4 ms·

> People's immune systems differ significantly and some are just stronger than others. Just a nit but stronger is a weird term to describe any functioning immu

> People's immune systems differ significantly and some are just stronger than others. Just a nit but stronger is a weird term to describe any functioning immune system. Some genetic trade-off that makes one immune system better suited to blocking COVID infection might make it worse at something else. > To be infected I have to be infected but you have to be susceptible. The COVID models basically ignored this and assumed a homogenous population. And the data used to drive those models is going to be a result of a real world average across the population, so that is fine for nation level decision making. Let’s say you’re using data from two traced cases, a case with high susceptibility (HS) family members and one with low susceptibility (LS) family members. The HS case was in contact with 4 family members, 3 of which became infected. The LS case was in contact with 4 family members and 1 became infected. Two infections, 50% transmission rate. That is fine for modeling and policy, unless we have some sort of a test to determine if you are HS/LS which could be used to make more granular policies. All of this is moot considering the the different variants we never reached some magical “herd immunity” anyways.

mike_hearn 3y ago
In theory yes. In practice no. The models weren't driven by data in the way you're thinking. Go take a look at their source code if you don't believe me. COVID-Sim had ~150 parameters, virtually all of which were made up guesses. Of the rest they were computed from random things like news reports from Chinese media, friends who happened to work in hospitals, etc. When more data did become available the models were often never updated. I reviewed the code of several models during the pandemic. Beyond having severe bugs, they also frequently used values from January in December or even well into 2021! As more data came in it became clear that the original assumptions were far too pessimistic, but epidemiology doesn't distinguish clearly between assumptions, simulation results and observational data. So they just kept using the old numbers. It got their results in the news and peer reviewers didn't care. > All of this is moot considering the the different variants we never reached some magical “herd immunity” anyways. Well, we don't really know why variants suddenly displaced each other so fast and so completely. It would appear that there's some kind of viral interference, in which you can't be infected with multiple respiratory viruses at once. That doesn't fit with any classical form of germ theory, but the COVID response was based on the assumption of high-school level germ theory being complete, even when collected sample data showed that couldn't be true.
- eightysixfour 3y ago
  > In theory yes. In practice no. The models weren't driven by data in the way you're thinking. Go take a look at their source code if you don't believe me. COVID-Sim had ~150 parameters, virtually all of which were made up guesses. The parameter’s my region’s epidemiologists were using for their models was certainly being updated as new data became available. I even had a conversation with one of them on Reddit about it. > Well, we don't really know why variants suddenly displaced each other so fast and so completely. It would appear that there's some kind of viral interference, in which you can't be infected with multiple respiratory viruses at once. Eh? It is the same natural selection, competitive process at large scale and in the body. You can be infected with multiple variants.

mike_hearn 3y ago

In theory yes. In practice no. The models weren't driven by data in the way you're thinking. Go take a look at their source code if you don't believe me. COVID-Sim had ~150 parameters, virtually all of which were made up guesses. Of the rest they were computed from random things like news reports from Chinese media, friends who happened to work in hospitals, etc. When more data did become available the models were often never updated. I reviewed the code of several models during the pandemic. Beyond having severe bugs, they also frequently used values from January in December or even well into 2021! As more data came in it became clear that the original assumptions were far too pessimistic, but epidemiology doesn't distinguish clearly between assumptions, simulation results and observational data. So they just kept using the old numbers. It got their results in the news and peer reviewers didn't care. > All of this is moot considering the the different variants we never reached some magical “herd immunity” anyways. Well, we don't really know why variants suddenly displaced each other so fast and so completely. It would appear that there's some kind of viral interference, in which you can't be infected with multiple respiratory viruses at once. That doesn't fit with any classical form of germ theory, but the COVID response was based on the assumption of high-school level germ theory being complete, even when collected sample data showed that couldn't be true.

eightysixfour 3y ago
> In theory yes. In practice no. The models weren't driven by data in the way you're thinking. Go take a look at their source code if you don't believe me. COVID-Sim had ~150 parameters, virtually all of which were made up guesses. The parameter’s my region’s epidemiologists were using for their models was certainly being updated as new data became available. I even had a conversation with one of them on Reddit about it. > Well, we don't really know why variants suddenly displaced each other so fast and so completely. It would appear that there's some kind of viral interference, in which you can't be infected with multiple respiratory viruses at once. Eh? It is the same natural selection, competitive process at large scale and in the body. You can be infected with multiple variants.