3 ms·

In a math classroom, you are right. But when we're in a context of talking about the disease on an internet forum, you are being needlessly pedantic and extreme

by stnmtn 5y ago

In a math classroom, you are right. But when we're in a context of talking about the disease on an internet forum, you are being needlessly pedantic and extremely nitpick-y if you expect people to describe the ways infections grows as "similar to a sigmoid function" rathan than the well-understood and extremely common "exponential growth" term.

oneoff786 5y ago
I don’t really care if you say sigmoid when you mean exponential, my point is that both of those are wrong. People shudder in fear of the phrase “exponential” but disease does not spread exponentially. It never does. That’s a laymen fairy tail and it does a great deal of harm in understanding the disease. Exponential is not well understood. People don’t know what it means, as evidenced by the fact that they consistently use it incorrectly.
- renaudg 5y ago
  >disease does not spread exponentially. It never does. That’s a laymen fairy tail and it does a great deal of harm in understanding the disease. >Exponential is not well understood. People don’t know what it means, as evidenced by the fact that they consistently use it incorrectly. Well, in typical Dunning-Krüger fashion you're lecturing us laymen about the wrong use of "exponential", but it seems that you are the layman when it comes to epidemiology modelling. Diseases of course spread in an initially exponential phase, not logarithmically as you boldly claim in another comment. Epidemiologists talk about exponential disease growth all the time, here's one example : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6962332/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6962332/
  - refytedy 5y ago
    Not the original commenter and agree with your sentiment. That said, from what we know about random graph models, once the transmission rate passes a certain threshold, the cumulative number of eventually infected individuals will NOT scale exponentially as the transmission rate increases. You're speaking about initial growth rate though, not convergent size of the once-infected population, which is definitely exponential and relevant if you care about minimizing simultaneous hospitalizations.
  - oneoff786 5y ago
    A brief initial exponential phase that doesn’t explain the average state of the system is a poor mental model. Epidemiologists do frequently use exponential growth in discussions of theoretical spreading, but they are not correct. It does not fucking matter. They’re just wrong. Is the long term spread of a disease exponential. No. We can say this with 100% confidence. Why? Because they do not increase exponentially over time. They QUICKLY hit ceilings and inflection points. This is the problem. If you call a system exponential because it looks like that at the beginning, your estimates of the middle and end will be hugely off. Massively wrong. It is obviously apparent looking at Covid that it does not spread exponentially. The impact of this understanding is enormous. In an exponential model, a virus that is twice as infectious could be expected to infect 4x as many people over the same period. If the model is logarithmic, twice as infectious will likely result in LESS THAN TWICE the number of people infected. Scary predictions that a small increase in spreading will exponentially increase the number of infected people do not hold up against observed real world data. We should be cheering on a more spreadable but less deadly disease. Scale matters, of course, but we aren’t seeing anything like that. If you want to come back to this in one year and check to see if omicron infected even twice the number of people that delta did over the same time period that’s a bet I’d surely take.

by stnmtn 5y ago

oneoff786 5y ago

I don’t really care if you say sigmoid when you mean exponential, my point is that both of those are wrong. People shudder in fear of the phrase “exponential” but disease does not spread exponentially. It never does. That’s a laymen fairy tail and it does a great deal of harm in understanding the disease. Exponential is not well understood. People don’t know what it means, as evidenced by the fact that they consistently use it incorrectly.

renaudg 5y ago
>disease does not spread exponentially. It never does. That’s a laymen fairy tail and it does a great deal of harm in understanding the disease. >Exponential is not well understood. People don’t know what it means, as evidenced by the fact that they consistently use it incorrectly. Well, in typical Dunning-Krüger fashion you're lecturing us laymen about the wrong use of "exponential", but it seems that you are the layman when it comes to epidemiology modelling. Diseases of course spread in an initially exponential phase, not logarithmically as you boldly claim in another comment. Epidemiologists talk about exponential disease growth all the time, here's one example : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6962332/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6962332/
- refytedy 5y ago
  Not the original commenter and agree with your sentiment. That said, from what we know about random graph models, once the transmission rate passes a certain threshold, the cumulative number of eventually infected individuals will NOT scale exponentially as the transmission rate increases. You're speaking about initial growth rate though, not convergent size of the once-infected population, which is definitely exponential and relevant if you care about minimizing simultaneous hospitalizations.
- oneoff786 5y ago
  A brief initial exponential phase that doesn’t explain the average state of the system is a poor mental model. Epidemiologists do frequently use exponential growth in discussions of theoretical spreading, but they are not correct. It does not fucking matter. They’re just wrong. Is the long term spread of a disease exponential. No. We can say this with 100% confidence. Why? Because they do not increase exponentially over time. They QUICKLY hit ceilings and inflection points. This is the problem. If you call a system exponential because it looks like that at the beginning, your estimates of the middle and end will be hugely off. Massively wrong. It is obviously apparent looking at Covid that it does not spread exponentially. The impact of this understanding is enormous. In an exponential model, a virus that is twice as infectious could be expected to infect 4x as many people over the same period. If the model is logarithmic, twice as infectious will likely result in LESS THAN TWICE the number of people infected. Scary predictions that a small increase in spreading will exponentially increase the number of infected people do not hold up against observed real world data. We should be cheering on a more spreadable but less deadly disease. Scale matters, of course, but we aren’t seeing anything like that. If you want to come back to this in one year and check to see if omicron infected even twice the number of people that delta did over the same time period that’s a bet I’d surely take.