Here’s a link to a very educational cartoon where it shows how damned complicated it is to come up with trustworthy answers. It’s put together by a team at 538 along with Mr. Weinersmith, creator of SMBC (a webcomic I follow as closely as I follow XKCD).
Now, time for me to go back to pandemic modeling for my patient base. Apologies if this has been posted and discussed elsewhere here recently. I’m keeping my nose out of most of the COVID stuff here because COVID itself is eating up my entire life and my dreams and I don’t have time or the emotional bandwith to cope with certain autodidacts who feel they grasp the situation and want to explain how it ought to be dealt with. This stuff is so fucking complicated, people.
Thanks to all for providing diversion for me in other threads here. Quality time here, in my hot tub, and on my bidet all help me cope with the stress of work and life and TP shortage.
Here is also the FiveThirtyEight Politics podcast with a discussion between Maggie Koerth, Laura Bronner, and Galen Druke on “Why Forecasting COVID-19 Is Harder Than Forecasting Elections” which covers much of the same issues in terms of uncertainties. FiveThirtyEight.com has actually produced an excellent series of articles in the last few weeks on many of the ongoing issues with trying to deal with the epidemic including Clare Malone’s article on how ”New York’s Inequalities Are Fueling COVID-19” which has implications for not only how other densely populated cities may experience outbreaks but also for prisons and other institutions where effective isolation is not possible. They’ve also done some work trying to address the longer term socioeconomic implications of the pandemic and lockdown, but this is such an unprecedented situation in modern society that they admit that what happens in the future is really guesswork, and depends strongly on the foresight and resolve of political leaders.
Take care of yourself, Qadgop the Mercotan. I know that your patients are particularly vulnerable and underserved because of their status, and are totally reliant on you and your staff for any support or treatment. But don’t take any risks you can avoid.
I wish they had put all that information into a bulletized list or FAQ. A pages-long web comic is really not the best medium for a data dump.
Does anyone happen to have access to all that information in a nice bulletized list that I can quickly read through without having to dance around all the pictures?
I had seen this before, but it was well worth another look.
Qadgop, I’ve thought of you often during this crisis. Our governor released a thousand nonviolent offenders from prison (with the expected blowback), and I wondered if your state was doing the same. I hope you and your family are safe.
Sorry about the blowhards here who’ve made it frustrating for you to post. You’re much more widely respected than they are, but I’m sure it’s aggravating to have to read the uninformed crud.
TL;DR: Accurate models require accurate input data, but we do not have accurate input data, because we simply do not know enough about the virus and disease right now.
[ul]
[li]Covid-19 kills some people <— This is a fact, and we know it to be true.[/li][li]We don’t know the death rate[/li][LIST]
[li]We have a good idea of how many people have died from it, but certainly not exact.[/li][li]We don’t know how many people have had it.[/li][li]The death rate is different in different populations, so even when we have a good estimate in one place (e.g. the Diamond Princess) it doesn’t necessarily apply to other places.[/li][li]As the medical system in any area reaches saturation, the death rate in that area will go up.[/li][/ul]
[li]We don’t know the infection rate (R0)[/li][ul]
[li]We don’t even have good information on the variables needed to estimate the infection rate.[/li][li]Rate of contact changes from place to place and person to person.[/li][li]Traits of the virus, such as how long it lives on surfaces and how it spreads in the air, are still only estimates.[/li][li]Some people may contract it easier than others.[/li][li]During an illness, we don’t know how long it is contagious, or when it is most contagious.[/li][/ul]
[li]We don’t know the effective reproduction rate[/li][ul]
[li]We don’t know about post-illness immunity[/li][/ul]
[li]All of those things we don’t know can be combined to make models with wildly different predictions[/li][/LIST]
My own editorialization: Of course Internet pundits can say why any particular estimates might be wrong, because they are all just estimates at this point. The people doing those estimates can probably think of even more reasons why they are wrong. Choosing a model because it supports the outcome or behavior you desire is not a recognized as a reliable way of choosing the best model. We are months behind where we should be in testing. The only thing that can answer many of these questions is lots and lots of testing. That will take time, like waiting for the next season of GoT kind of time.
I really appreciate that summary. Somewhat in line with your editorialization, I find it ironic that so many people who denounced the “do nothing” models as wildly inaccurate and not worth changing our behavior over now point to the “hey, we did something and it’s working” models as rock-solid predictions for the future and proof positive that the virus has run its course and we can lift all these restrictions, completely missing the point that behavior has the ability to affect the reliability of those very models.
