3 ms·

The debate this article addresses might be the best televised debate of all time. The best I've ever seen, at least. Most debates miss out obvious arguments on

by Vecr 3y ago

The debate this article addresses might be the best televised debate of all time. The best I've ever seen, at least. Most debates miss out obvious arguments on both sides. I've got an issue with the article. > "So the only way it could possibly be a lab leak is if the lab leaked sometime in late November, infected exactly one lab worker, that worker went straight to the wet market, infected a vendor, then went home, quarantined, recovered, and all other cases were downstream of that first infected wet market vendor. This is unparsimonious." I don't think that's true, as there's a lot of overdispersion in the spread of this disease. Yeah, I don't think a strict quarantine is likely, but it's not suspicious at all if someone infects 0 or only 1 person.

Scandiravian 3y ago
I don't think I fully understand your argument. Could you expand on it a bit more?
- Vecr 3y ago
  My argument is that infecting zero or one person is common. https://en.wikipedia.org/wiki/Overdispersion https://en.wikipedia.org/wiki/Overdispersion It's why you can get superspreaders without making the r0 a crazy high number. Most people don't transmit much if at all, but a few people transmit a huge amount. You can look at the actual curves if you want.
  - Scandiravian 3y ago
    Thanks for clarifying it for me I think that the argument you're quoting was not so much about how many people would have been infected by this hypothetical patient zero, but that the combined probability of all those events is much smaller than the likelihood of an infected animal transmitting the disease to anyone at the market That's where the whole discussion about 1/10.000 comes from, since it would require the one person that gets infected to be a worker from the wet market compared to a colleague or a random passer-by at another location I think that the main point Peter is making is that probability of transmission from racoon dog * cumulative number of people close to racoon dog >> combined probability of those events If I've misunderstood your point, I'd appreciate if you could point out where I went wrong
    - Vecr 3y ago
      You've got it mostly right but in this case there would be no racoon dog in the chain. It would be a lab leak. I'm about 30% on that, but what I say is a worker from the lab infecting people at the wet market (or infecting someone who goes to the market, etc) takes up most of that probability.
      - Scandiravian 3y ago
        I think I phrased myself poorly, the racoon dog part is the probability where it's a Zoonotic origin The probability of someone getting infected from a racoon dog is much higher than the joined probability required for the lab leak. Someone from the lab only infecting a single person is only a part of the probability of the lab leak, then there's the likelihood of them either going to the market or being close enough to the vendor to cause an infection at some other place I think Peter's argument is that given the amount of conditionals for that to occur it has a low likelihood compared to a racoon dog infecting just one of the people at the market that day
lamontcg 3y ago
Generally if someone only infects 1 person they're going to infect someone that they're close to. The lab leak of SARS-CoV-1 back in 2004 put two of the researchers in the hospital and killed the mother of one of the researchers. That is what a lab leak looks like. The two researchers had taken the train home and nobody really worried about that and there was no subsequent pandemic from that casual contact. So for SARS-CoV-2 you have to add up the probabilities for: 1. This researcher has never been identified [okay, its China so maybe they suppressed this] 2. This researcher did not infect anyone close to them [now you're talking about a lot more suppression] 3. This researcher infected one or more people at one site through casual contact [this is much more unlikely than people think] 4. That infection happened to be exactly in the location most likely for zoonosis (not at a bar, church, restaurant, shopping mall, etc or any other place in the city). [this is quite unlikely, now we're talking many thousands of possible sites] When you compare it against the coincidence that it spilled over in the same city as WIV that isn't going to actually be all that unlikely. You don't have the total area of China in your denominator for probabilities. The virus wasn't terribly good at replicating in humans initially and needed a city and not a rural population. Then you look at cities in southern/central china near where the bats are and the odds are probably closer to 1-in-12 than 1-in-a-million. It also isn't any coincidence that WIV was built closer to the bats and not in Beijing. So, SARS-CoV-1 spilled over in Foshan in the 12th largest city in China in Guangdong province, while SARS-CoV-2 spilled over in Wuhan in the 7th largest city in China in the Hubei province. Those are places you expect for a pandemic epicenter. As an exercise, roll 1d12 (or 1d20 if you're more skeptical) twice and if you get a '1' on either throw there's your coincidental spillover in the same city as WIV. A bit weird, but not even p < 0.05. Stuff like that happens.
- Vecr 3y ago
  > 1. This researcher has never been identified [okay, its China so maybe they suppressed this] > 2. This researcher did not infect anyone close to them [now you're talking about a lot more suppression] >3. This researcher infected one or more people at one site through casual contact [this is much more unlikely than people think] See, this is what I have a problem with. It assumes that the symptoms this person/these people had were bad enough for samples to be tested later (pretty much hospitalization at that point, or maybe a hospital visit if they saved samples for some other reason), and it assumes things about the transmission of the virus that I'm not sure were correct at the time. It's a different disease from SARS, it's demonstrably harder to track and contain. People don't need obvious symptoms to transmit, and by the time most people end up in the hospital the transmission probability has gone down, due to various things including better precautions, but mostly because their immune system has started clearing the virus. What kills people most of the time is off-target immune responses, not the virus itself.

by Vecr 3y ago

Scandiravian 3y ago

I don't think I fully understand your argument. Could you expand on it a bit more?

Vecr 3y ago
My argument is that infecting zero or one person is common. https://en.wikipedia.org/wiki/Overdispersion https://en.wikipedia.org/wiki/Overdispersion It's why you can get superspreaders without making the r0 a crazy high number. Most people don't transmit much if at all, but a few people transmit a huge amount. You can look at the actual curves if you want.
- Scandiravian 3y ago
  Thanks for clarifying it for me I think that the argument you're quoting was not so much about how many people would have been infected by this hypothetical patient zero, but that the combined probability of all those events is much smaller than the likelihood of an infected animal transmitting the disease to anyone at the market That's where the whole discussion about 1/10.000 comes from, since it would require the one person that gets infected to be a worker from the wet market compared to a colleague or a random passer-by at another location I think that the main point Peter is making is that probability of transmission from racoon dog * cumulative number of people close to racoon dog >> combined probability of those events If I've misunderstood your point, I'd appreciate if you could point out where I went wrong
  - Vecr 3y ago
    You've got it mostly right but in this case there would be no racoon dog in the chain. It would be a lab leak. I'm about 30% on that, but what I say is a worker from the lab infecting people at the wet market (or infecting someone who goes to the market, etc) takes up most of that probability.
    - Scandiravian 3y ago
      I think I phrased myself poorly, the racoon dog part is the probability where it's a Zoonotic origin The probability of someone getting infected from a racoon dog is much higher than the joined probability required for the lab leak. Someone from the lab only infecting a single person is only a part of the probability of the lab leak, then there's the likelihood of them either going to the market or being close enough to the vendor to cause an infection at some other place I think Peter's argument is that given the amount of conditionals for that to occur it has a low likelihood compared to a racoon dog infecting just one of the people at the market that day

lamontcg 3y ago

Generally if someone only infects 1 person they're going to infect someone that they're close to. The lab leak of SARS-CoV-1 back in 2004 put two of the researchers in the hospital and killed the mother of one of the researchers. That is what a lab leak looks like. The two researchers had taken the train home and nobody really worried about that and there was no subsequent pandemic from that casual contact. So for SARS-CoV-2 you have to add up the probabilities for: 1. This researcher has never been identified [okay, its China so maybe they suppressed this] 2. This researcher did not infect anyone close to them [now you're talking about a lot more suppression] 3. This researcher infected one or more people at one site through casual contact [this is much more unlikely than people think] 4. That infection happened to be exactly in the location most likely for zoonosis (not at a bar, church, restaurant, shopping mall, etc or any other place in the city). [this is quite unlikely, now we're talking many thousands of possible sites] When you compare it against the coincidence that it spilled over in the same city as WIV that isn't going to actually be all that unlikely. You don't have the total area of China in your denominator for probabilities. The virus wasn't terribly good at replicating in humans initially and needed a city and not a rural population. Then you look at cities in southern/central china near where the bats are and the odds are probably closer to 1-in-12 than 1-in-a-million. It also isn't any coincidence that WIV was built closer to the bats and not in Beijing. So, SARS-CoV-1 spilled over in Foshan in the 12th largest city in China in Guangdong province, while SARS-CoV-2 spilled over in Wuhan in the 7th largest city in China in the Hubei province. Those are places you expect for a pandemic epicenter. As an exercise, roll 1d12 (or 1d20 if you're more skeptical) twice and if you get a '1' on either throw there's your coincidental spillover in the same city as WIV. A bit weird, but not even p < 0.05. Stuff like that happens.

Vecr 3y ago
> 1. This researcher has never been identified [okay, its China so maybe they suppressed this] > 2. This researcher did not infect anyone close to them [now you're talking about a lot more suppression] >3. This researcher infected one or more people at one site through casual contact [this is much more unlikely than people think] See, this is what I have a problem with. It assumes that the symptoms this person/these people had were bad enough for samples to be tested later (pretty much hospitalization at that point, or maybe a hospital visit if they saved samples for some other reason), and it assumes things about the transmission of the virus that I'm not sure were correct at the time. It's a different disease from SARS, it's demonstrably harder to track and contain. People don't need obvious symptoms to transmit, and by the time most people end up in the hospital the transmission probability has gone down, due to various things including better precautions, but mostly because their immune system has started clearing the virus. What kills people most of the time is off-target immune responses, not the virus itself.