6 ms·

I've upvoted because the article encourages proactive thinking so that we don't just sit in our homes congratulating ourselves how well we socially distanced fr

I've upvoted because the article encourages proactive thinking so that we don't just sit in our homes congratulating ourselves how well we socially distanced from each other. What is sadly missing is some back-of-the-napkin math estimating the required capacity increase. Like, is it closer to x2 or x100? is it feasible at all? What is the damage to economy from doing this compared to lock-down for N months?

PeterisP 7y ago
I believe that nobody really expects to meet the gap between actual capacity and "required" capacity, i.e. that it can't and won't get met - nonetheless, raising the capacity to save lives will save more lives; even if it's not sufficient. It's worth noting that every early "flatten the curve" illustration example that I saw had the lower curve still above the capacity, just not nearly as much, however by the time it reached mass media, it had somehow morphed to the lower curve at or below capacity... probably because they thought it sells the idea better? but at the cost of misleading the public about what's feasible.
- dunkelheit 7y ago
  > I believe that nobody really expects to meet the gap between actual capacity and "required" capacity Well, that's exactly my problem - I don't really know what kind of gap to expect. It is all very qualitative at this point but scenarios (and thus the best course of action) are wildly different depending on how exactly do the relevant quantities compare to each other. Of course "we don't know yet" is a perfectly valid answer but then it must be clearly stated in the article that this is just a speculation describing a plausible scenario.
  - PeterisP 7y ago
    I read some articles that did the math on that, but I can't easily find them now. IIRC it was an order of magnitude difference - i.e. flattening the curve enough to meet capacity would require to stretch it out to something 10 years (certainly an unreasonable goal), so flattening it out over a single year would be something like 10 times over the current capacity. It's hard to estimate how much reasonable actions can increase capacity - 20%? 50%? 100%? But nowhere close to 10x increase.
    - kazagistar 7y ago
      Again, irrelevant. Capacity increase is lives saved. Flattening the curve is a multiplier on that, and is thus also lives saved. "Enough" doesn't matter as much as the fact that "more" scales linearly. Every bed of capacity you add is at least one life saved, even if you never get close to enough.
    - jopsen 7y ago
      flattening the curve also gives us more time to come up with a vaccine. If we're talking about high casualty numbers, stretching it a year or two, while we wait for a vaccine _might_ be feasible.
Zenst 7y ago
One variable that is still not known and would make a huge difference is how many people are asymptomatic and with that catch this and have no symptoms. That effects infection rates and also getting a better grasp upon how many recovered as well as infected. Not just a case of testing those for infection, that cat left the bag in every country on that end. Need to also test for those who are now immune without being aware. Otherwise may well see spreads more than expected due to these people spreading the virus when they have no symptoms and also prevent paralysation of those who are immune, isolating when they don't even know they had it already. So knowing that variable will become key in tackling this and currently we have no idea upon that.
- dougmwne 7y ago
  There have been a couple of population studies to try to measure asymptomatic infections. It appears that about 50%+ of people are symptomatic, so unfortunately significant percentages of people are going to need hospitalization. This study shows the number of symptomatic patients broken out by age group on the Diamond Princess which was a very well-studied population. https://www.medrxiv.org/content/10.1101/2020.03.04.20031104v1.full.pdf https://www.medrxiv.org/content/10.1101/2020.03.04.20031104v...
gus_massa 7y ago
From the numbers of a previous article the answer is x17. https://medium.com/@joschabach/flattening-the-curve-is-a-deadly-delusion-eea324fe9727 https://medium.com/@joschabach/flattening-the-curve-is-a-dea... (HN discussion https://news.ycombinator.com/item?id=22575356 https://news.ycombinator.com/item?id=22575356 ) I prefer a rounded number to not give an illusion of precision, so x20, but there are too many approximations and unknown factor, so it's hopefully closer to x10, but it may be x30. That article was heavily flagged because it has a bad title and a pessimistic message, but it was the first article that I saw with an approximate calculation of the size of the problem.
- dunkelheit 7y ago
  Thanks. Although the author is not an expert and he kind of lost me when he used normal distribution to model the curve, it still looks like the right ballpark. That's a lot.
  - gus_massa 7y ago
    He could have replaced the normal distribution by any other roundly curve and the result doesn't change too much. For example with the SRI model you don't get a Gaussian https://simple.wikipedia.org/wiki/SIR_model https://simple.wikipedia.org/wiki/SIR_model , but the graphic is close enough. It's less peaky than the Gaussian so it should make things easier, but the reduction probably will be a small factor, something between 1 and 2.

PeterisP 7y ago

I believe that nobody really expects to meet the gap between actual capacity and "required" capacity, i.e. that it can't and won't get met - nonetheless, raising the capacity to save lives will save more lives; even if it's not sufficient. It's worth noting that every early "flatten the curve" illustration example that I saw had the lower curve still above the capacity, just not nearly as much, however by the time it reached mass media, it had somehow morphed to the lower curve at or below capacity... probably because they thought it sells the idea better? but at the cost of misleading the public about what's feasible.

dunkelheit 7y ago
> I believe that nobody really expects to meet the gap between actual capacity and "required" capacity Well, that's exactly my problem - I don't really know what kind of gap to expect. It is all very qualitative at this point but scenarios (and thus the best course of action) are wildly different depending on how exactly do the relevant quantities compare to each other. Of course "we don't know yet" is a perfectly valid answer but then it must be clearly stated in the article that this is just a speculation describing a plausible scenario.
- PeterisP 7y ago
  I read some articles that did the math on that, but I can't easily find them now. IIRC it was an order of magnitude difference - i.e. flattening the curve enough to meet capacity would require to stretch it out to something 10 years (certainly an unreasonable goal), so flattening it out over a single year would be something like 10 times over the current capacity. It's hard to estimate how much reasonable actions can increase capacity - 20%? 50%? 100%? But nowhere close to 10x increase.
  - kazagistar 7y ago
    Again, irrelevant. Capacity increase is lives saved. Flattening the curve is a multiplier on that, and is thus also lives saved. "Enough" doesn't matter as much as the fact that "more" scales linearly. Every bed of capacity you add is at least one life saved, even if you never get close to enough.
  - jopsen 7y ago
    flattening the curve also gives us more time to come up with a vaccine. If we're talking about high casualty numbers, stretching it a year or two, while we wait for a vaccine _might_ be feasible.

Zenst 7y ago

One variable that is still not known and would make a huge difference is how many people are asymptomatic and with that catch this and have no symptoms. That effects infection rates and also getting a better grasp upon how many recovered as well as infected. Not just a case of testing those for infection, that cat left the bag in every country on that end. Need to also test for those who are now immune without being aware. Otherwise may well see spreads more than expected due to these people spreading the virus when they have no symptoms and also prevent paralysation of those who are immune, isolating when they don't even know they had it already. So knowing that variable will become key in tackling this and currently we have no idea upon that.

dougmwne 7y ago
There have been a couple of population studies to try to measure asymptomatic infections. It appears that about 50%+ of people are symptomatic, so unfortunately significant percentages of people are going to need hospitalization. This study shows the number of symptomatic patients broken out by age group on the Diamond Princess which was a very well-studied population. https://www.medrxiv.org/content/10.1101/2020.03.04.20031104v1.full.pdf https://www.medrxiv.org/content/10.1101/2020.03.04.20031104v...

gus_massa 7y ago

From the numbers of a previous article the answer is x17. https://medium.com/@joschabach/flattening-the-curve-is-a-deadly-delusion-eea324fe9727 https://medium.com/@joschabach/flattening-the-curve-is-a-dea... (HN discussion https://news.ycombinator.com/item?id=22575356 https://news.ycombinator.com/item?id=22575356 ) I prefer a rounded number to not give an illusion of precision, so x20, but there are too many approximations and unknown factor, so it's hopefully closer to x10, but it may be x30. That article was heavily flagged because it has a bad title and a pessimistic message, but it was the first article that I saw with an approximate calculation of the size of the problem.

dunkelheit 7y ago
Thanks. Although the author is not an expert and he kind of lost me when he used normal distribution to model the curve, it still looks like the right ballpark. That's a lot.
- gus_massa 7y ago
  He could have replaced the normal distribution by any other roundly curve and the result doesn't change too much. For example with the SRI model you don't get a Gaussian https://simple.wikipedia.org/wiki/SIR_model https://simple.wikipedia.org/wiki/SIR_model , but the graphic is close enough. It's less peaky than the Gaussian so it should make things easier, but the reduction probably will be a small factor, something between 1 and 2.