That’s the new variant for this week. I am not worried about it, but there is little other news on the Covid front and I’m still attempting not to reuse titles. As time goes by the events are still new but they tend to rhyme. This update is inessential.

In addition to various non-Covid stuff I’m putting at the end, which I anticipate will slowly grow as a section over time, this week’s other post is Criticism of EA Criticism Contest. If you find that is relevant to your interests, with an additional copy on the EA forum. If it is not relevant to your interests you can safely skip it.

Executive Summary

  1. BA.2.75 is potentially a thing, unclear if it is a meaningful thing yet.
  2. Spike in deaths that is likely a data artifact.
  3. A reminder we could be working faster towards

Let’s run the numbers.

The Numbers


Prediction from last week: 700k cases (+5%) and 2,250 deaths (-1%)

Results: 746k cases (+10%) and 2,885 deaths (+27%).

Prediction for next week: 750k cases (+1%) and 2,450 deaths (-15%).

With BA.5 taking over it makes sense for there to be a modest rise in total cases about now, so even though I think the jump was partly a holiday backfill my guess is we catch back up there. For deaths the details look a lot like backfills and spikes in reporting rather than an actual increase in deaths, so my guess is that it largely reverts but uncertainty prevents predicting a full reversion.


Diving into the state-by-state data shows a bunch of places that are getting increasingly lazy with when they report deaths. A bunch of them happened to report substantial numbers of deaths this week (and Florida, which reported zero, was adjusted while the others weren’t because it’s not obvious how to do that properly). The regional numbers end up looking mostly sane aside from the Western spike.


An estimate that at least in the UK we’re catching almost none of the child cases anymore, and not that many adult ones either. Seems likely to be right.

BA.2.75 Our Price Cheap

BA.2.75 is suddenly something people are talking about as rising rapidly, a subvariant of BA.2. We will not be calling it Centaurus. Here we go again?

Go go gadget log scale.

And yet:

My guess is this probably isn’t going to be an impactful thing, to the extent it is a thing at all. If it is, we’ll have plenty of warning signs that are much louder than this. Local surges are nothing new by themselves.

Universal Variant Vaccines Our Price Cheap

Patrick Collison of Stripe and Fast Grants writes that with a modest lifting of regulatory barriers and a small (~1bb) investment we could greatly accelerate universal Covid vaccines that would protect against all current and future variants.

There isn’t that much to say here. Developing such a vaccine would be a big win. Helping it along and speeding up deployment would be cheap. I do not doubt the premise here.

We are not doing it because we are not interested in doing helpful things regarding Covid. That is a shame. It would be good if there was a way to do helpful things but not unhelpful things, but I do not know what that would be.


Mina thread reminding us that BA.5 isn’t more immune evasive except insofar that it is different from previous strains. Against a not-yet-infected individual there’s no difference.

New White House briefing on pandemic strategy for BA.5. Same stuff. Take-homes (or, really, remain-at-homes):

I do not know what change of strategy could be expected to be net useful.

Bloomberg opinion piece on ‘How bad is BA.5? Bad enough.’ Meaning standard Long Covid generic worries, basically.

Atlantic article about BA.5. Headline is that it ‘is a test of our pandemic priorities’ and that seems right. When people say that, it is usually because they think the pandemic should be our priority. Luckily our priorities lie elsewhere.

Thread about BA.5 trying to give impression that there is an ‘inflection point’ of concern now that it is 70% of cases, despite no rise in cases and no rise in hospitalization or death. Mostly it makes the case that there isn’t anything to worry about here, both explicitly and even more strongly by implication.

Zeynep thread on B.5 and its reproduction rate, pointing out that one can’t simply multiply reproduction advantages of successive variants against each other. B.5 is growing, but not at a super alarming pace.

If Covid Infects a Human

And the human does not notice or care, does it still count?

This is not the first time this question has come up.

There are two ways to read this graph. One is that it is possible that BA.4 and BA.5 put dramatically more virus per case into the wastewater. The other is that we are now missing a much higher percentage of cases, and only catching maybe 25% as many as before. Presumably we’re mostly or entirely missing cases.

This also suggests it’s that we are missing cases:

This reminds me of analysis on human biases. Naturally, human beings have a lot of biases, largely because humans that have them have historically had higher reproductive fitness than those that didn’t. Part of this is that many biases act to cancel each other out in key situations.

In this case, some (relatively worried) people are using heuristics based on case counts to figure out risk and what to do about it. What we don’t want to do is fix the error they are making and account for all these extra cases, because they are making the same error in the other direction by failing to take into account that there are now so many cases that are entirely harmless.

Wastewater data is still great and we should invest in more of it. They’re improving the technology to make this information even better and easier to extract.

The best wastewater news is that a commenter pointed me to this aggregation site. Seems like an excellent resource.

Still, the distinction between cases and infections has turned out to serve us well by effectively taking into account (lack of) severity. We should be happy about that.

CNN calls the latest variant the ‘worst’ one, Nate Silver reminds us that this is insulting to people who know that the infection fatality rate is much, much lower.

The CDC, daring to not Do More about all these infections.

In Other News

Novavax vaccine remains illegal for the moment, but this will soon change. Presumably minimal impact but good to see. Continues be to be nuts that life saving medicines remain illegal long after we know they are going to be legal in the future.

Interesting thread about attempts to predict where Covid mutations may be heading in the future. The topline question, however, of ‘why can’t we use these predictions to have the vaccine stay ahead of mutations’ is rather obvious. We can’t use predictions to stay ahead because even if we had perfect predictions it would change nothing. Work could not usefully begin before the new variant had already largely taken over, the FDA and other authorities would not permit it. Every vaccine update will be delayed until after most of its value has expired.

Claim that ‘public health officials are not ‘following the science.’’ You see, we don’t have evidence these new vaccine shots work because there is so little risk to begin with we can’t get p<0.05 and that means we can’t ever know anything, so instead (not stated, but as the practical effect would be) we should treat these people as high risk due to them not being vaccinated. The true point of such vaccinations is to change the allocation of blame and worry so life can happen. Alas, other people have trouble saying this out loud.

If you do get Covid, Zeynep reminds you to make sure if at all possible to give yourself the necessary recovery time. The best way to recover is to take it easy and not rush back into things. To the extent there are worries about longer term effects, this is one thing that will help. This is also when you find out which things you do are really ‘work’ and where you need to take it easy, versus which ones aren’t and thus you don’t.

Should we expect Paxlovid resistance? Post says it is inevitable, but given how rarely we use it there would have to be little in the way of trade-offs to get that immunity. We shall see.

NPR gaslights us that experts ‘knew all along’ that Covid-19 was airborne and ventilation and filtration were important ‘and now the public is catching on.’

New study suggests that the damage from shutting down schools has been worse than almost anyone expected.” The number of people claiming they expected it is large.

Chinese reject vaccine mandate as unreasonable, force it to be withdrawn within about two days. Model this contrast to their widespread tolerance for other measures. Is it the children who are wrong?

Thread pointing out that while installing new air ventilation systems can be expensive (while still being worthwhile) the lowest hanging fruit is that usually there are already systems in place and we haven’t done proper maintenance, which is far cheaper.

White House is developing a plan to ‘let’ everyone get a second booster. Somehow we are still having conversations like this.

Simultaneously, ‘there’s a lot of talk about whether we should universally prescribe a second booster.’

STOP IT. We shouldn’t be simultaneously debating whether to make something mandatory and also whether to stop making it forbidden. We can do one without the other, and we obviously should be doing that. I do not intend to get a second booster, and I’d strongly oppose trying to pressure people to get it, but it seems even crazier that people need permission to get it. If you want another booster, there’s infinite supply, why shouldn’t you be able to get another booster?

Definitely don’t be like the director of the CDC and say that every six month old should get the vaccine any more than you should prevent them from getting it.

Also in stop it news, permanent midnight advocates never quit.

I wonder what she’s like at the beach.

Thread from Chise, data on updated Moderna vaccine continues to look great.

Attempt to set up a formal debate on the safety of Covid vaccines (HT: Robin Hanson). I will not be participating in any form but figured I’d pass it along.

Trump’s greatest achievement was Operation Warp Speed and the vaccines, so of course Trump is afraid to mention the vaccines for fear of alienating his base.

Kansas City, here we come cause where else are we gonna go.

The Blue Jays are still coming out way behind given the whole year they couldn’t even play games in their stadiums.

One way of looking at what happened.

And counterpoint.

New Long Covid study starting costing 1.2 billion. It is prospective. It aims to involve more children than adults for reasons I cannot fathom.

“We put this together to not miss anything,” Koroshetz said. “It’s kind of like a battleship. That’s part of the problem.”

Thus, it has taken over a year to get started, and costs over a billion dollars. Despite that, I still don’t know how they expect to get the answers they seek at the end of the four years, as I expect the core hard problem of correlation not to go away. And for everything to have changed by the time we get to four years later. That doesn’t mean we shouldn’t make the attempt, but I do not have high hopes.

Not Covid

Not in particular, but also that, what it means to say you have a plan edition, sounds correct, no notes.

A call to reform IRBs, which are badly in need of reform.

Recommending a post (and whole blog) by an old friend I knew by a different name who I lost touch with, and who became a medical examiner. She tells it like it is.

Story of the publication of a short story. It’s rough out there and the hourly even when you succeed seems terrible. If I ever write fiction sounds like posting it to the blog is the way to go.

Bloomberg post on AB5 and the trucking situation in California. Looks likely to be all trucked up at least for a while. It doesn’t look good. But only one way to know for sure.

Related: Democrats declare themselves party of freedom to do, believe and say whatever the Democrats have decided is good this week, and from being allowed to do otherwise.

A development conversation, by Noah Smith, from Twitter.

93% of data claimed to be available on request was not available upon request, says data that may or may not be available on request.

FDA Delenda Est, Mokeypox edition.

No, seriously, we have a million doses of vaccine sitting in a warehouse, that have been fully inspected by the EMA, waiting for the FDA while it fails to do its own inspection that it insists upon for no reason. As a result, at-risk people who want to get vaccinated for monkeypox cannot get a shot.

So yes, very much a pox on the FDA.

Addendum: If you are not furious already–and you should be–remember that during COVID the FDA suspended factory inspections around the world creating shortages of life-saving cancer drugs and other pharmaceuticals. As I wrote then “Grocery store workers are working, meat packers are working, hell, bars and restaurants are open in many parts of the country but FDA inspectors aren’t inspecting. It boggles the mind.”

Proposed FDA reform knows as the VALID Act. It’s 433 pages long which makes me highly suspicious this is trying too hard rather than taking a blowtorch to that which requires a blowtorch. No idea if these things collectively would do much of anything or not, but mostly it is presumably steps in the right direction given that steps aren’t going to be taken to address the underlying issues. My guess is it’s a meaningful step forward but nothing like what we actually need, but it wouldn’t shock me if it wasn’t.

Despite that (and while making it clear this is probably something it is fine to throw under the bus to get the upside if the upside is there), the first ‘additionally this bill will’ item is:

Strengthen oversight of cosmetics and dietary supplements

Meet the new bill, same as the old bill. Whatever would we do if there was insufficiently strong oversight of cosmetics?

Do not ask questions for which you do not wish to know the answer.

The AI Legal Alignment problem.

There is an explicit exception to prevent the contradiction in this case. That won’t always be the case. As an example, it seems like the UK is at least kind of banning the internet via the UK Online Safety Bill that requires monitoring everything in real time in case anything might be harmful to a child and requiring tons of paperwork outlining all one’s plans and business models in advance and generally making life impossible for anyone who isn’t a tech giant, in the name of cracking down on tech giants? Such is the way of regulatory capture everywhere. What happens when the UK forces you to inspect that which the EU requires you to not inspect?

And the AI Deliberate Misalignment problem.

That is quite the nightmare.


New Comment
10 comments, sorted by Click to highlight new comments since: Today at 3:58 AM

Just a reminder that this paper provides a cute way to estimate the "true" prevalence of Covid (or indeed any pathogen) in a community: geometric mean of case rate and positivity rate. found that the state,cases and test positivity rates.

Thus for instance based on today's numbers for the NYC metro area on

50.4 per 100K = .000504 13.7% positivity rate = 0.137

True prevalence = 0.831%

Regarding "A call to reform IRBs, which are badly in need of reform."

I'm happy to say that University of California, Irvine made some very positive changes a few years ago for social sciences research that lessened the burdens of the IRB. The national requirements (as given by the hhs) are actually pretty permissive for things that carry basically no risk for participants. Relevant for social sciences researchers are these two chunky paragraphs of exemptions taken from[1]. The gist is that research in which you do something that any normal person would be completely fine with, and then ask people questions on a survey in which you don't ask them their name do not require IRB approval. UCI then permitted researchers to self-determine if this applies to their research. They still have to fill out an IRB application, but it is very abridged and once you have it filled out you just stick it somewhere you can remember it in case you get audited and can start conducting research. It was wonderful. My first IRB experience had a four month turn around because they forgot about my application, and then afterwards I could get it done in between classes.

  1. ^

    (2) Research that only includes interactions involving educational tests (cognitive, diagnostic, aptitude, achievement), survey procedures, interview procedures, or observation of public behavior (including visual or auditory recording) if at least one of the following criteria is met:

    (i) The information obtained is recorded by the investigator in such a manner that the identity of the human subjects cannot readily be ascertained, directly or through identifiers linked to the subjects;
    (ii) Any disclosure of the human subjects’ responses outside the research would not reasonably place the subjects at risk of criminal or civil liability or be damaging to the subjects’ financial standing, employability, educational advancement, or reputation; or
    (iii) The information obtained is recorded by the investigator in such a manner that the identity of the human subjects can readily be ascertained, directly or through identifiers linked to the subjects, and an IRB conducts a limited IRB review to make the determination required by §46.111(a)(7).

    (3) (i) Research involving benign behavioral interventions in conjunction with the collection of information from an adult subject through verbal or written responses (including data entry) or audiovisual recording if the subject prospectively agrees to the intervention and information collection and at least one of the following criteria is met: (A) The information obtained is recorded by the investigator in such a manner that the identity of the human subjects cannot readily be ascertained, directly or through identifiers linked to the subjects; (B) Any disclosure of the human subjects’ responses outside the research would not reasonably place the subjects at risk of criminal or civil liability or be damaging to the subjects’ financial standing, employability, educational advancement, or reputation; or (C) The information obtained is recorded by the investigator in such a manner that the identity of the human subjects can readily be ascertained, directly or through identifiers linked to the subjects, and an IRB conducts a limited IRB review to make the determination required by §46.111(a)(7). (ii) For the purpose of this provision, benign behavioral interventions are brief in duration, harmless, painless, not physically invasive, not likely to have a significant adverse lasting impact on the subjects, and the investigator has no reason to think the subjects will find the interventions offensive or embarrassing. Provided all such criteria are met, examples of such benign behavioral interventions would include having the subjects play an online game, having them solve puzzles under various noise conditions, or having them decide how to allocate a nominal amount of received cash between themselves and someone else.(iii) If the research involves deceiving the subjects regarding the nature or purposes of the research, this exemption is not applicable unless the subject authorizes the deception through a prospective agreement to participate in research in circumstances in which the subject is informed that he or she will be unaware of or misled regarding the nature or purposes of the research.

[+][comment deleted]4mo 10

Apologies for asking an object level question, but I probably have Covid and I'm in the UK which is about to experience a nasty heatwave. Do we have a Covid survival guide somewhere?

(EDIT: I lived lol)

Most people don't realize that 6 hours of sleep, dehydration, and drinking the night before reduce IQ, memory, and focus substantially. But that doesn't change the fact that those things are easily and permanently fixable, if a little inconvenient. 

Brain damage, on the other hand, is permanently unfixable. But with a P100 mask, eating outdoors, and washing hands before touching face, avoiding permanent brain damage is a little inconvenient.

These mitigations will have to be carried out forever. You will never be able to dine indoors anywhere except your own home, always have to wear a P100 mask all the time outside of your home/a few places you deem okay, and you will never be able to touch your face before washing your hands again. People really struggled with that last one, especially; so many of us touch our faces without even realizing it.

Zvi's talked a good number of times about the quality of the long COVID studies and doesn't have much nice to say about them. Suffice it to say that the risk of permanent brain damage from COVID is not close to 1. Cost-benefit-wise, for many people, having to P100-mask-up/always eat outside except at home/basically never touch your face again is not worth it compared to the risks. It certainly isn't for me.

ETA: Forgot about relative risks; many of the other things we do have risk of permanent brain damage that might be comparable to that from long COVID. Driving is one example, traumatic brain injuries are definitely possible. We could reduce this risk by wearing helmets while driving, yet no one does.

We could reduce this risk by wearing helmets while driving, yet no one does.

Arguments like this, while strictly about consistency, tend to implicitly take the correctness of the status quo for granted. Unless you have thoroughly evaluated all of your current habits, this is unjustified. I'd prefer to be inconsistent and sometimes right than consistently wrong. 

That's a fair take; it reminds me of alcohol and tobacco, which were largely "grandfathered" in. We probably wouldn't allow them if they were discovered today.

"Luckily our priorities lie elsewhere" - why is that lucky?

A theme of Zvi's pandemic coverage has been that people's (especially those with authority) reactions to the pandemic have been largely counterproductive (e.g. school masking, mandatory boosters, vaccine approvals, etc.).

So if authorities are incapable of acting with regards to the pandemic in helpful ways, it's lucky that their priorities are not in acting with regards to the pandemic.

New to LessWrong?