Would you bet on a real money version of a market like this? Why or why not?

If it’s conditional on market odds, what odds range would you bet at?

(Assume it’s not necessarily the same single judge as here, maybe it’s a different person or a panel of judges, but that it’s a similar >98% likelihood in one direction criterion for resolution)

@PeterMillerc030 Probably not because if I use the Good Judgement folks as a reference, the best they could reach was 79%.

@PeterMillerc030 I'd make a chained bet on this scenario:

1. Someone will eventually find and publish a bat or other non-human animal sarbecovirus with a functional furin cleavage site.

2. Many people will unfortunately* call it a smoking gun for natural origins.

3. It will have zero impact on the popularity of the lab leak theory, even though the most popular lab leak theory has centered on the SARS-CoV-2 furin cleavage site for years.

4. Alina Chan will write another Op-Ed telling us why lab leak is more likely than ever.

@zcoli We saw a little bit of that with BRZ batCoV, because people saw 'bat virus has FCS'. It died soon after. Perhaps lab origin will cease to be an animating force in Trumpist politics in a few years. I have no idea what will cause people to change their minds. If they do, it will be for some random reason. At that point, I don't see Alina Chan maintaining her relevance.

@PeterMillerc030 Good question. Objectively I would like to accept zoonotic origin. But I have difficulty letting go of LL because of (dis-) trust:

• It is hard to trust the CCP who have a conflict of interest

• It is hard to trust the GoF researchers who have an obvious conflict of interest

• I'm still irritated how the researchers suppressed any opening for LL.

To bet real money I would have to believe in an outcome that disproved my defensible bias. Alas, I don't think such evidence will ever come.

I admit that this is indefensible and subjective.

@BW See other comment. What would change my mind if the GoF and Covid researchers admitted that they were too hasty with denying LL. A statement like this would be sufficient: "We still believe zoonotic is correct, but we were too hasty in denying LL in suspicious circumstances with insufficient evidence. We admit that we were too close minded."

@uair01 PO found the engineered hypothesis to be low probability but said ‘it is currently impossible to prove or disprove the other theories of origin described here’. They came out strongly against lab scenarios. Given the media hype for all but one quarter, it is hard to believe that lab origin voices have been censored.

there is nothing about PO that is not supported in the years since.

@zcoli I’d be willing to bet some money against step #1.

I.e. we could wait 5 years and I’d bet there will still be no other sarbecovirus with a functional FCS at s1/s2.

That hasn’t been found in bats in ~20 years of sampling, odds are low it would be found in 5 more years. If we started sampling other animals the odds might go up, but I’m not aware of many studies like that.

Conditional on #1 being true, though, I think 2, 3, and 4 are correct. Maybe a few lab leakers would change their minds but Alina would not. And the general popularity of the lab leak theory has less to do with the technical details, so it might not change much.

@BW I think you’d probably agree that the brazilian virus had little to no relevance on covid origins, but I guess you’re right about the impact on the conversation—some people portrayed that as hugely important new evidence (“bat virus with FCS”), a few people tweeted that lab leak theory had fallen apart, most people didn’t notice or care. Most people reacted more strongly to random quotes from “Fauci’s diary”.

@uair01 The most influential papers arguing for a market origin were written by guys (Worobey and Pekar) who weren’t convinced by the market origin theory as of mid-2021 but changed their minds in late 2021/early 2022. Neither one of them works on gain of function research.

@uair01 When you say you wouldn’t bet real money, does that mean you wouldn’t bet at any price?

@PeterMillerc030 yeah from my POV it's the opposite; the thing that common wisdom shut down irrationally was origin in Huanan-market-linked wildlife trade, based on a very low number of suspected unlinked cases and some bad reporting by SCMP.

@BW Thanks, that's a very interesting statistic.

@PeterMillerc030 Thanks for the correction. I believe you. Answer to your other question:

I think I would bet money on bets like:

• A virus will be detected in the next 5 years that has a very high similarity to the covid virus (Yes)

• There will be a leak, hack, deserter or whistleblower that proves lab leak (No)

These feel more verifiable.

@uair01 FYI, the closest full bat virus genome to Sars1 has 96.2% similarity while that for sars2 has 96.8% similarity. For sars1 a near identical match to a human isolate was found in a civet at a spillover site. I would not place any high odds on that happening in the next five years.

@BW Yeah, possible that you could scour Laos/Yunnan and find a 97.5% similar bat virus, or something like that, but I’m not sure how that would change the conversation.

A 99.8% similar virus in a raccoon dog would have had a big impact on the conversation in 2020, but you’re not going to find that today. If you find an infected animal on a Chinese farm it’s going to have some modern variant it caught from people. I doubt you can find some wild group of raccoon dogs that kept reinfecting each other for 6 years.

There was a period of time during zero covid in China where they could have done serology on farms (or wildlife traders), but that window has passed.

The only chance at this point is if China found that in virus in 2020 and didn’t disclose it. But, of course, lots of people wouldn’t believe that disclosure even if it were true.

@PeterMillerc030 if a 99.8% similar virus were found in an animal in China, the same people would be more confident in lab leak.

They’d say it was probably fake because nothing so similar had been found for SARS-CoV (accuracy of this statement doesn’t matter).

@zcoli Claude eval, for all of AI's limitations, at least does not suffer from some human biases. For instance, the idea that there is secret intelligence. This seems obvious, but somehow Claude processed this better after seeing both sides of the arguments. "Low confidence on both sides of a 5-3 split is the signature of absent evidence, not withheld evidence; a dispositive holding produces convergence, not division."

https://zenodo.org/records/4511826

Investigation into Wuhan Institute of Virology Pathogen Databases

On the 12th Sep 2019, the main database of samples and viral sequences of the Wuhan Institute of Virology went offline. Eventually every single of the 16 virus databases managed by the WIV was taken offline. Here we show how these databases may provide essential clues at to the origins of SARS-CoV-2 and review the circumstances in which they were taken offline.  

@0xseraphim I uploaded Claude’s comparison of the two sides’ claims. The conclusion is that both sides are overstating their claims on the database. But the evidence is weak and does not change P(LL) <0.2%. You can take a look in section 3.39 on page 70 in the current version.

@BW I think... that maybe I'm ok... with not looking at a doc filled with 70+ pages of AI slop.... 🤣

@0xseraphim ok I presumed you were trying to say something with a link to the very old database story.

@BW that's ok! you can give me the short version in your own words if you're up to it :)

@0xseraphim it is a very common pattern. Prominent lab leakers claim something that is bullshit. They get called out. They admit it was bullshit but useful idiots keep spreading it even 6 years later.

@DavidKochanov what are you referring to?

@0xseraphim It was never much active. It is alleged to have been publicly accessible containing incriminating sequences that the whole world missed before it got shut down. Shi’s accounts are corroborated by Daszak to Markolin and Jane Qiu’s reporting independently. And the Zhihua Chen/Coroldo demonstration that substantiates internal testing of database only based on Beijing IP which is where WIV traffic geolocates to.

@BW I have heard this too. And yet, the timing in hindsight looks so very suspicious when lined up with the estimated original contagion event. Not that it is, by itself, a smoking gun.

@0xseraphim so you do not believe the November sick researcher story? Because one cannot believe it leaked twice. Whose estimate though? It is not a mainstream opinion.

@BW If it really did come from a lab, the main smoking guns to look for would essentially be:

- A laboratory possessed SARS-CoV-2 or an extremely close precursor before the outbreak

- A laboratory worker was infected before the first known community cases

- Laboratory records showed a containment failure involving the relevant virus

- Research records documented construction of a virus matching SARS-CoV-2

- The viral genome contained clear evidence of artificial modification

- The earliest infections formed a chain leading from laboratory personnel to the wider community

If you steelman the hypothesis that there was some kind of event at the wuhan lab around september 2019 (ignoring stories about later events for now); then assume they go through some big kneejerk reaction to cover it up; taking that database offline makes a lot of sense within that framing. Even if that original database was never used much to begin with. The fact that they took it offline at that exact moment is a coincidence that really stands out (nonwithstanding any alleged incriminating sequences it may or may not have contained)

@BW you'd basically want to erase as much data as you could as fast as possible and then implement a total information blackout regarding any new related events

@0xseraphim that is why you have to look at it in the context of all the other things you mentioned. Which are also evaluated. I’m afraid there is no short cut to discussing that. One can have inferences and assign likelihoods.

@BW I guess so? But there's so many plausible pieces of evidence to consider. Bayesian analysis isn't going to help you all that much here without having better data to analyse

@0xseraphim anybody interested can take a look at the rebuttals to the issues you raised whether or not the numbers have meaning.

@BW It's like the furin cleavage site. Go assign a prior to that lol

You can ask three different statisticians to mathswash it independently and come up with three completely different numbers

@BW I'm not saying "the numbers" you're talking about are useless. It's just that they're probably not going to be very convincing in the absence of new data

@0xseraphim Addressed in detail and Claude ends up giving a mild likelihood for natural origin. Everything you have mentioned is addressed.

@BW I can ask gpt-2 to address it as well ahaha

@BW Claude circa October 2026 is no Oracle. You still gotta use your protein brain buddy

@0xseraphim go ahead. It was a closed book study as I have said. Comparison of provided arguments only.

@BW but you're aware that... no one cares right?

@0xseraphim ok bye

@BW 🤣

@BW to be clear, people are interested in what you have to say

It's just that people probably aren't going to care all that much about a massive dump of Opus slop. Maybe there's some insights in there you think stand out?

@0xseraphim I started with that in the other post if you are looking for something that can be said in two minutes.

@BW you mean this?

What are the insights that stand out in there?

@BW in case you missed it, AI’s role is not to provide insights but to act as a judge.

@BW Ok, but it probably isn't going to give better numbers than other sources right? Because it doesn't have better data, and it doesn't have superhuman intelligence. I guess I follow you if you're making the claim that it's more "neutral". But I wouldn't assign much value to it compared to an aggregate of numbers from other sources.

@0xseraphim sure interpret it however you want.

@BW by all means if your game plan is to respond to anyone posting comments on this market page with references to your doc, feel free to do so! You might find them to be unconvinced however ahaha

@0xseraphim no such game plan but nobody debating here is going to be convinced of anything by the other side anyway.

@BW yeah I guess you're right there

@BW tell me about the November sick researcher story

@0xseraphim sorry bud, I am going to drop out of this.

@BW alright have a good one

@0xseraphim The genomic diversity of SARS-CoV-2 rules out an outbreak in September. The end.

@0xseraphim Also, weird no one ever reports how many of those databases are available online today, to the extent that they ever were available.

@zcoli CIA guy 'Credible reporting does indicate that the pandemic may have begun in November 2019 in Wuhan, China and was the product of a lab incident'
But the way the game works is to say September database to hide leak, November researchers for concealment, CIA says 'lab leak', 'cryptic' transmission since September that results in two lineages and hope nobody notices that these are all very different things.

@BW The only things we know about CIA investigations:

• What some of the analysts told the nutty whistleblower Erdman and convinced him that Fauci secretly visited Langley and that analysts were paid off to change their conclusions. All false.

• That one person who briefed Fauci had a theory about a pangolin virus studied at WIV. The study they had somehow heard about was published later and had nothing to do with COVID-19 origins.

• Not clear there is even a single favored lab leak theory within the CIA, but clear it’s different than both the favored DOE and favored FBI theory.

@DavidKochanov

Prominent lab leakers claim something that is bullshit. They get called out. They admit it was bullshit but useful idiots keep spreading it even 6 years later.

And then when you tell people why it’s bullshit, they complain it’s too hard to read all the details and either disappear or change the subject to some other bullshit claim.

@zcoli Not including the stuff that’s consensus for the whole IC (no SARS-CoV-2 at WIV until end of December) or near consensus including CIA (virus not modified in any lab).

However, the IC investigation products we’ve seen declassified are largely characterized by bad logic and dumb mistakes. So conclusions leaning natural origins are as worthless as those leaning lab origin until we see something that’s serious.

@PeterMillerc030 Or they just disappear and pretend it never happened…