9 ms·

For the mathematically-minded, it is worth noting that the Asian flu had an R naught of 1.7 while Covid had an R naught of a bit more than 2. Keeping in mind t

For the mathematically-minded, it is worth noting that the Asian flu had an R naught of 1.7 while Covid had an R naught of a bit more than 2. Keeping in mind these are fat-tailed distributions. So an R naught of 2 is a estimate of a true R naught, and because using sample averages to estimate means (true R naught is a mean) is unstable (the fatter the tail, the less moments of the distribution there are), once the reported number starts getting upwards of 2, the less confidence you can have that that reported number represents the true R naught. You have know idea if the superspreaders are infecting 50 or 500 people. At the beginning of the pandemic, we were seeing R naughts estimates at 2.5. That's shockingly high. People were right to take extreme members in the face of that kind of uncertainty. These people enforcing their values with 20-20 hindsight when the decisions were made given limited data at the time are causing harm to our collective ability to make decisions in emergencies.

RansomStark 3y ago
For those that are interest in the nuances of modelling infections, I'm going to leave a Wikipedia extract here. "R_{0} is not a biological constant for a pathogen as it is also affected by other factors such as environmental conditions and the behaviour of the infected population. R_{0} values are usually estimated from mathematical models, and the estimated values are dependent on the model used and values of other parameters. Thus values given in the literature only make sense in the given context and it is recommended not to use obsolete values or compare values based on different models. R_{0} does not by itself give an estimate of how fast an infection spreads in the population". https://en.m.wikipedia.org/wiki/Basic_reproduction_number https://en.m.wikipedia.org/wiki/Basic_reproduction_number
- mike_hearn 3y ago
  > it is recommended not to ... compare values based on different models i.e. R0 numbers are worthless. You can't compare these values between pathogens, you can't even compare them between different academic groups studying the same pathogen or between outbreaks of the same pathogen as modeled by different versions of the same software. There's no agreed way to measure this value empirically. Instead they write a program to simulate the spread of a virus through population and then (in effect) brute force one or more of the inputs until the results of the simulation match what happened so far. These variables are often given names that sorta reflect an intuition about what they're supposed to do, but they're just names. The actual values are arbitrary and there's no effort to ensure the computed values connect to the names in any plausible way. A funny example of this was a COVID model from University College London that used household size as a free variable. The model used this value to achieve curve fit for each country in the study, so they reported that the average Brit lives with seven other people (the paper has 73 citations). https://miro.medium.com/v2/resize:fit:1400/format:webp/1*sEvI1SgBvOydVhtBGN-NAA.png https://miro.medium.com/v2/resize:fit:1400/format:webp/1*sEv... More seriously, estimates of R0 for SARS-CoV-2 have ranged from 1.5 to nearly 6, maybe even higher. It's not even clear what the definition is because it's not time bounded in any way. So not a value that's really knowable, like the speed of light is.
  - selimthegrim 3y ago
    I met the guy (Nick Hengartner) who did the 6 (really 5.7) estimate and knew his colleagues and I can tell you the paper was a lot more careful than that.
    - mike_hearn 3y ago
      Want to explain how they computed it to us? Because if R0 is a real thing then either the people estimating R0=1.something were wrong, or Nick was wrong, and so somewhere some methodologies need to change. I read a fair few modelling papers and R0 (under varying names) was always being used by the model as a free variable used to achieve fit to a historical dataset. It was never being established via lab work, as you might expect given its definition.
      - selimthegrim 3y ago
        https://arxiv.org/abs/2002.03268 https://arxiv.org/abs/2002.03268
        mike_hearn 3y ago
        So they did exactly what we're saying they do? The paper sets up some ad-hoc models unique to this academic group and then reversed them to find values of R0 that could fit. The problem is underdetermined so there are a huge array of values that could work. Key line: > Overall, we report R0 values are likely be between 4.7 and 6.6 with a CI between 2.8 to 11.3 That's an extremely wide CI by any measure. It's a bit unclear how this is meant to be a contradiction. It looks like a good example of the issue. There is no universal theory or method for computing R0. Every single epidemiologist has their own unique approach which is then often discarded in time for the next paper, making the numbers incomparable.
        tripletao 3y ago
        There's some minor variation due to different curve-fitting approaches, but the big variation (e.g. R0 from 3.7–203.3 for measles, per my other comment here) is real, simply because the environment or human behavior varied. As others have repeatedly noted, R0 is a function not just of the pathogen, but also of its environment, including the behavior of its human hosts. Earlier, you wrote: > It [R0] was never being established via lab work, as you might expect given its definition. If you expected that "lab work" could establish R0, then you've grossly misunderstood its meaning and definition. It seems like you're looking for a concept of "R0 but for the pathogen alone, independent of environment and human behavior". That just doesn't exist though, any more than you could define the growth rate of a plant independent of weather and soil fertility.
DrThunder 3y ago
R naught measures transmission not severity of the disease once getting it.
freedude 3y ago
The R Naught is not as complicated as it sounds. If you get infected and pass the virus on to one person, and him to one person, and that person to another person, and this pattern persists throughout society, you have an infection rate of 1. If you pass it on to two people, and on down the line, you have an R Naught of 2. And so on it goes. If it falls below one and finally to 0, the pandemic qualifies as endemic. The infection rate is always conjectural, not really empirical. It’s impossible to discern without universalized, random, and thoroughly accurate testing, tracing, and tracking. Those conditions have never been met in any country or any pandemic. So what seems to be a measure of an existing reality is really true only in theory, not realistically discernible in the midst of a pandemic. At best, it is an estimate. Masks are ineffective at lowering the R Naught. Social Distancing might.

RansomStark 3y ago

For those that are interest in the nuances of modelling infections, I'm going to leave a Wikipedia extract here. "R_{0} is not a biological constant for a pathogen as it is also affected by other factors such as environmental conditions and the behaviour of the infected population. R_{0} values are usually estimated from mathematical models, and the estimated values are dependent on the model used and values of other parameters. Thus values given in the literature only make sense in the given context and it is recommended not to use obsolete values or compare values based on different models. R_{0} does not by itself give an estimate of how fast an infection spreads in the population". https://en.m.wikipedia.org/wiki/Basic_reproduction_number https://en.m.wikipedia.org/wiki/Basic_reproduction_number

mike_hearn 3y ago
> it is recommended not to ... compare values based on different models i.e. R0 numbers are worthless. You can't compare these values between pathogens, you can't even compare them between different academic groups studying the same pathogen or between outbreaks of the same pathogen as modeled by different versions of the same software. There's no agreed way to measure this value empirically. Instead they write a program to simulate the spread of a virus through population and then (in effect) brute force one or more of the inputs until the results of the simulation match what happened so far. These variables are often given names that sorta reflect an intuition about what they're supposed to do, but they're just names. The actual values are arbitrary and there's no effort to ensure the computed values connect to the names in any plausible way. A funny example of this was a COVID model from University College London that used household size as a free variable. The model used this value to achieve curve fit for each country in the study, so they reported that the average Brit lives with seven other people (the paper has 73 citations). https://miro.medium.com/v2/resize:fit:1400/format:webp/1*sEvI1SgBvOydVhtBGN-NAA.png https://miro.medium.com/v2/resize:fit:1400/format:webp/1*sEv... More seriously, estimates of R0 for SARS-CoV-2 have ranged from 1.5 to nearly 6, maybe even higher. It's not even clear what the definition is because it's not time bounded in any way. So not a value that's really knowable, like the speed of light is.
- selimthegrim 3y ago
  I met the guy (Nick Hengartner) who did the 6 (really 5.7) estimate and knew his colleagues and I can tell you the paper was a lot more careful than that.
  - mike_hearn 3y ago
    Want to explain how they computed it to us? Because if R0 is a real thing then either the people estimating R0=1.something were wrong, or Nick was wrong, and so somewhere some methodologies need to change. I read a fair few modelling papers and R0 (under varying names) was always being used by the model as a free variable used to achieve fit to a historical dataset. It was never being established via lab work, as you might expect given its definition.
    - selimthegrim 3y ago
      https://arxiv.org/abs/2002.03268 https://arxiv.org/abs/2002.03268
      - mike_hearn 3y ago
        So they did exactly what we're saying they do? The paper sets up some ad-hoc models unique to this academic group and then reversed them to find values of R0 that could fit. The problem is underdetermined so there are a huge array of values that could work. Key line: > Overall, we report R0 values are likely be between 4.7 and 6.6 with a CI between 2.8 to 11.3 That's an extremely wide CI by any measure. It's a bit unclear how this is meant to be a contradiction. It looks like a good example of the issue. There is no universal theory or method for computing R0. Every single epidemiologist has their own unique approach which is then often discarded in time for the next paper, making the numbers incomparable.
        tripletao 3y ago
        There's some minor variation due to different curve-fitting approaches, but the big variation (e.g. R0 from 3.7–203.3 for measles, per my other comment here) is real, simply because the environment or human behavior varied. As others have repeatedly noted, R0 is a function not just of the pathogen, but also of its environment, including the behavior of its human hosts. Earlier, you wrote: > It [R0] was never being established via lab work, as you might expect given its definition. If you expected that "lab work" could establish R0, then you've grossly misunderstood its meaning and definition. It seems like you're looking for a concept of "R0 but for the pathogen alone, independent of environment and human behavior". That just doesn't exist though, any more than you could define the growth rate of a plant independent of weather and soil fertility.

DrThunder 3y ago

R naught measures transmission not severity of the disease once getting it.

freedude 3y ago

The R Naught is not as complicated as it sounds. If you get infected and pass the virus on to one person, and him to one person, and that person to another person, and this pattern persists throughout society, you have an infection rate of 1. If you pass it on to two people, and on down the line, you have an R Naught of 2. And so on it goes. If it falls below one and finally to 0, the pandemic qualifies as endemic. The infection rate is always conjectural, not really empirical. It’s impossible to discern without universalized, random, and thoroughly accurate testing, tracing, and tracking. Those conditions have never been met in any country or any pandemic. So what seems to be a measure of an existing reality is really true only in theory, not realistically discernible in the midst of a pandemic. At best, it is an estimate. Masks are ineffective at lowering the R Naught. Social Distancing might.