It would be very valuable to take a particular “Alarm” and see how many true positives, false positives, true negatives and false negatives it would have produced over the past 20 years.
I did this with a few diseases! Two investigations got their own post, which are linked above. Others just got mentions in the v. 2 post, linked at the very top.
The problem here os “fog of war”: we can’t know for sure all R0, rout of transmission and other parameters for sure before a pandemic will reach high stages. This will result either in the frequent false alarms, or there will be no early warning.
I updated the post to distinguish between respiratory droplets and a fully airborne route of transmission.
We still don’t know whether COVID-19 (or SARS) is transmitted only through respiratory droplets, or whether it is airborne more broadly in smaller droplets beyond a range of 6′ or so. If that’s the key difference between a widely transmissible disease and one that’s less of a threat, and there’s a lot of variability among respiratory illnesses, then that would make it less likely a priori that a novel respiratory illness is highly contagious.
On the other hand, it might be that respiratory droplets vs. fully airborne is not the key difference, or of most diseases spreadable by droplets are also aerosolized.
To tell what we’re dealing with, we might look for suggestive case studies. There was a choir that met in the early days of COVID-19 where none of the members were sick, they all stayed 6′ apart, but half the choir still caught COVID-19 from each other. This suggests COVID-19 is fully airborne, though of course it’s not hard evidence.
But in general, this model is designed to help with ‘fog of war’. Since we can’t know these factors for sure, we use what evidence is available to reason under uncertainty. My historical research both into COVID-19 and historical diseases suggested to me that this model is fairly well-calibrated, but there just aren’t enough data points to know for sure. Even if not, though, it at least serves as guideposts for future reasoning, and I think that’s valuable.
We still don’t know whether COVID-19 (or SARS) is transmitted only through respiratory droplets, or whether it is airborne more broadly in smaller droplets beyond a range of 6′ or so.
And if it is airbone in smaller droplets whether it is outside of specific medical procedures airbone in that way.
This doesn’t mean those pandemics, with their millions of lives lost, were unimportant. It simply means that they did not suddenly and severely shake the world economy, the way COVID-19 is doing right now.
Not sure this comment changes things much but would point out that the world economies in 2020 were a lot more integrated than in just about all other periods of time one could consider pandemic impacts. In that sense I think the impact, and particularly to things like financial asset prices, may be reduced in a less integrated world.
What might be more relevant here, but I’m not sure how to apply any adjustment factors to your x/14 scale approach is how the observation of tight integration (think all supply chain roads lead to China) and the degree of decoupling that seems to be occurring would impact some of those points.
But I do like the approach in that it does kind of keep it simple in approach and could be applied by anyone that just wanted to think about things without have a good background in math or modeling.
I suspect that the problem isn’t a lack of data as much as it is a problem of remaining constantly vigilant (allusion intended) and being willing to pay the price of false positives. For example, if H7N9 scored highly initially, would you be OK with your family thinking that you are a fool for selling your retirement funds? This is getting into the topic of “Shut up and multiply.”
It would be very valuable to take a particular “Alarm” and see how many true positives, false positives, true negatives and false negatives it would have produced over the past 20 years.
I did this with a few diseases! Two investigations got their own post, which are linked above. Others just got mentions in the v. 2 post, linked at the very top.
The problem here os “fog of war”: we can’t know for sure all R0, rout of transmission and other parameters for sure before a pandemic will reach high stages. This will result either in the frequent false alarms, or there will be no early warning.
I updated the post to distinguish between respiratory droplets and a fully airborne route of transmission.
We still don’t know whether COVID-19 (or SARS) is transmitted only through respiratory droplets, or whether it is airborne more broadly in smaller droplets beyond a range of 6′ or so. If that’s the key difference between a widely transmissible disease and one that’s less of a threat, and there’s a lot of variability among respiratory illnesses, then that would make it less likely a priori that a novel respiratory illness is highly contagious.
On the other hand, it might be that respiratory droplets vs. fully airborne is not the key difference, or of most diseases spreadable by droplets are also aerosolized.
To tell what we’re dealing with, we might look for suggestive case studies. There was a choir that met in the early days of COVID-19 where none of the members were sick, they all stayed 6′ apart, but half the choir still caught COVID-19 from each other. This suggests COVID-19 is fully airborne, though of course it’s not hard evidence.
But in general, this model is designed to help with ‘fog of war’. Since we can’t know these factors for sure, we use what evidence is available to reason under uncertainty. My historical research both into COVID-19 and historical diseases suggested to me that this model is fairly well-calibrated, but there just aren’t enough data points to know for sure. Even if not, though, it at least serves as guideposts for future reasoning, and I think that’s valuable.
And if it is airbone in smaller droplets whether it is outside of specific medical procedures airbone in that way.
Not sure this comment changes things much but would point out that the world economies in 2020 were a lot more integrated than in just about all other periods of time one could consider pandemic impacts. In that sense I think the impact, and particularly to things like financial asset prices, may be reduced in a less integrated world.
What might be more relevant here, but I’m not sure how to apply any adjustment factors to your x/14 scale approach is how the observation of tight integration (think all supply chain roads lead to China) and the degree of decoupling that seems to be occurring would impact some of those points.
But I do like the approach in that it does kind of keep it simple in approach and could be applied by anyone that just wanted to think about things without have a good background in math or modeling.
You’ve inspired me to think about what goes into the mental model of someone who does this professionally such as an epidemiologist or https://firstwatch.net/category/health-intelligence/outbreaks/, which a coworker maintains (conflict of interest alert).
I suspect that the problem isn’t a lack of data as much as it is a problem of remaining constantly vigilant (allusion intended) and being willing to pay the price of false positives. For example, if H7N9 scored highly initially, would you be OK with your family thinking that you are a fool for selling your retirement funds? This is getting into the topic of “Shut up and multiply.”