[Epistemic status: US-centric. gathering various impressions even though I haven't deeply investigated this topic and am not an expert, so it's all in one place and others can better use or critique the claims.]
1. Figure things out for yourself
Figure out what risk level you're comfortable with, and use microCOVID.org to get a sense of what policy makes sense for you personally.
Consider sharing your Fermi estimates with others (e.g., make a small Facebook group or chat for the purpose, or a LW thread if you're happy to do it publicly). This is a way to compare notes, get feedback, and share info without there needing to be a single big Policy Proposal For Everyone blog post people are passing around.
2. Be much more wary of COVID when hospitals are full
Keep an eye on confirmed COVID-19 cases, hospitalizations, and deaths in your area, and put effort into avoiding catching or transmitting COVID-19 if it looks like hospitals in your area will be overloaded 2-4 weeks from now.
3. Consider mostly not worrying about it
For young healthy vaccinated people in places with relatively good health care and relatively high vaccination rates (e.g., the US), if it doesn't look like your local hospitals will be overloaded in the next few weeks, then I think COVID-19 is mostly not worth worrying about right now.
(Likely exceptions: you're spending a lot of face-to-face time with an immunocompromised friend; you're visiting your seventy-five-year-old parents for the holidays in two weeks; etc.)
I emphasize this point mostly because my friends are in left/liberal spaces, where I think social forces encourage people to voice their "worry more" thoughts and keep quiet about their "worry less" thoughts.
This makes it extra valuable to speak up about "worry less" when you think it's true. (Though, again, all of this is too individual-specific for a public blog post to be able to say much about the exactly correct level of "worry".)
People in my social circle have mostly been taking large precautions throughout the pandemic. I see rationalist friends worrying a lot about whether it's OK to host get-togethers without rapid-testing all attendees (which I'd tentatively guess is not worth the trouble for the vast majority of gatherings). I see non-rationalist friends posting about how sad they are to not get to see friends again until things 'go back to normal'.
What I don't see are rough quantitative arguments for why the benefits are worth the costs here, especially ones that take into account the importance of social distancing's emotional costs (cf. Scott Alexander's Things I Learned Writing The Lockdown Post).
I don't see explanations of what 'going back to normal' looks like, and why we should expect that to happen at all (cf. Alyssa Vance on "we should distance until X").
Daniel Filan tells me that he initially locked down in response to COVID, but when he took the time to do a quick Fermi on the risks of COVID (around August 2020), he was surprised to find how costly his precautions were compared to their benefit:
[... M]y estimate [in late 2020] was that I should pay up to tens of cents to avoid a uCOVID.
I wasn't really modelling externalities well, and I'm still not totally sure how to do that right.
[...] TBC, the Fermi was something like "look up p(death | covid) given my age and sex, then estimate cost of other side effects as equal to cost of death, then add the cost of being normally sick for one week".
[T]hat being said, paying $100k to definitely not get covid seems pretty pricey.
My current guess: the cost per uCOVID [in late 2020] came out to under 10c, and I rounded up for caution or something.
See also Connor Flexman's Delta Strain: Fact Dump and Some Policy Takeaways, which (with a bunch of caveats and uncertainty) estimates that given Delta's ubiquity, a healthy vaccinated thirty-year-old who would otherwise live a full life now loses something like the equivalent of 1 hour of life for every 1,000–5,000 microCOVIDs they get.
I think the vaccines are remarkably effective (both vs. infection and vs. severe-symptoms-given-infection, which correlates with long COVID and death risks), and COVID wasn't a large risk for young healthy people in the first place (though Delta is a bigger risk than Alpha, ignoring effects of vaccination).
(Added Aug. 24: I think COVID risk for young vaccinated people isn't that different from the risk you face from other widespread viruses. In particular, my impression is that many (if not all?) viruses cause long-term symptoms similar to long COVID. Though at least in unvaccinated people, it seems that COVID causes long-term issues more often than the typical virus.
I know some people whose conclusion from this is "I'll try to avoid all viruses", including people who were taking such precautions pre-COVID. For healthy individuals, I don't have a strong view on whether that's more or less reasonable than mostly-ignoring this risk. But I am suspicious of the view that only and exactly COVID is worth worrying about.)
4. If you do worry about it...
... I would especially prioritize doing things outdoors.
I also think in-person friend groups, events, and group houses should consider restructuring to cater to people with this or that risk tolerance. The typical especially at-risk person, IMO, shouldn't be trying to get a huge community of people to all match their needs. Nor should they be bunkering down to "wait out COVID" and avoiding in-person socializing until then (unless they really don't want or need in-person socializing). Rather, I'd propose planning around the assumption that things will stay at least this risky for years to come, and build connections with other people who want to keep their risk low on that timescale.
(Added Aug. 24: The older you are, the higher your COVID risk generally is. Mayo Clinic says other major risk factors are: obesity, diabetes, heart disease (including hypertension), lung problems, cancer, sickle cell anemia, weakened immune system, chronic kidney or liver disease, and Down syndrome.)
5. Try to get triple-vaccinated
J&J seems to be much less effective than Pfizer/Moderna, so people who received J&J should especially prioritize getting two mRNA shots ASAP. But mRNA recipients should probably also get a third shot if their second shot was 4-6 months ago.
Will Eden re 'should people get a third shot, and if they had Pfizer/Moderna the first time should they try to have J&J this time around?':
The mechanisms for cell entry are a bit different, but they all end up using RNA to encode the S protein on the cell surface, so the nature of the immunity is basically the exact same mechanism. J&J is also going to immunize you against the vector virus, making it harder to receive future AAV vaccines/gene editing in the future, so personally I would stick to the mRNA vaccines only.
From the Israeli data it seems like a third shot is probably most effective around 5 months later for little or no lapse in coverage? I would recommend it personally and have to others already. Doing third shots is in trials, as is mixing different vaccines, both appear to work well.
In response to Zvi Mowshowitz's criticism of the Israeli data, Will adds:
I do think there’s a reasonable point about the denominator problem, and how outbreaks began in more heavily vaccinated areas thus skewing the results downwards. But he also admits this means there are enough vaccine breakthroughs to cause a pandemic!
On the flip side, the discrepancy between protection against symptomatic COVID vs hospitalization/death I don’t agree with him on. I think a straightforward view of the data, especially the effectiveness of vaccine vs month of administration released by Israel, suggests that you need high circulating antibodies to prevent the infection from taking root at all, but prevention of severe disease is more reliant on cellular immunity and your body mounting a quick response after exposure, which is the benefit provided by vaccination. (On the other hand, this is why I want mucosal immunity - an optimal vaccine would even prevent infection + spread, and clearly the current vaccines don’t.)
Moderna seems to be better than Pfizer, perhaps because it uses a higher dose.
The United States will begin widely distributing Covid-19 booster shots next month as new data shows that vaccine protection wanes over time[...] U.S. agencies are preparing to offer booster shots to all eligible Americans beginning the week of Sept. 20, starting eight months after their second dose of Pfizer or Moderna’s vaccines.
But this seems too slow to me, given the next section. (My family and peers would nearly all have to wait till Nov/Dec/Jan.) So I still suggest trying to get a third shot sooner.
6. Right now is the safe time
COVID seems to be quite seasonal -- it spreads a lot more in the winter. To think about the coming months, I looked at COVID rates where I'm living:
Based on this, I expect this coming Nov/Dec/Jan/Feb to be way worse again around SF. Except the baseline is much higher now, so I expect the elevated case rate to likewise be much higher.
I think most people who track these things are looking at the current COVID rates, seeing they're a lot higher than in the past, and concluding that we should be buckling down temporarily right now.
Except that the rise is because Delta took over, and Delta is the new normal going forward. So if you're planning to throw some parties before March 2022, now is very likely the best time to do it. If you're planning to "buckle down temporarily" and "temporarily" means "a few months" rather than "a few years", you may want to reconsider whether things will actually look any better five or ten months from now.
Eyeballing NYC, it looks like things may be bad in some parts of the US even in May?
Beyond seasonality, other things that may make things worse in the future include:
- Fading vaccine efficacy, if that's a thing. Even if you personally get more shots, others may not.
- In general, my tentative default guess (if something like the status quo continues) is that people's willingness to get more COVID booster shots will drop off a lot over the coming months and years. (Because it will feel more like an annoying regular chore than like a miracle cure.)
- New COVID strains.
- Regulatory hurdles inhibiting and delaying vaccine roll-out.
I'm not especially worried about long COVID, because I expect long COVID frequency and severity to track symptom severity pretty closely -- if vaccines protect against severe symptoms, they'll tend to reduce long COVID risk too.
Even in the best-case scenario, there's a delay between "new COVID strain evolves" and "new vaccine targeted at that strain rolls out," and a further delay before we get data on long COVID in people infected by that strain (either for vaccinated or unvaccinated people), which makes it harder to be confident about the risks.
Or perhaps not. The future could hold a lot of things. But in my own planning, I'll be acting like most of the probability mass is on 'things stay about the same for a long time, or get worse'. I mostly am planning around the expectation that this is life going forward, and I'm not going to shut down my life indefinitely -- but I can plan around taking more risks during warmer months, and fewer risks during colder ones.
7. Small exposures are better than large exposures, and maybe better than no exposure?
This is a probably a relatively good time to get infected if the trend "COVID evolves to be more dangerous and high-viral-load" continues. Getting infected with a lower viral load is better, and may confer a lot of immunity against future variants.
By the same logic, if you expect to get infected with Delta regardless in the next few months (as a large portion of the US presumably will), it's better if your exposures tend to be in low-risk settings where you'll get a small initial viral load. (And it's better to get sick when you know hospitals won't be overloaded.)
8. Be aware of possible signs of COVID
(Added Aug. 24.)
According to CBS, an August CDC study found that the most common symptoms of Delta are still "cough, headache, sore throat, myalgia [muscle pain], and fever". Oddly, the COVID Symptom Study instead found that the most common Delta symptoms were runny nose, headache, sneezing, sore throat, and loss of smell.
But: don't assume people are COVID-free just because they're showing no symptoms. COVID is often asymptomatic (or the symptoms are hard to notice), and people with COVID transmit the illness a lot before they start showing symptoms.
If you might have COVID, self-isolate and try to get tested. PCR tests are the most reliable, but they have to be sent to a lab and normally take days to give back a result. Also, Connor Flexman suspects that PCR tests may still have a ~40% false negative rate, which is much higher than usually advertised. Antigen tests have something like a ~50% false negative rate, but can be done fully at home and give much faster results (within a few minutes). LAMP tests like Lucira are also fast, and are in-between PCR and antigen tests in efficacy, according to Connor. (Flagging that these are very off-the-cuff numbers with minimal due diligence done; I'll revise them if the situation becomes clearer.)
Consider taking zinc lozenges soon after COVID exposure or symptom onset. (They should taste bad if they're working.)
9. Be prepared for if you get pretty sick
Things to buy now for if you get sick: Pedialyte or gatorade powder, acetaminophen or ibuprofen, oral thermometers, and a finger pulse oximeter.
Maybe also: over-the-counter inhalers, a humidifier, mucinex/guaifenesin, pseudoephedrin.
10. Take care of yourself if you get sick
(section in progress)