4 ms·

It's not an impossible coincidence, but what prior estimate of the probability that an arbitrary pandemic is lab-origin are you applying that to? Of the ~dozen

It's not an impossible coincidence, but what prior estimate of the probability that an arbitrary pandemic is lab-origin are you applying that to? Of the ~dozen pandemics in the last fifty years, one (the 1977 flu pandemic[1], which killed ~700k people) was probably lab-origin. A 1/12 prior plus a 1/10 coincidence in favor of lab origin is pretty close to even odds, so what evidence takes it back down? You can't just say "because pandemics are usually natural", since that's what the 1/12 prior already captures. You present many reasonable arguments that natural origin is possible, and I agree that it is. That doesn't mean lab origin is impossible though, and it doesn't mean the possibility is negligible unless your prior is very heavily biased against it. Considering that a lab-origin pandemic has probably happened before (and has even been dismissed as clearly natural by the WHO[2], before the later scientific consensus corrected that!), I don't see where you're getting the confidence you've expressed in other comments that the origin of this one is certainly natural. As to exact numbers, I agree that it's reasonable to consider only the urban population, since emergence would probably be in a city. There's no reason to limit the population to China, though--natural emergence elsewhere in SE Asia would also have been possible, and the WIV's research program is distinctive not just in China but in the world. It seems like our disagreement hinges mostly on that first prior (considering that a pandemic emerged, but no other information), though. 1. https://www.nejm.org/doi/full/10.1056/NEJMra0904322 https://www.nejm.org/doi/full/10.1056/NEJMra0904322 2. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2395678/pdf/bullwho00443-0095.pdf https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2395678/pdf/bul...

lamontcg 5y ago
The 1977 flu pandemic was a known virus though which was released. It was highly plausible that the virus could have been in a lab. This virus is literally unlike any other known virus. The closest known virus to it is RaTG13 which is only 96% homologous and the 4% differences across all its basepairs is very high. We have a closer relationship to apes than SARS-CoV-2 does to RATG13. And we had SARS-1 in 2003 and H1N1 in 2009 and MERS in 2012, where there's no lab involvement at all. Not to mention Ebola, Zika. It estimated that some 30+ viruses have spilled over from animals to humans since the 1970s, and there's been one lab leak. And we know that Yunnan province has bats with very similar viruses, with one virus having been identified which is almost certainly the precursor to SARS-CoV-1 and another virus which is the cousin of SARS-CoV-2, and we have a spillover event that happened in 2003 to Guangdong with SARS-CoV-1 showing that this can certainly occur without the involvement of any labs. What remains that is objectionable about the spillover theory of SARS-CoV-2 is the "wild coincidence" of the WIV lab being in the same city as where the spillover happened. This time. But a large city is likely where it would have been identified first. Estimates from the genetic clock of the virus is that the spillover happened sometime as early as October. It would have been spreading in rural areas before it hit the nearest big city and caused a superspreading event and was noticed. To accurately determine the likelihood that the spillover was in Wuhan and not somewhere else you need to accurately know the distribution of the bats in China, the distribution of the sarbecoviruses in the bats, and the intermediate animals that might have been involved and their movement and the movement of humans. We know from SARS-CoV-2 that the virus jumping from bats in Yunnan to Guangdong can happen. We can expect it would be unlikely for the virus to leapfrog all the way to Beijing. We don't have all that information, so it now becomes pulling numbers out of butts time. To me that feels more like rolling a natural 20 on 1d20. With SARS-CoV-1 we rolled a 6 and it turned up in Guangdong. If we rolled a 11 or 12 maybe it winds up in Congquing. But we have no idea what the intermediate species might be, which might make spillovers in Congquing unlikely if those animals aren't farmed near there. And we still have no idea of the mechanism by which SARS-CoV-2 wound up in Guangdong, but we know it did, and it didn't involve a lab.
- tripletao 5y ago
  It seems like you have a lot more confidence than I do that we know everything the WIV was working on? I agree that SARS-CoV-2 is unlikely to be derived (naturally, artificially, or otherwise) from any known virus, including RaTG13; but a distinctive characteristic of the WIV's research was their sampling of novel viruses from remote bat caves that no other humans routinely entered. That presents a pathway for emergence of novel, naturally-evolved pathogens beyond what normal agricultural etc. activity already risked, whether through a researcher infected in the field or through a literal lab accident back in Wuhan. So specifically considering their group, the novelty of the pathogen doesn't seem to me like significant evidence against lab origin. The public focus on gain-of-function research actually seems excessive to me--while I do think it's possible that the origin of SARS-CoV-2 involves lab manipulation (e.g., genetic engineering or serial passage) of the virus, the WIV's library of novel natural viruses seems to me like at least as big a part of the danger. In any case, their database went offline on Sep 12 2019, around when that genetic clock suggests SARS-Cov-2 began spreading in humans: https://www.researchgate.net/publication/349073738_An_investigation_into_the_WIV_databases_that_were_taken_offline https://www.researchgate.net/publication/349073738_An_invest... That timing could be coincidence, and their flimsy explanation for its continued unavailability ("hacking attempts") could just be the reflexive secrecy of an authoritarian regime. It's a second weird coincidence, though. The WIV has claimed to have published everything they have to GenBank; but Deigin et al. were able to assemble a novel Merbecovirus from contamination of agricultural datasets from a team that shared sequencing equipment with the WIV: https://arxiv.org/abs/2104.01533 https://arxiv.org/abs/2104.01533 That Merbecovirus obviously has nothing to do with SARS-CoV-2; but if they had one unpublished virus, it seems reasonable to wonder if they have more. If you're going to extend the field of zoonoses to include events that didn't become major pandemics, then it seems like you should extend the field of lab leaks too? But let's say negligible probability of lab origin is p < 1/1000, which in an expected value sense is still roughly a 9/11 worth of deaths. Then even taking a prior of p = 1/30 that an arbitrary pandemic is lab-origin and considering the emergence in Wuhan, I don't see how you get to negligible unless the novelty of the pathogen decreases that by a factor of >100 (which per above I don't see).

lamontcg 5y ago

The 1977 flu pandemic was a known virus though which was released. It was highly plausible that the virus could have been in a lab. This virus is literally unlike any other known virus. The closest known virus to it is RaTG13 which is only 96% homologous and the 4% differences across all its basepairs is very high. We have a closer relationship to apes than SARS-CoV-2 does to RATG13. And we had SARS-1 in 2003 and H1N1 in 2009 and MERS in 2012, where there's no lab involvement at all. Not to mention Ebola, Zika. It estimated that some 30+ viruses have spilled over from animals to humans since the 1970s, and there's been one lab leak. And we know that Yunnan province has bats with very similar viruses, with one virus having been identified which is almost certainly the precursor to SARS-CoV-1 and another virus which is the cousin of SARS-CoV-2, and we have a spillover event that happened in 2003 to Guangdong with SARS-CoV-1 showing that this can certainly occur without the involvement of any labs. What remains that is objectionable about the spillover theory of SARS-CoV-2 is the "wild coincidence" of the WIV lab being in the same city as where the spillover happened. This time. But a large city is likely where it would have been identified first. Estimates from the genetic clock of the virus is that the spillover happened sometime as early as October. It would have been spreading in rural areas before it hit the nearest big city and caused a superspreading event and was noticed. To accurately determine the likelihood that the spillover was in Wuhan and not somewhere else you need to accurately know the distribution of the bats in China, the distribution of the sarbecoviruses in the bats, and the intermediate animals that might have been involved and their movement and the movement of humans. We know from SARS-CoV-2 that the virus jumping from bats in Yunnan to Guangdong can happen. We can expect it would be unlikely for the virus to leapfrog all the way to Beijing. We don't have all that information, so it now becomes pulling numbers out of butts time. To me that feels more like rolling a natural 20 on 1d20. With SARS-CoV-1 we rolled a 6 and it turned up in Guangdong. If we rolled a 11 or 12 maybe it winds up in Congquing. But we have no idea what the intermediate species might be, which might make spillovers in Congquing unlikely if those animals aren't farmed near there. And we still have no idea of the mechanism by which SARS-CoV-2 wound up in Guangdong, but we know it did, and it didn't involve a lab.

tripletao 5y ago
It seems like you have a lot more confidence than I do that we know everything the WIV was working on? I agree that SARS-CoV-2 is unlikely to be derived (naturally, artificially, or otherwise) from any known virus, including RaTG13; but a distinctive characteristic of the WIV's research was their sampling of novel viruses from remote bat caves that no other humans routinely entered. That presents a pathway for emergence of novel, naturally-evolved pathogens beyond what normal agricultural etc. activity already risked, whether through a researcher infected in the field or through a literal lab accident back in Wuhan. So specifically considering their group, the novelty of the pathogen doesn't seem to me like significant evidence against lab origin. The public focus on gain-of-function research actually seems excessive to me--while I do think it's possible that the origin of SARS-CoV-2 involves lab manipulation (e.g., genetic engineering or serial passage) of the virus, the WIV's library of novel natural viruses seems to me like at least as big a part of the danger. In any case, their database went offline on Sep 12 2019, around when that genetic clock suggests SARS-Cov-2 began spreading in humans: https://www.researchgate.net/publication/349073738_An_investigation_into_the_WIV_databases_that_were_taken_offline https://www.researchgate.net/publication/349073738_An_invest... That timing could be coincidence, and their flimsy explanation for its continued unavailability ("hacking attempts") could just be the reflexive secrecy of an authoritarian regime. It's a second weird coincidence, though. The WIV has claimed to have published everything they have to GenBank; but Deigin et al. were able to assemble a novel Merbecovirus from contamination of agricultural datasets from a team that shared sequencing equipment with the WIV: https://arxiv.org/abs/2104.01533 https://arxiv.org/abs/2104.01533 That Merbecovirus obviously has nothing to do with SARS-CoV-2; but if they had one unpublished virus, it seems reasonable to wonder if they have more. If you're going to extend the field of zoonoses to include events that didn't become major pandemics, then it seems like you should extend the field of lab leaks too? But let's say negligible probability of lab origin is p < 1/1000, which in an expected value sense is still roughly a 9/11 worth of deaths. Then even taking a prior of p = 1/30 that an arbitrary pandemic is lab-origin and considering the emergence in Wuhan, I don't see how you get to negligible unless the novelty of the pathogen decreases that by a factor of >100 (which per above I don't see).