5 ms·

Except that Japan has high population density, no significant lockdown, contact tracing that misses ~80% of the cases (estimating the total case count from thei

Except that Japan has high population density, no significant lockdown, contact tracing that misses ~80% of the cases (estimating the total case count from their death count, using the IFR from serology elsewhere), and 7 deaths per million. So maybe it's the masks? The general culture of hygiene? The weather? Something else? There's a lot we don't know here. I wish people would admit and accept that, instead of looking for false confidence, finding it, and then arguing with people who found a different falsely confident answer.

akiselev 6y ago
> Except that Japan has high population density, no significant lockdown, contact tracing that misses ~80% of the cases (estimating the total case count from their death count, using the IFR from serology elsewhere), and 7 deaths per million. So maybe it's the masks? The general culture of hygiene? The weather? Something else? According to Worldometer Japan has tested ~284k people while New York and California have combined tested almost four million despite together having half the population of Japan. Last time I saw excess mortality statistics a few weeks ago they were lagging with several prefectures missing and bad normalization for aging and other factors. Assuming the numbers are accurate, I'm more interested in how South Dakota and Wyoming are doing so much worse than Japan with such a low population density. I suspect the answer lies in the crumbling rural health infrastructure. > There's a lot we don't know here. I wish people would admit and accept that, instead of turning this into yet another polarized political debate. Mortality, population density, and basic division are not political and if by themselves they're polarizing, then we're too far gone already.
- tripletao 6y ago
  I'm not sure what you intend by those test counts? Japan has 874 dead. Assuming IFR of 1%, that implies about 87k infected. But they report 17k confirmed cases, so they're missing ~80% of the cases. I mentioned this to head off any suggestion that Japan's trace and test operation was responsible for their success, since at best it's decreasing R by that ~20%. Perhaps Japan is failing to catch some deaths too; but they're an open society with freedom of speech, so I don't expect their numbers to be grossly wrong. So what could South Dakota and Wyoming's public health system be doing that's worse than Japan? I don't know the answer, but it's not clear to me at all that the difference is in their government responses vs. some aspect of their initial situations. I gave deaths per capita because that's the probability that someone in that state will have died of coronavirus, which is what most people care about directly. Your (deaths/people)/(people/area) has no corresponding direct significance. To then call that "the relevant number" overstates the importance of population density, and also models population density badly when it's non-uniform (as you note for NYC). I agree that population density is an important factor, just not as determinative as your post would seem to imply. Oh, and looks like you quoted my post before I edited it, sorry. But I believe false confidence is polarizing and would like to see less of it, as uncomfortable as the true uncertainty may be.
  - akiselev 6y ago
    > I'm not sure what you intend by those test counts? Japan has 874 dead. I'm saying I believe their numbers are inaccurate because they're undertesting and stats I could use to validate them (overall number of deaths to calculate seasonal excess mortality) are also currently inaccurate for unrelated reasons. I agree that the number of deaths will be impossible to hide in an open society like Japan but for bureaucratic reasons it will take a while to compile and validate those stats before normalizing them enough to compare between years let alone nations. Until then, I'm not convinced and any argument based on Japan as an example is also suspect. > To then call that "the relevant number" overstates the importance of population density, and also models population density badly when it's non-uniform (as you note for NYC). Uhm, what? Deaths per million lumps Wyoming in with New York City, making it even less accurate to what people care about. Manhattan will not be reopening on the same schedule as some town on the outskirts of Fargo. Someone in the middle of a dense city does not face the same risk as someone whose closest neighbor is 2 miles away. Also, it's a virus. There's almost a century of medicine and public health expertise that shows that all things being equal, density is the dominant factor. I thought ending quarantine is what people cared about. Your false confidence about Japan's comparability to the US is not helping. Population density is why NYC and Wyoming aren't comparable yet you're trying to compare another country.
    - tripletao 6y ago
      So what do you think explains the difference between Japan and NY? It's not population density. Even if Japan's death count is understated, there is zero chance that all of Japan's medical staff and press have somehow failed to notice and report NYC-level mortality. And I'm not sure what you think is my "false confidence about Japan's comparability to the US"? I certainly don't believe Japan's response would work equally well in NY, considering that NY applied stricter measures and ended up with 200x the death rate. I believe the susceptibility of individual regions to the coronavirus varies greatly with factors we don't yet understand, including population density, cultural and behavioral factors, weather, maybe air pollution, maybe genetics. I'd guess these factors explain much more of the variation in mortality by region than variation in the governments' responses, though I'm not sure. And you're missing my point about deaths per capita. That's an intrinsically meaningful number, because people care about the probability that they die (or get seriously sick, etc.). Other numbers are valuable only to the extent they predict something people intrinsically care about. Population density is one such input to the prediction, but it's clearly not the only factor, and your simplistic focus on that isn't useful.

akiselev 6y ago

> Except that Japan has high population density, no significant lockdown, contact tracing that misses ~80% of the cases (estimating the total case count from their death count, using the IFR from serology elsewhere), and 7 deaths per million. So maybe it's the masks? The general culture of hygiene? The weather? Something else? According to Worldometer Japan has tested ~284k people while New York and California have combined tested almost four million despite together having half the population of Japan. Last time I saw excess mortality statistics a few weeks ago they were lagging with several prefectures missing and bad normalization for aging and other factors. Assuming the numbers are accurate, I'm more interested in how South Dakota and Wyoming are doing so much worse than Japan with such a low population density. I suspect the answer lies in the crumbling rural health infrastructure. > There's a lot we don't know here. I wish people would admit and accept that, instead of turning this into yet another polarized political debate. Mortality, population density, and basic division are not political and if by themselves they're polarizing, then we're too far gone already.

tripletao 6y ago
I'm not sure what you intend by those test counts? Japan has 874 dead. Assuming IFR of 1%, that implies about 87k infected. But they report 17k confirmed cases, so they're missing ~80% of the cases. I mentioned this to head off any suggestion that Japan's trace and test operation was responsible for their success, since at best it's decreasing R by that ~20%. Perhaps Japan is failing to catch some deaths too; but they're an open society with freedom of speech, so I don't expect their numbers to be grossly wrong. So what could South Dakota and Wyoming's public health system be doing that's worse than Japan? I don't know the answer, but it's not clear to me at all that the difference is in their government responses vs. some aspect of their initial situations. I gave deaths per capita because that's the probability that someone in that state will have died of coronavirus, which is what most people care about directly. Your (deaths/people)/(people/area) has no corresponding direct significance. To then call that "the relevant number" overstates the importance of population density, and also models population density badly when it's non-uniform (as you note for NYC). I agree that population density is an important factor, just not as determinative as your post would seem to imply. Oh, and looks like you quoted my post before I edited it, sorry. But I believe false confidence is polarizing and would like to see less of it, as uncomfortable as the true uncertainty may be.
- akiselev 6y ago
  > I'm not sure what you intend by those test counts? Japan has 874 dead. I'm saying I believe their numbers are inaccurate because they're undertesting and stats I could use to validate them (overall number of deaths to calculate seasonal excess mortality) are also currently inaccurate for unrelated reasons. I agree that the number of deaths will be impossible to hide in an open society like Japan but for bureaucratic reasons it will take a while to compile and validate those stats before normalizing them enough to compare between years let alone nations. Until then, I'm not convinced and any argument based on Japan as an example is also suspect. > To then call that "the relevant number" overstates the importance of population density, and also models population density badly when it's non-uniform (as you note for NYC). Uhm, what? Deaths per million lumps Wyoming in with New York City, making it even less accurate to what people care about. Manhattan will not be reopening on the same schedule as some town on the outskirts of Fargo. Someone in the middle of a dense city does not face the same risk as someone whose closest neighbor is 2 miles away. Also, it's a virus. There's almost a century of medicine and public health expertise that shows that all things being equal, density is the dominant factor. I thought ending quarantine is what people cared about. Your false confidence about Japan's comparability to the US is not helping. Population density is why NYC and Wyoming aren't comparable yet you're trying to compare another country.
  - tripletao 6y ago
    So what do you think explains the difference between Japan and NY? It's not population density. Even if Japan's death count is understated, there is zero chance that all of Japan's medical staff and press have somehow failed to notice and report NYC-level mortality. And I'm not sure what you think is my "false confidence about Japan's comparability to the US"? I certainly don't believe Japan's response would work equally well in NY, considering that NY applied stricter measures and ended up with 200x the death rate. I believe the susceptibility of individual regions to the coronavirus varies greatly with factors we don't yet understand, including population density, cultural and behavioral factors, weather, maybe air pollution, maybe genetics. I'd guess these factors explain much more of the variation in mortality by region than variation in the governments' responses, though I'm not sure. And you're missing my point about deaths per capita. That's an intrinsically meaningful number, because people care about the probability that they die (or get seriously sick, etc.). Other numbers are valuable only to the extent they predict something people intrinsically care about. Population density is one such input to the prediction, but it's clearly not the only factor, and your simplistic focus on that isn't useful.