Demaneuf and de Maistre suggested one natural sarbecovirus per 10 years, but didn't specify a serious one. Rootclaim said their 20 years value was generous. I just copied their 20 years value. Based on prior history, you'd think that 20 years is too low. Based on changing human activities, it's really hard to say.

Subtotal (Annual odds a coronavirus pandemic starts at Huanan)

.002 historically for BSL-3 labs in USA. Demaneuf and de Maistre said .02 was the "worst case scenario". Ron Fouchier thinks it's 1 in 10 million for his lab.

Interviews with many scientists say that they're surprised anyone would work on a virus only 80% similar to SARS, and would focus on closer viruses.

Priors: ratio between lab/natural pandemic for Wuhan novel sarbecovirus

Rootclaim gave a range of values from 0.4% to 4% in debate 1, and then changed it to 10% in debate 3 based on his higher exponential growth theory

Distribution of positive environmental samples and cases within market

Maybe I should count this against zoonosis, but I'm not sure. The one lineage A sample at the market does have a human case associated with that shop. Also, in Saar's concept of steelmanning, this is really easy to steelman -- the first lineage A patient could have been a customer, not a vendor.

It's important to avoid double counting. Since the judges conditioned the natural priors on a serious coronavirus pandemic, this is probably included, because it's a feature that makes the pandemic serious. Since my prior was 1 in 20 years, and there's was closer to 1 in 100 years, and serious coronavirus pandemics might be less common, I'm going to add a 5 here.

I think I gave a range during the debate, and said that my argument would still win with Saar's 100X ratio here. Since the debate ended, I realized that no lab in history has added FCS via insertion, they always did so by mutation. On the other hand, nature does 12 NT insertions with some regularity. So I'm now closer to 1 on this value than I was before, and this could even lean natural.

this is based on the logic that the original virus could have had different codons for the leading serine. With 6 codon choices, 1 would be in frame.

I gave a range between maybe 0.5 and 5.0 for this, not obvious to me how important it is, but it's obviously not the 1000X for lab that some people claim.

I'm not sure what Rootclaim even meant here. Is this just whether covid binds well to human ACE2? That's just a precondition for a pandemic, any sarbecovirus that infects people would bind to ACE2. Eric interpreted this to mean that covid binds best to human ACE2, of all species. But I don't think that's even true -- various papers disagree and Rootclaim cherry picked the one paper that argued that. Also, even if that were true, it would be in conflict with Rootclaim's 50% prior that any old virus could be used to start the pandemic, because a starting virus with human optimal ACE2 binding would surely be a much rarer thing than that.

I think this could honestly lean natural, perhaps even strongly so -- it's probably a sign that the virus jumped species, and it's unlikely that a lab would do an experiment with a novel backbone, and then a furin cleavage site, and then also an N-glycan change, all without any prior publications or prior leaks. But the arguments here are complicated and the odds Rootclaim used were low, so I mostly just ignored this.

This was separate from no secret virus and "coverup would have to succeed". Realistically, I might have gone overboard with counting secrecy in many different ways.

I did make a mistake here, with regards to the humanized mice, those would not cause the mutation I said. So maybe my value should be different? I do think we probably still need some factor here, as the virus could not be in culture for long.

Even Yuri doesn't buy this argument anymore. And if they really made SARS2 with an FCS insertion, I think they wouldn't disclose RATG13 at all, it would make the lab's mistakes too obvious, so this probably should lean natural.

It is proven that China hid data about the market (like the raccoon dog DNA), and I theorize that they still have more hidden/unpublished data. So perhaps I should not count secrecy so highly. On the other hand, there really hasn't been anything to turn up for lab leak. No e-mails. No trustable evidence of sick employees. No scientists outside of China know about secret work at the Wuhan lab. And there's been lots and lots of searching.