But only one side of the debate points to the stuff made up in a geopolitical battle over our heads as evidence to support their case.

@PeterMillerc030 Please check finding 7 in Gilles' report.

@Akzzz123 How did finding 7 in Gilles' report make you feel?

@PeterMillerc030 Do you have anything to add?

@Akzzz123 Gilles is a prolific writer. Rootclaim's strategy was to say, "Gilles wrote a 200 page report on early cases, which we did not read, but we think might contain useful information". Saar requested the judges read it.

In our conversation, you said that Gilles' report proved the existence of ~90 hidden cases, and 9 in November. I disproved that and you agreed that Gilles is wrong on that point.

Now you are saying that Gilles' report still has some other important information in it. If you'd like to spell out what else you think is important, I might choose to look into those additional claims.

@PeterMillerc030 This whole thing of appealing to anecdotal maybe-case X is kinda absurd. There are closer to 100 than to 10 unascertained and/or unreported infections for every COVID-19 case reported (the fraction of reported cases that are symptomatic COVID-19 cases being so high makes that clear enough). The question is what's the specific evidence of a case early enough and with associated epidemiological or genomic data to provide evidence that shifts conclusions. Along those lines, what's been uncovered by investigations from the crowd and media mainly supports Huanan market origins and somewhat earlier infection with lineage B than lineage A. Probably the best example is being able to add a black triangle for the earliest likely COVID-19 case at the same place as a red square indicating isolation of lineage B SARS-CoV-2.

More related questions