zlacker

Wuhan is the capital of Hubei and is going to attract the surrounding area (if a virus jumped to humans in Sidney, NY it would probably be detected in NYC). Distance from Yunnan will also be a factor and the virus jumping to humans from Yunnan to cities anywhere in the north will be much less likely. A proper model would look the migration of people and animals between different cities and provinces in China.

And again there's the SARS-CoV-1 where bat viruses in Yunnan jumped to people in Guangdong, 700+ miles away. And you'd have to look at the whole population around Wuhan where it is the closest major city (and Wuhan itself is the largest major city in central China, which makes it comparable to massive cities like NYC).

The probability is likely closer to 10%, and SARS-CoV-1 already lost that die roll and wound up in Guangdong. This time you rolled and got a 10% result. That isn't very unsurprising. Your p value is not significant.

And, yeah, I don't think that they've done everything they can to survey broadly. The studies that they've done tend to collect a bunch of bats from a particular cave and sequence the viruses in them and then draw broad conclusions about the whole province. There's only two sarbecoviruses sequences they've found in Hubei. I'd be much more comfortable with statements that SARS-CoV-2-like sarbecoviruses aren't found in Hubei if they'd sequenced 100 of them.

EDIT: Yeah so Wuhan is 6% of the population of all cities in China of roughly its size. But Shanghai is father away so its 24M should be weighted less, so with Tianjin and Nanjing. Beijing is way further away and its weighting should be fairly negligible. Chengdu and Chongqing are slightly closer. Of the top 10 Guangzhou and Shenzhen are both similarly close and in Guangdong where the SARS-CoV-1 spillover happened. My 10% gut estimate still looks pretty good -- "pick a random top 20 city in China somewhat near Yunnan where the bats are" and Wuhan is not that improbable at all.

>>lamont+(OP)
For emphasis, I'm not claiming the emergence in Wuhan (or the presence of an FCS, or the higher affinity for human ACE2 than for any other known animal, or the CCP's apparent coverup of something, or any other single factor) is determinative in itself. But there are multiple factors and no clear mechanism for dependence between them. Those should all be considered together, in a Bayesian analysis or whatever other statistical framework you prefer.

As to specific numbers, 10% seems too low to me--Wuhan is 6% of the population of Chinese cities at least as big as Wuhan, disregarding smaller cities entirely, including Foshan where the original SARS emerged, and disregarding Dr. Shi's expectation that origin closer to Yunnan was more likely. But at least we're within an order of magnitude of each other.

Knowing nothing whatsoever except that a pandemic emerged, what's your prior for lab vs. natural origin? It depends how you count, but we've had perhaps a dozen pandemics in the last fifty years, and one (the 1977 flu) was near-certainly lab-origin. So even with extreme optimism as to lab safety improvements, I can't see how you'd estimate less than e.g. 1%. For SARS-CoV-2, you'd perhaps adjust down for the novelty of the pathogen (since most lab work is with known pathogens--though most of the WIV's seems not to have been). It doesn't take that many p = 10% coincidences up to get to even odds or better, though.