mindstalk: (riboku)
mindstalk ([personal profile] mindstalk) wrote2022-01-08 12:03 am

personal zero covid, musings

I've seen various people say "everyone's going to get omicron" -- usually in a smug way, along with "covid is endemic", implying that it's foolish to try to avoid it. Usually these people are healthy and not obviously possessed of frail people they care about. At any rate, I'm inclined to try to prove them wrong, and avoid getting it until they roll out new vaccines or prove that it's really "just a cold" or something. Which means, basically, no unmasked contact with people until rates have come down a lot. But come down how much?


Here's one approach: an 80 year lifespan is close to 30,000 days, so if the daily case rate were 1 in 30,000, or 33 in a million, that would (very simplistically) imply a near 100% chance of covid in the lifespan. Halve the rate to aim for 50% chance; halve again since the real infection rate is greater than the reported rate, and we're at 8 cases per million. So if the rate is that low, then it's safe to eat out (dining inside)... every day? Sometimes? Not really clear, but there's at least a feeling that such a rate is approaching safety.

Of course, the real infection rate might be 4x the reported rate, not 2x. Or more. OTOH, I'm already halfway through my lifespan, so can maybe regain a factor of 2.

It's tempting to give myself credit for a safer lifestyle -- living alone, working at home, not going to bars -- but the main point of this exercise is to gauge when it would be safe to live unsafely again, e.g. dining out. And just as, for STD purposes, when you have unsafe sex with someone you're in a sense having sex with everyone they've had sex with in the past year, so for covid purposes, once you mingle unmasked with strangers, you're breathing not just their air but all the air they've shared over the past week, and all the other safety in your life stops mattering. Or matters in limiting how often you roll the dice of infection, but still, you're rolling at all.

Hmm, (I'm thinking out loud, as it were), that may help clarify things. One unmasked interaction is one die roll. If I eat out every 2-3 days, that's fewer die rolls than doing so every day, fewer engagements with the infection network. If I go to a bar, that's like one very intense interaction, or a lot of die rolls.

Https://microcovid.org is a much more sophisticated approach to the whole question, purporting to give risks for various scenarios; whether it's more accurate than just eyeballing the local infection rate, I can't say.

8 per million is pretty stringent. Right now only 3 'countries' whose testing I might trust qualify: China, Taiwan, Hong Kong. Japan did until a couple days ago, but is rapidly doubling: 2 days ago 19, yesterday 33. Korea has been holding the line around 80 per million. Even when things were calmer few qualified, though New Zealand and Vietnam were among that number, and Korea.

By contrast, currently daily infections are hitting 5000 per million in many places, and probably hitting the limits of testing capacity.

I haven't actually lived that stringently; when I had dim sum in Toronto, the local case rate was I dunno, maybe 50? 100? OTOH I did that once, somewhat nervously. I'd also eaten out (inside) in SF and Quebec, with I forget what case rates, but usually with open doors or almost no other diners.

At 50 per million, eating out twice a week is 100 die rolls, for roughly 0.5% chance of covid in a year.

Well, the melatonin is kicking in, so I'll end it here.
elusiveat: (Default)

[personal profile] elusiveat 2022-01-08 01:01 pm (UTC)(link)
And just as, for STD purposes, when you have unsafe sex with someone you're in a sense having sex with everyone they've had sex with in the past year, so for covid purposes, once you mingle unmasked with strangers, you're breathing not just their air but all the air they've shared over the past week, and all the other safety in your life stops mattering. Or matters in limiting how often you roll the dice of infection, but still, you're rolling at all.

This is a good first approximation, but if you have good information about the status of the people you're interacting with, the individual risk calculation is slightly different due to incubation periods. Just as an exposure event more than two weeks past is unlikely to have a bearing on the infectiousness of a person you're interacting with today, an exposure event less than three days ago also is unlikely to have an effect. So, if you could trust a set of people to observe stringent precautions for a two week interval beforehand, you should be able to hang out with each of them within a span of a weekend without too much increase to your risk profile, even if they are doing something similar with a *different* group of friends. Network trust is a problem, though, and on COVID precautions (moreso than STDs) I think that a lot of people really don't have a clear idea of what behaviors are and aren't risky.

As with STDs, testing frequency also changes things. The institution where I teach is going to be testing all members of the campus community *three times per week* until the current surge ends (this is in addition to mandatory masking and mandatory boosting). That changes the risk profile even for interacting with people who engage in riskier behaviors outside of the campus setting.
squirrelitude: (Default)

[personal profile] squirrelitude 2022-01-09 04:27 am (UTC)(link)
Not sure I followed your estimates. Let me try something else.

Let's say that on an average day, on transit and in offices and restaurants and so forth, you have 100 infection-time-window opportunities. Some of those "opportunities" would be multiple from the same person ("transmission many times over") so you'd want to discount that, but some of the interactions would also be fractional. Let's just settle on "you meet 100 people each day and if any of them is infectious you get infected".

Maybe each positive person is infectious for 4 days. (I'm squaring up the bell curve here, and playing very loose with numbers.) For each case, there are 4 person-days contributing to risk, if they don't isolate *at all*. So if there's 1 case per 250 per day, that's a chance = 1 of catching it each day.

What kind of risk do you want? 0.50 / (40 * 365.25) = 3.4E-5 chance per day? 250 / 3.4E-5 = 1 in 7.4 million case rate.

...hah, OK, that comes out to about your estimate, doesn't it. :-)

Of course, it's going to come in waves; it may not really circulate at all for half each year if your area has a strong seasonality to it. And certain behaviors will be higher or lower risk. So average rates are only the beginning.
squirrelitude: (Default)

[personal profile] squirrelitude 2022-01-09 02:56 pm (UTC)(link)
Ah, yeah, I think I get it -- if everyone had the same risk, what's the rate that would get you a 50% change of ever contracting it. Makes sense.

(Also, I bungled my own math in trying to approach it from the other direction, so I'll see if I can fix that.)
squirrelitude: (Default)

[personal profile] squirrelitude 2022-01-09 02:59 pm (UTC)(link)
Corrections: Chance = 1 would require 1 case per 400, not per 250. (Brain: 100 and 1000 are the same thing, right? And × is the same as ÷?) Not a huge correction factor, though; the resulting 1 in 12 million is the same order of magnitude as 1 in 7.4 million.