(UPDATE 9/6/21: now that more is known about the Delta variant, I'm being more conservative than I articulated below, and managing to a weekly risk budget of 400 microcovids. The main difference beyond what I articulated below is that I'm no longer interacting freely with fully vaccinated people, but rather considering how risky various interactions are and moderating those accordingly or masking/distancing. I may relax that again based on my determination of how big an issue long covid is, which I'm exploring here: https://www.lesswrong.com/posts/jfHZR6Ykmc5DBSLCp/cliffs-notes-how-much-should-fully-vaccinated-people-care)


As someone who is now several months past my second dose of Pfizer, and who lives in San Francisco, which opened up completely earlier this month, I've been debating what events / spaces are still too risky to spend time in. Here are the rules of thumb I've been following and the thinking that informed them; would be interested in input!

I imagine I could make further progress with further thinking (and/or maybe I should just pick a risk budget and use the microCOVID calculator to stick to it), but figured that since I've done this amount of research and thinking, it might be helpful to some and low-cost for others to help fill in some gaps.


  • I am ok interacting freely (unmasked, indoors) with fully vaccinated people
  • In "mixed" environments (where I could be interacting with unvaccinated people), I will still take precautions:
    • Outdoors: wear a mask (KN95 or better) OR distance
    • Indoors: wear a mask (KN95 or better) AND distance
  • I will avoid events / spaces where those precautions are impossible
  • Open questions: 
    • How risky is it to loosen precautions in environments where not everyone is fully-vaccinated?
      • I'm not quite sure how to use the microcovid calculator to gauge risk in "mixed" settings - does anyone know if they incorporate vaccination rates in your area when you select "I don't know" in response to "Their vaccine"?
        • It seems like they might, because for a given scenario, answering that question with "Yes" yielded 40 microCOVIDs, "No" yielded 400, and "I don't know" yielded 100, which is much closer to "Yes" than "No". But I haven't thought though the math.
    • How risky is it to do risky activities (say, kissing, sharing drinks) with people who are fully vaccinated but regularly doing those activities with unvaccinated people?


(#1 and 2 drive my decision to take the precautions I am still taking. #3 drives my decision to relax precautions otherwise.)

(1) Current vaccines are less effective against the Delta variant (confidence: medium-high)

  • The mRNA vaccines are in the 80+% range and J&J's in the 60% effectiveness range against symptomatic COVID for the Delta variant (De-Lin 6/23/21)
  • The WHO is urging fully vaccinated people to continue to wear masks given the Delta variant (CNBC 6/25/21), and the WSJ reported that about half of adults infected in an outbreak of the delta variant in Israel were fully vaccinated with Pfizer, prompting the government there to reimpose and indoor mask requirement and other measures (WSJ 6/25/21)
  • However:
    • Public health experts cited by the NYT say the Delta variant is "unlikely to pose much risk to people who have been fully vaccinated."
      • “If you’re fully vaccinated, I would largely not worry about it,” said Dr. Ashish K. Jha, dean of the Brown University School of Public Health.
      • According to one recent study, the Pfizer-BioNTech vaccine was 88 percent effective at protecting against symptomatic disease caused by Delta, nearly matching its 93 percent effectiveness against the Alpha variant. But a single dose of the vaccine was just 33 percent effective against Delta, the study found.
      • “Fully immunized individuals should do well with this new phase of the epidemic,” said Dr. Peter Hotez, dean of the National School of Tropical Medicine at Baylor College of Medicine. “However, the protection offered by a single dose appears low, and of course if you are not at all vaccinated, consider yourself at high risk.” (NYT 6/22/21)
      • That is despite the following, mentioned earlier in the same article:
        • "Delta, formerly known as B.1.617.2, is believed to be the most transmissible variant yet, spreading more easily than both the original strain of the virus and the Alpha variant first identified in Britain. Public health officials there have said that Delta could be 50 percent more contagious than Alpha, though precise estimates of its infectiousness vary.
        • Other evidence suggests that the variant may be able to partially evade the antibodies made by the body after a coronavirus infection or vaccination. And the variant may also render certain monoclonal antibody treatments less effective, the C.D.C. notes.
        • Delta may also cause more severe illness. A recent Scottish study, for instance, found that people infected by the Delta variant were roughly twice as likely to be hospitalized than were those infected with Alpha. But uncertainties remain, scientists said."
  • Fully vaccinated people tend to have milder symptoms, avoiding hospitalization and death - it seems the call for masks is driven largely by the goal of preventing community transmission [to the unvaccinated, I assume?] (De-Lin 6/26/21)

(2) There may be long-term adverse effects from COVID even if it doesn't require hospitalization, but the risk is low (confidence: low)

  • This article claims the risk is low but understudied: https://www.huffpost.com/entry/will-people-who-get-covid-post-vaccination-have-long-term-symptoms_l_60afda9be4b0ead279660672 (5/30/21)
  • The following paper Zvi linked in his last roundup did not look at vaccination (which, per the article above, may be protective against long COVID), nor did it give a good sense of the practical impact of the effects it mentions:
    • "Paper documents loss of grey matter in the brain after getting Covid-19, including for those who were not hospitalized – hospitalization did not seem to impact the magnitude of this effect. You do not want to get Covid-19. Given the timing this does not provide information on vaccinated people who then still got infected, nor does it differentiate between severity levels beyond whether someone was hospitalized. I do not have a good sense of what size impact one should expect from the effect observed here – it’s easy for this type of thing to be quite impactful, and also easy for it to sound scary while not having much impact at all." (Zvi 6/24/21)

(3) Prior to the Delta variant being a major concern, prevailing guidance was that it was fine for fully-vaccinated people to interact in close quarters indoors without masks. My understanding is that this hasn't changed for fully-vaccinated people since then. (confidence: medium)

  • NYT 4/22/21: In a crowd, mask. And when you are with unvaxxed people 2 of 3 outdoors, distanced, and masked. Keep one in a pocket going out.
    • When to mask
      • If you [or the other person are?] not vaccinated, make sure your activity meets two out of the following three conditions: outdoors, distanced and masked
      • And masks are still advised for both the vaccinated and unvaccinated if you find yourself in an outdoor crowd, according to the new C.D.C. guidelines. Standing shoulder to shoulder with strangers during an outdoor concert or a protest could increase risk, particularly for the unvaccinated.Recently while hiking without a mask, Dr. Marr said she still made an effort to keep her distance from large groups when the trail got crowded.
      • “If I was passing by a solo hiker it didn’t concern me,” said Dr. Marr. “But if I passed by a group of 10 hikers in a row, I stepped further off the path. The risk is still low, but at some point there could be a large enough pack of people that the risk could become appreciable.”
  • AAMC 3/2/21: ok to hang w/fully vaccinated folks w/o masking, not unvaxxed, probably not indoor restaurants
    • Can two people who are fully vaccinated be with each other without masking? “The answer is a firm and definitive yes,” Gandhi says.
    • “If you are around other people who have been vaccinated, you can take that mask off and enjoy being with them,” Ranney adds.
    • Can a vaccinated person be around an unvaccinated person without masking? No. “While I’m almost positive that vaccination is going to take away transmission, if you had a little viral RNA in your nose, we would never want a vaccinated person to pass that on to an unvaccinated person,” Gandhi says. “So, mask around the unvaccinated until they are vaccinated too.”
    • Can two vaccinated people enjoy dinner in an indoor restaurant? Maybe. “Because these are such powerful vaccines, you should be able to go out to eat, especially right now, when most of the restaurants have extra ventilation and spacing and the waiters are all wearing masks,” Gandhi says.
    • Ranney is a bit more cautious. “A couple of months from now, when most of us are vaccinated, going back to restaurants is going to be very safe,” she says. “But right now, given the high rates of COVID in the community and the fact that the vaccines are not 100% effective, that would not be my first choice.” Instead, she would recommend having your vaccinated friends over for dinner in your home.
    • Can you hug your grandkids? While the vaccines aren’t likely to be available to children until later in 2021, Ranney plans to reunite her parents and children as soon as her parents are fully vaccinated. “The risk to them of getting really sick from COVID is quite low. But the risk to them emotionally from continuing to be separated from my kids is higher. It’s a risk-benefit equation. We’re not going to completely eliminate COVID from the world, so two weeks after my parents get their second dose, we’re going to get together.”
    • Can you fly or go to the gym? “If vaccinated, you can go to an indoor gym, fly (maintaining your masking for others), and start opening up your life,” Gandhi says.
  • Good examples of the sorts of precautions to take in a variety of specific situations:
  • Vaccine's protectiveness
    • Very few breakthrough infections [though variants TBD...?] (NYT 4/22/21)
      • The U.S. Centers for Disease Control and Prevention recently reported just 5,800 cases of breakthrough infections among 75 million vaccinated people. And the C.D.C. has said vaccinated friends and family members can safely spend time together, indoors and outdoors, without masks.
      • [Though note - variants TBD]
    • Vaxxed unlikely to spread virus. Variants TBD (JHSPH 4/8/21)
      • Details
        • Can we say with any degree of certainty that vaccinated people are unlikely to spread COVID to unvaccinated individuals?
          • The emerging data confirms what many of us thought would be the case—that not only do the vaccines stop symptomatic COVID, but they also make it highly unlikely that someone can even be infected at all. I think the preponderance of the evidence supports the fact that vaccinated individuals are not able to spread the virus.
        • What is known about the variants of concern and their ability to infect fully vaccinated individuals?
          • When it comes to variants, it is likely the case that it depends on the variant and depends on the vaccine. The B.1.1.7 variant (first identified in the U.K.) is one that vaccines appear robust against. The more concerning B.1.351 variant (first identified in South Africa) appears to pose problems for the J&J vaccine but not where it counts in terms of presenting serious disease, hospitalization, and death. Data on the P.1 variant (first identified in Brazil) is forthcoming, but I suspect it will be similar to the B1.351 variant.
    • Variant impact
      • Even with variants, that guidance holds (Nat Geo 4/21/21)
        • “If you’re vaccinated, you can pretty much assume that you are protected against severe disease and very likely protected against enough infection to transmit, but because we have these variants emerging and the fact that we’re not even close to herd immunity, people should still be taking precautions,” Morrison says.Interacting with other vaccinated people without masks makes sense, but she also agrees with the CDC recommendation for vaccinated people to visit without masks or social distancing only with low-risk unvaccinated people in a single household. With so many infections still occurring daily, that limitation further reduces the likelihood of vaccinated people picking up and spreading infections from an unvaccinated home.“The real worry is for the unvaccinated people you come into contact with,” she adds. “Even if the potential for them to pass it on to you is low, it’s not zero.” Similarly, an infected vaccinated person has lower—but not a zero—likelihood of infecting others who aren’t vaccinated or have conditions or medications suppressing their immune systems.
      • Preliminary indication they protect against UK (B.1.1.7), impact severity with S Af (B.1.351) and Brazilian (P.1) (AAMC 3/2/21)
        • Myth #4: The variants are going to get us anyway, vaccines or not.For the last few months, concern has been growing that a number of SARS-CoV-2 variants — mutated versions of the virus that seem to be contributing to greater numbers of hospitalizations and deaths in some parts of the world — could render the vaccines impotent.The important thing to remember is that not all variants are the same, says Frieman. Emerging data from Pfizer, Moderna, and Johnson & Johnson, as well as the vaccine candidates from Novavax and AstraZeneca that are not yet approved by the Food and Drug Administration, suggest that all of these vaccines are highly protective against both the original virus and the so-called U.K. variant — also known as the B.1.1.7 variant — that is projected to become the dominant strain in the United States by the end of March.“The general consensus is that if you are vaccinated with any of the vaccines that we have now [including the vaccines from Novavax and AstraZeneca], you are protected against the 2020 strains and the U.K. variant,” Frieman notes.The so-called South African variant, technically known as B.1.351, and the Brazilian variant, known as P.1, are similar and do not seem to be as easily neutralized by the vaccines, Sette says. But that doesn’t mean the vaccines are completely useless. “We have been doing some calculations and it seems that most of the pieces that the T cells recognize are not changed in the variants,” he says. “What that means is that the T cell response [induced by the vaccine] may not prevent infection, but it can impact disease severity.”Gandhi agrees. “We are talking about the variants as if our T cell responses stimulated by the vaccine are not important. But they’re very important in preventing severe disease. And that matters because we never would have been in this mess with SARS-CoV-2 if it didn’t cause severe disease.”
  • Official CDC guidelines (4/2/21)
  • Sources


New Answer
Ask Related Question
New Comment

4 Answers sorted by

You omit to give your age, which is highly relevant. Take the risks from the below paper* and then deduct another 95%+ to account for being fully vaccinated. Unless you’re either very elderly or seriously unwell (on the order of having leukaemia not just being mildly asthmatic) I suggest that the risk level is now low enough that it should not be driving your decisions, in the same way that you’re not dedicating this much effort to avoiding flu. (No, Covid isn’t flu, but when you are fully vaccinated then the risk level becomes comparable.)

It sounds like you’ve been overthinking this a lot. It’s time to live your life, see friends, enjoy yourself, and live again. There are more important thigs in life than squeezing out every last micromort of risk at the expense of all joy and of everything that makes life worth living.

*Our analysis finds a exponential relationship between age and IFR for COVID-19. The estimated age-specific IFR is very low for children and younger adults (e.g., 0.002% at age 10 and 0.01% at age 25) but increases progressively to 0.4% at age 55, 1.4% at age 65, 4.6% at age 75, and 15% at age 85. Moreover, our results indicate that about 90% of the variation in population IFR across geographical locations reflects differences in the age composition of the population and the extent to which relatively vulnerable age groups were exposed to the virus


IFR is not the only thing that matters. Avoiding long term heart and brain damage is also important. 

7Dumbledore's Army2y
Point stands, I think. Once you’re fully vaccinated the risk - including risk of post-viral fatigue - is in the range we normally consider tolerable. More generally, you need to balance risk reduction against actually enjoying your life. I would rather live a rich life than extend a grey and joyless existence, even if it means tolerating a small risk that said life will end early. That calculus shifts in the presence of large risks, but we aren’t talking about large risks now. I would encourage the OP, and everyone else who is vaccinated and still panicking worrying excessively about now-small risks, to look at the big picture and ask if they need a sense of perspective.

I'd find it helpful if folks had evidence to share about the level of risks other than death. (E.g. the risk of post-viral fatigue.) I agree that you need to balance risk reduction against actually enjoying your life, but I've been able to do that to my satisfaction and am interested in assessing the marginal risk of the items I noted in my post. I didn't go much into the benefit side in the post, because that varies by individual, and I feel pretty capable of assessing the marginal benefit for myself. (I can assure you that my life is quite rich, and nothing close to grey and joyless, even though I haven't leapt into indoor activities with unvaccinated people). 

Panicking isn't useful. Having discussions about how to effectively deal with risk isn't panicking, talking about panicking is strawmanning. 
1Dumbledore's Army2y
Ok, reworded to something else.
6Neel Nanda2y
Do you have a source for this? I've seen good data about hospitalization and risk of death, but nothing about long COVID. They probably correlate, but I've seen suggestive data that they correlate less than I'd intuitively expect. It definitely doesn't feel like there's enough data to be confident in saying 'this is now a silly thing to care about or spend mental energy on'. Though I'd mostly agree if you live in an area with very low case counts.
6Dumbledore's Army2y
Sorry, no source - but given the vaccines massively (>90%) reduce risk of other harms ie death & hospitalisation, I think the null hypothesis has to be that vaccines also massively reduce risk of long COVID.  I also start with a prior that a lot of discussion about long-COVID is low quality and I think it's an example of post-viral fatigue rather than some brand-new thing. It gets lots of media hype, like anything vaguely scary and covid-related, but hard data seems to be hard to come by and often very low quality. 
5Neel Nanda2y
The [best source I've found] (https://institute.global/policy/hidden-pandemic-long-covid [https://institute.global/policy/hidden-pandemic-long-covid]) finds a 30% reduction in P(Long COVID | infection after 2 vaccine doses). Infection reduction is about 85%, so total risk reduction is about 90%, MUCH less than the risk reduction for hospitalisation. The study is based on 3,000 infected patients, all over 60, unclear how it generalises to younger people. In general, there is SOME good quality research on long COVID, and it seems obvious to me that it is a legitimate thing and respects a good fraction of the harm of the pandemic. Even if overall research is much less high quality than I want.
1Dumbledore's Army2y
Thanks for the source, I hadn’t seen it before. 90% risk reduction is still an order of magnitude, seems like a big deal to me.  One point to be aware of: I notice they don’t distinguish between the different vaccines, they just give a population-wide figure. The UK has used a combination of Pfizer and AstraZeneca, and on other metrics eg efficacy against symptomatic infection or hospitalisation, AZ is slightly to moderately worse than Pfizer. Assuming the same pattern holds for long covid, I would assume a >90% risk reduction for Pfizer (and Moderna with similar mRNA technology), which is the read-across relevant to American readers.
3Matthew Barnett2y
They said "No, Covid isn’t flu, but when you are fully vaccinated then the risk level becomes comparable." So, what's the evidence that long term heart and brain damage of COVID-19 is worse than the flu (or the cold for that matter) after you're already vaccinated?
The general way to deal with drugs is to put the of burden of evidence on the drug that it helps with certain conditions (it's called the precautionary principle in medicine). Nobody gathered evidence that any of the vaccines help against long COVID. The vaccines do seem to help against outcomes like hospitalization but if we look at an issue like brain damage [https://www.medrxiv.org/content/10.1101/2021.06.11.21258690v1]there's no statistical significant difference between whether or not the person was hospitalized. It's been a while since I read up on heart damage and from what I remember that also didn't need hospitalization to occur. 
Nobody could have within the time frame. Have you noticed how there's always one bunch of people complaining that everything has been slowed down by bureaucracy, and another saying that nothing has been tested thoroughly enough?
The sentence said nothing about requiring testing. We could just say: "You get 5$ extra per vaccine dose if you show it helps reduces long COVID by 90%".
How do they show something wthout testing it?
They are not required to show it to get their vaccine to market. Generally, the idea is to make it easier to bring vaccines to market and then pay extra for proof that the vaccine does desirable things.
Specifically, you are paying them extra money to show it.
Yes, you are paying money for important information. That's different then slowing down response via bureaucracy. 
You are not, because money can't buy time in the required sense. If the purchasers of the drug can't afford to wait to study the long term impact, it is no good paying the providers extra, because there is no way they can accelerate time. You are assuming the very lesswrongian assumption that all bureaucracy is unnecessary. It's more complicated than that. If you remove the checks and balances, you don't get the same results faster , you get worse results faster.
That's just wrong. You can run multiple studies. Moderna/Pfizer didn't have information about whether the vaccine reduces transmission in the first trial that lead to bringing the vaccine to market. It's information they gathered in later trials and there's no reason why the couldn't have run tests for long COVID on patients of those trials.  Recruiting more patients for clinical trials accelerates the trial and costs money. Recruiting the amount of patients that allowed Moderna and Pfizer to get their vaccines approved when they did cost hundreds of millions of dollars.  I'm not sure why you want to strawman. I never said that I reject all bureaucracy. I'm just for less of it and smarter regulation. I'd love to see a law that criminalizes intentionally witholding information about biosafety breaches from the public. As Pakinson described, he British foreign service managed to get by with orders of magnitude less bureaucrats when they had an actual empire to manage then they have employed afterwards. Bureaucracy grows like cancer and is hard to reduce.  Hydra manages to have bureaucracy that does independent quality testing. The FDA doesn't manage to do any independend quality testing and thus fails to remove fraudulent products like those of Ranbaxy from the market within reasonable trimeframes and manages to approve drugs where their scientific advisory panel says they don't work.  The FDA combines a lot of resistence to bringing drugs to the market with little action to provide actual safety. In this case we got vaccines with extremely high side effects compared to the vaccines that we usually use as a result of the regulation. While they might not cause lasting harm, being ill for a day isn't nothing.  Without regulation we would have used well understood and easy to scale up vaccine technology earlier in 2020. The regulation we have only allowed for vaccines with patent protected technology to be effectively brought to market and unfortunately that came
I'm aware that parallelism is how you usually speeds things up. I am saying that it doesnt work in cases where you are studing a long term phenomenon. They also didn't mind millions starving. You can see increased bureaucracy and regulation as being a reflection of putting an increased value on individual wellbeing. I don't know if that's right , but you could consider it. That's hindsight fallacy.
You can't study effects of COVID a few years out. At the same time we could now have information about what 6-month after infection effects the vaccines prevent. I can also consider that increased bureaucracy and regulation is due to God making it happen. There are a lot of bad explanations that I can consider.  If the increased bureaucracy is due to increased value of individual wellbeing, we would only see it in situations where the point of the regulation is increased wellbeing. Few people think about tax law as being primarily about wellbeing, yet the complexity of it grows constantly.  Just like cancer grows naturally bureaucracy does as well. Pakinson did good work on describing how it works.  Hindsight fallacy would be saying that it would have been predictable when the pandemic started that the process leads to vaccines with higher side-effects. What I said was just that it did lead to vaccines with higher side-effects. That's an observation that does come from hindsight and it would have been possible for regulation to produce no net damage in this case. At the same time I have written about how regulation increases side-effects of drugs before, so it's not completely a thesis that comes out of hindsight.  If we look at vaccines, vaccines that get developed in a way where the inventor of the vaccine vaccinates himself early are more likely to be safe then ones that get validated through clinical trials where increasing the chance of the trials finding a clinical effect is more important then reducing side effects.  You could write a regulation that the first human in which a new vaccine gets tested as to be the CEO of the vaccine company to create skin-in-the-game. Such regulation wouldn't slow down vaccine development but would help with safety.
No,not even for five extra dollars a dose. You could also consider that the truth lies somewhere between. Which is to say : "if the increased bureaucracy is entirely due to increased value of individual wellbeing..." I am finding that hard to parse. How are you defining "safe", how are you checking that they actually took their miracle cure, and why are you placing so much confidence on a single (at best) data point? It's easy to justify having some non-zero level of regulation by looking at the quackery prevalant in the nineteenth early twentieth century. And claiming to have benefited from a cure you had never personally taken is quackery 101.
1[comment deleted]2y
I completely lost my sense and smell and it did return over the next few months, for the record. Therefore, I wouldn’t consider that damage final in all cases.
I am super skeptical of that whole brain damage thing.  Brains change, from all kinds of things.  I can't help but notice that everywhere they see statistically significant differences is downstream of smell and taste, and actually closely resembles previously described brain changes in people with chronic rhinitis that blocks the sense of smell through ordinary means.

I'm in my mid-30s, and I'd say, moderately asthmatic, which probably falls into the same risk category you had in mind. I'm not sure what led you to believe that I've been avoiding seeing friends or enjoying myself, and squeezing out risks at the expense of everything that makes life worth living -- refraining from indoor events with people who're unvaccinated hasn't had much of an impact on my quality of life, but it will have a bigger impact on my quality of life now that events are loosening restrictions. Hence my post.

I think it's clear that IFR is low... (read more)

Your interventions seem to be based on the idea aerosols accumulation doesn't matter much. My current assumptions is that aerosols accumulation matters a lot for COVID-19. This means that when indoors ventilation and running air filters is important and plausibly more important then masks/distance. 

More dakka with airfilters is an option if you are worried about transmission when hosting a party. 

When it comes to masks it's worth noting that different masks provide different protection. The study in military recruits suggests that while cloth masks might reduce your chances to spread the disease they don't reduce your change of getting infected.

Thanks. I'd say that my interventions are based on the assumption that I can't personally do much to impact aerosol accumulation in the indoor events I want to attend. I think that I may be underestimating the extent to which I can influence / screen for that (and perhaps overestimating the efficacy of masks and distancing indoors), so thanks for raising this. 

When taking an Uber, it's quite easy to say "Hey, I want to have the windows open".  You likely won't get a quantity of air filters that makes an indoors event like an outdoor event outside of a meeting between rationalists, but you can still voice the preference to open windows at many events. 

Don't French kiss people who are symptomatic and known to be infected.

Or, more reasonably, if you know someone is infected OR symptomatic avoid "Sharing their air."

Once you account for the lack of community cases (if 1 in 10,000 people are infected, as is currently approximately the case in the vaccinated parts of the world) then having a close interaction with 100 people at a gathering of any size has less than .1% chance of even including an infected individual.

If you want zero risk, wear an elastomeric respirator or DIY PAPR. Masks probably don't work against the variants (masks wiped out the flu but not the massive fall/winter covid wave).

There's no good evidence that outdoor transmission is even a thing, and mechanistic reasons (aerosols can't accumulate outdoors) also cast doubt on the idea of outdoor transmission. In places where lots of people are vaccinated or have immunity by having been infected, the risk is miniscule even in outdoor crowds. If you're extremely risk-averse, wear a respirator or PAPR or avoid crowds altogether.

Masks probably don't work against the variants (masks wiped out the flu but not the massive fall/winter covid wave).

This seems like the wrong inference. The R0 of flu is something like 1.2, the R0 of Alpha was about 4 (at pre-COVID levels of social distancing). 'Masks work' looks like masks reducing R0 by some factor. If this reduces R0 to below 1, it wipes out the disease, if it remains above 1 you will still get a massive wave. Because the R0 of flu is so much lower, 'flu was wiped out but COVID wasn't' is approximately 0 evidence about the effectiveness of masks.

For example, this paper found a 25% reduction in R0 from universal mask wearing. That would reduce flu to 0.9 and wipe it out, but reduce Alpha to 3, which is still very virulent. Yet, it is still obviously correct to wear a mask

It is obviously correct to wear a mask only if you do not have access to a respirator or PAPR. Better protection could have wiped out covid just like masks wiped out the flu. While masks may have offered some better-than-nothing protection, it was grossly inadequate protection.
3Neel Nanda2y
Sure, I'd agree with this. Things like N95s and P100s are much better than cloth or surgical masks.

"Don't work" seems like the wrong phrase. Unless the newer variants have evolved teleportation, masks will be about as effective as they ever were at reducing the extent to which respiratory droplets carry the virus from one person to another. I bet they produce about the same reduction in R as they used to.

Against newer more transmissible variants, widespread mask-wearing may not be enough to stop exponential growth among the unvaccinated. But "aren't enough" and "don't work" are two very different things.

"Don't work" in the sense of "masks wiped out the flu but not the massive fall/winter covid wave." One can argue that the last covid wave could have been worse without masks, but that's pure speculation. The newer variants might have defeated masks by producing more virus particles, for instance; no teleportation required. Whatever the mechanism, it seems clear that above a certain threshold of infectivity (compared to the original covid variants and last season's flu variants), masks don't work, or if you're picky about it, masks don't work that well.

Again, I don't think it's clear that "don't work" is at all a good way to say it.

Consider the following scenario, which I expect has something like the right shape although all the concrete numbers are made up and probably wrong (and of course the numbers aren't as deterministic as this makes it sound): the effect of wearing masks is that the number of virus particles you get hit with from being near someone infected is 3x lower; the likelihood of an infection taking hold is greater when the number of virus particles is larger; newer variants make infected people produce 2x more virus particles; maskless, the infection rate is high enough for exponential growth when the number of virus particles is >= half the number passed on (maskless) by the original variant.

Then (1) the original variant could be effectively stopped just by masks (you get N/3 virus particles, which is smaller than N/2); (2) the new variant can't be (you get 2N/3 virus particles, which is larger than N/2); (3) the masks are still reducing virus transmission by the same factor as ever.

In this scenario the new variant needs something more than masks to stop it (e.g., more distancing, vaccination, testing-and-tra... (read more)

If the choice is between wearing a mask and nothing, wearing a mask would probably be better than nothing. If the choice is between wearing a mask and a respirator or PAPR, choosing to wear a mask significantly increases risk, because we already know that masks offer poor protection. We know this because masks failed to stop the last covid wave. In the wider context of covid waves: if everyone wore a respirator or PAPR starting before the last covid wave, covid would have been wiped out.
OK, sure: other more elaborate barriers between your face and the outside world are more effective than surgical masks, bits of cloth, etc. No question. But, again, the situation isn't that masks don't work. It's that other things work even better. (I would bet fairly heavily that surgical masks + full vaccination[1] are "enough", given a modest level of general caution otherwise, even for the latest and most infectious strains. Surgical masks are much less hassle than respirators or PAPRs. I do not think I would recommend that fully vaccinated people who aren't extra-vulnerable or extra-anxious should go out and get N95 respirators and the like rather than making do with masks.) [1] With, say, Pfizer or Moderna or (less good but still probably enough) AstraZeneca. The same may be true for others but my ignorance about them is greater. You need a model of the world that's less black-and-white than "X works" versus "X doesn'tt work". Any given intervention reduces transmission by a certain amount. Depending on how transmissible the currently-relevant strain of the virus is, different combinations of interventions will be sufficient or not. Collapsing all that to "works" versus "doesn't work" is a bad idea; it makes it harder to think clearly.
At this point in the pandemic, the level of protection offered by masks is so uncertain (10% 1%? 0.1%?) and likely to be so small that masks are little more than a Hail Mary for those that don't have access to vaccines, respirators, or PAPRs. While a Hail Mary doesn't technically mean that it "doesn't work," it's pretty close, and making a big deal about these distinctions is becoming a little pedantic. For the vaccinated, recommending masks (which may or may not offer a tiny extra bit of protection) over vastly more effective respirators due to a small-to-none hassle factor seems a bit silly. It's much more reasonable to recommend respirators or nothing.
You said something about "pure speculation" earlier, but I think that's what you're engaging in here. What makes you think that masks offer "10%? 1%? 0.1%?" protection? Indeed, what do you mean by those numbers? What actual model of transmission leads you to think this? [EDITED to add:] Actually, maybe I misunderstood you; is the "10%? 1%? 0.1%?" meant to be an amount of protection (in which case, why also say "and likely to be so small that ..." -- aren't you saying the same thing twice?) or a probability of any protection at all (in which case, what are you smoking?)? Again, what actual model of transmission is this based on? What masks do is to reduce the fraction (viruses breathed in) / (viruses breathed out), by blocking the passage of droplets or changing the pattern of air flow. Unless the later strains have evolved teleportation or something, it seems unlikely that the factor by which this fraction has reduced in any given situation is much different now from before. I've had trouble finding really convincing figures, but it seems like the typical factor for a surgical mask is somewhere on the order of a 3x-10x reduction in the "outward" direction, larger when speaking or coughing than when just breathing normally (which is good, because speaking and coughing make you emit a lot more viruses if infected). "Inward" protection seems to be somewhat less -- maybe 2x? Home-made cloth masks appear to be substantially inferior to (ordinary, cheap) surgical masks. This would mean e.g. that the amount of time you need to spend near an infected person in order to get infected yourself is 3-10x greater if they're masked, 6-20x greater if you both are. If you think that somehow none of this works any more because of the newer more infectious strains, or that it's all nonsense and actually masks never had any substantial effect at all, then you should either give credible evidence that it doesn't or explain a plausible way in which it would have stopped working (or
We know that masks have poor performance, because while masks seem to have eliminated the flu and to have stopped or significantly slowed down the spread of the original variants, especially in Asia, masks failed to have the same effect on the newer variants which caused the fall/winter wave. And since Delta is even more contagious than those variants, masks will be even less effective than they were during the last wave. How you can claim that this is not evidence is beyond me. I also doubt that it's useful or even possible to accurately calculate the efficacy of masks for the latest variants without doing lab work. How would you quantify the decrease in efficacy caused by increased infectivity due to better binding of the variants to human cells? How much more virus particles do these variants produce, if any, and how does that relate to mask efficacy? Is mask efficacy a spectrum or are there thresholds that suddenly render masks 0% effective in most or many situations? But since we already know that masks offer poor performance anyway, quantify exactly how poor the performance is (10%, 1%, 0.1%, or whatever) doesn't seem to be all that important. Masks have become a dangerous distraction from far more effective interventions.
You still seem to be assuming that "poor performance" = "not on their own sufficient to stop the latest variants growing exponentially", and that is just unambiguously wrong. Do you, or do you not, have any information that isn't broadly consistent with the following crude model? 1. An infected person emits virus particles at some (somewhat random) rate, more when speaking or coughing than when breathing normally. 2. If you breathe in virus particles, there is some probability that you get infected; the probability is higher when the number of virus particles is higher. 3. If the infected person is wearing a mask, then the rate at which they emit virus particles is reduced by a constant factor somewhere in the vicinity of 5x. 4. If the not-yet-infected person is wearing a mask, then the number of virus particles reaching them is reduced by a constant factor somewhere in the vicinity of 2x. 5. Newer variants are more infectious, meaning some combination of (a) infected people emit more virus particles, or (b) the probability of infection for a given intake of virus particles is higher. If this model is somewhere near the truth, then the only way for masks to be near-useless ("10%, 1%, 0.1%", as you put it -- but I asked you to explain what these numbers are supposed to mean and you didn't, and I would still like you to) is if the newer variants cause such a colossal increase in the number of virus particles put out by an infected person, or in how effectively infectious they are, that being near them even briefly basically guarantees getting infected. Because otherwise, if you and they both wear masks then that means something like a 10x increase in how much exposure you can have before getting infected, and if the unmasked figure would be (say) 2 minutes' conversation at a 2m distance, then that would turn into 20 minutes, and I at least have plenty of conversations that are longer than 2 minutes but shorter than 20 minutes. Do you have good reason to believe tha
Given the fact that we already know that masks have poor performance based on the what I've already mentioned, models are pointless for most situations. If you're referring to modelling a strategy of maximizing personal (rather than public) protection with a poor performing tool, models could help you do that, but in the case of masks, it will turn out that most strategies are impractical because 1) there will be too many variables to keep track of, 2) some variables will be impossible or hard to obtain, and 3) some variables will be hard to control even with perfect knowledge. With or without a mask, if the distance between people is far enough, infection will be avoided regardless of infectivity due to dilution. If the distance is between two people, you may be able to calculate a minimum safe distance if you know all of the variables. Some of these variables are room size, ventilation, infectivity, mask type, rates of breathing and vocalization, and vaccination status. You'd also need to know if the room was previously occupied and by whom. Some of these these variables will be known but some will not be. You'll also need to recompute these variables once they change. If you're dealing with a simple model with two people in which nothing changes, this strategy might work. But real world cases are almost nothing like this. What if you go to another room or another person walks in? Is the ventilation the same? Is the person vaccinated? What kind of mask are they wearing? How many people were in the room before you walked in (aerosols can become suspended for hours even if the people that generated them are no longer around)? Modeling this stuff quickly becomes impractical, and if you can avoid that by wearing a respirator, why bother? "10%, 1%, 0.1%" was meant to poke fun at the attempt of precisely quantifying the poor performance of masks and is not based on any data.
I don't think further discussion in this thread is likely to prove fruitful.

Thanks! I'm not particularly worried about outdoor things (though your comments on crowds help fill in a gap where I had some uncertainty). More curious whether to attend indoor events without a vaccination restriction.

2 comments, sorted by Click to highlight new comments since: Today at 7:22 AM

WSJ reported that about half of adults infected in an outbreak of the delta variant in Israel were fully vaccinated with Pfizer,

Isn't that what we'd expect with their high vaccination rates, even if the vaccine was still really effective but not 100% effective?

That's a good point, so I guess the more important statistic is the lower effectiveness of the vaccines against the Delta variant. But would need to crunch some numbers to figure out how much that decrease in effectiveness matters in practical terms.