Customized COVID-19 risk analysis as a high value area

post by Askell · 2020-07-23T20:28:55.516Z · score: 40 (17 votes) · EA · GW · 6 comments

At the moment, I think that many effective altruists are spending a lot of their time doing COVID risk analyses part-time. This is for a couple of reasons: people in the EA community are more likely to have the kind of disposition that makes you want to dig into the research on COVID-19, and many EAs live in housing situations that are not always accounted for in the current recommendations (e.g. many live in the bay area where there are a lot of group houses).

There are a couple of major downsides of people doing this work themselves: it's time-costly, and it is mentally draining and stressful. It's also wasteful if a lot of this analysis work ends up getting replicated privately across many people.

At the same time, if households don't have ways of analyzing risks and deciding on acceptable behaviors, it's pretty likely that people will get complacent and make decision that put people in the house at risk, and so the cycle continues.

Here's a thing I'd like to see exist: a small group of people with relevant experience coming together to produce trustworthy COVID-19 related risk evaluations in a central place that people in the EA community can look to. This might include:

• Keeping up with the latest research on behavior risks and presenting this in a way that is easy for individuals and households to digest (e.g. fitting them into neat categories or levels of risk, comparing different risks)
• Writing weekly summaries of the outlook, progress on vaccines and treatments, etc. so that people are less likely to compulsively news search
• Creating template quarantine procedures for houses of 5, 10, 20 people, etc.
• Creating template "scenario plans" that could be used if someone gets sick, ot if infection rates rise above a certain amount in an area, etc.
• Having somewhat regular Q&As to answer questions people might have, as soliciting requests for content that people would find useful

It could also be useful to offer specific things on a contracting basis (since it probably doesn't make sense to collectively fund private information), e.g.:

• Reviewing existing house quarantine procedures or personal plans for gaps, inconsistencies, etc.
• Being on-call to run an emergency risk evaluation of a quarantine violation or to answer urgent questions

The idea is that this information would be trustworthy, up-to-date, and tailored to regions and housing situations that people in the EA community are in but that might not be taken into account by existing analyses. I think it would be relatively important that this doesn't involve giving advice, but instead involves the categorization and explanation of risks, and lets individuals and households set their own level of acceptable risk.

This would reduce the time people need to spend evaluating COVID-19 risks, it would help people to act in accordance with their own risk tolerance, it would create publicly accessible information that's available to people outside of the EA community, and it would reduce stress around COVID-19 risk analysis.

How would I do this? I'd probably have one of the EA organizations hire a couple of full time researchers who have already been looking into COVID-19 a lot and can show that their work is reliable, as well as someone who can help present this information well on a website that people can be directed to. If the project wasn't going to offer bespoke advice on a contracting basis, I might also make this a place where people can offer/request these kinds of risk analysis services. (I guess an MVP of this whole project might be for people who have done a lot of COVID-19 research to offer to do risk analysis contracting.)

People might object that this information already exists, but I haven't been able to find a place that brings it together in the way I outline here, and my hunch is that a lot of people are still spending a lot of time searching it out. But I hope people will link to such resources in the comments if I'm wrong.

Anyway, this is all just an idea that I have and perhaps I'm mistaken about its value. I'm posting here to get a general sense of whether this is something that other EAs think would be valuable, how it could be improved, ways in which it's a terrible idea, and so on.

6 comments

Comments sorted by top scores.

comment by catherio · 2020-07-24T02:31:07.258Z · score: 15 (9 votes) · EA(p) · GW(p)

Strong agree!

Our house has a custom model & "points system" for group houses that we are working as fast as we can to release legibly (constrained by most of us having full-time jobs etc.).

(Basically: quantifying different activities in terms of microcovids i.e. literal 1-in-a-million chance of getting COVID)

We are already spending our spare time on informal custom consults. We really don't want to be doing this as an ongoing "job" in the long run, but we really DO want the information to be out there to help people.

We would LOVE LOVE LOVE to train people in what we know, so they can run consults!

Please reach out to me (ideally on messenger, or my username @mit.edu) if you would like to help us turn our V1 version into something scalably usable by EAs & group houses!


In the meantime, we have a stack of sources here that feed into our microcovid estimates, which you can consult: https://docs.google.com/presentation/d/1pkYTQA5hR-52pUbfGqjTFR1yieocOmEj1O0RnV2x6jI/edit#slide=id.g8b0e9b222b_0_179


Creating template "scenario plans" that could be used if someone gets sick, ot if infection rates rise above a certain amount in an area, etc.

Here is Ibasho's isolation protocol, with clear criteria about what counts as "sick" etc: https://docs.google.com/document/d/1OpWFFoUB4gULcpOZMZL4A1xdDfu-qvgSQDCFi8dXa6E/edit#


Writing weekly summaries of the outlook, progress on vaccines and treatments, etc. so that people are less likely to compulsively news search

We are already doing this and sharing on a messenger thread. Anywhere already existing that we could post them to be useful?

comment by Davidmanheim · 2020-07-24T11:14:46.150Z · score: 4 (4 votes) · EA(p) · GW(p)

This sounds slightly related to something 1DaySooner is just starting, which is a risk model for a HCT, which will look at the risk of death, and hopefully also of long term disability. Ideally, it would also consider the probability conditional on rescue therapies being available or becoming available. To do that, we're focusing on a population subset, but the basis for the model is data that includes multiple ages, so extending that is easy.

It is likely that this model can be plugged into models for the other portions of the risk, isolation, etc. and it might be useful to collaborate. It's also an important project on its own, so if there are people interested in working with us on that, I'd be happy to find more volunteers familiar with R and data analysis.

comment by catherio · 2020-07-24T18:10:13.009Z · score: 7 (3 votes) · EA(p) · GW(p)

Yes David, we would love to build off that risk model and include it in our group house microcovid estimates project. Knowing how to value a "microcovid" is an important step in choosing how many microcovids per year you should select as your tolerance.

comment by Davidmanheim · 2020-07-29T13:30:18.692Z · score: 2 (2 votes) · EA(p) · GW(p)

This is still under very active development, but the github repository is here, and a toy version of what we'd like to produce with better estimates is here, as a Rshiny App.

comment by catherio · 2020-07-24T02:41:39.036Z · score: 1 (1 votes) · EA(p) · GW(p)

(combined with other comment)

comment by catherio · 2020-07-24T02:39:35.155Z · score: 1 (1 votes) · EA(p) · GW(p)

(combined with other comment)