The case numbers this week were clearly bad news. The raw count was somewhat bad news, and the positive test percentage increase was very bad news. It would be easy to treat the whole shift as fully ‘real,’ attribute it all to Delta, and panic.

I do not think that is the correct interpretation. What we are seeing matches what we saw a year ago, so a lot of this is a seasonal and regional change that has nothing to do with Delta. It’s also likely that some of the shift in percentages comes from data being wonky rather than the underlying conditions. Not only do we have alternative explanations, the size of the shift doesn’t match the incremental change in the amount of Delta out there, even if (as I suspect) it’s a rather dramatic takeover, with Delta’s share of the pandemic in America rising 25%+ in a single week.

It’s also going to be tempting to attribute seasonal weather effects to the local vaccination rate, since the two are highly correlated. Differences in vaccination rates in different areas matter a lot, but that’s not central to what is happening this week.

Thus I am still expecting some regional outbreaks, and am still not expecting nationwide problems, but one must ask about whether the winter is going to bring trouble the way last winter did. The hopeful answer is that vaccinations will be far enough along by then to not matter, and the second hopeful answer is that even if it’s bad it won’t be anywhere near as bad as last time. The vaccination numbers this week were quite good.

Still, I miss the confidence I had two weeks ago. Let’s run the numbers.

The Numbers

Predictions

Prediction from last week: Positivity rate of 1.8% (unchanged) and deaths fall by 8%.

Result: Positivity rate of 2.4% (up 0.6%) and deaths decline by 8%.

The case numbers reflect a <10% jump in cases, yet we have a 30% jump in positivity rate. This suggests some combination of a decline in testing and quirky data. I’m no longer confident that the positivity rate is the best measure of the state of the pandemic in America, and am relying on case counts more. A lot of that is no longer fearing that case counts are being manipulated the way they clearly were some of last year, or worries about testing supplies.

Prediction for next week: Positivity rate of 2.7% (up 0.3%) and deaths decline by 5%.

I expect a lot of this effect to be seasonality rather than Delta, but cases likely will rise for a bit. Deaths should still be declining somewhat.

Deaths

Date	WEST	MIDWEST	SOUTH	NORTHEAST	TOTAL
May 20-May 26	615	948	1279	631	3473
May 27-June 2	527	838	1170	456	2991
June 3-June 9	720	817	915	431	2883
Jun 10-Jun 16	368	611	961	314	2254
Jun 17-Jun 23	529	443	831	263	2066
Jun 24-Jun 30	550	459	706	186	1901

This is exactly on track as an average, and I’ve come to accept that week to week numbers in regions bounce around. If anything, it’s surprising that the death numbers nationwide have been so steady when there’s clearly a lot of data collection timing issues and random fluctuations going on. Deaths lag by several weeks, so the bad news from the last few weeks hasn’t had an impact yet, nor has Delta’s increased lethality mattered much yet, but it’s likely that will stop soon, and we won’t get much lower than this for a while.

Cases

Date	WEST	MIDWEST	SOUTH	NORTHEAST	TOTAL
May 13-May 19	39,601	45,030	63,529	34,309	182,469
May 20-May 26	33,890	34,694	48,973	24,849	142,406
May 27-June 2	31,172	20,044	33,293	14,660	99,169
Jun 3-Jun 9	25,987	18,267	32,545	11,540	88,339
Jun 10-Jun 16	23,700	14,472	25,752	8,177	72,101
Jun 17-Jun 23	23,854	12,801	26,456	6,464	69,575
Jun 24-Jun 30	23,246	14,521	31,773	6,388	75,928

I found a new site this week that offers a great view of the data, so here’s the week over week changes by state:

It’s hard to read here, so consider clicking through to see it there. Those are dramatic increases in southern states, and the map makes it very clear that these differences are not primarily about how many people each state has vaccinated so far. The big difference is north versus south, and it looks a lot like this map:

Things have been record-breakingly hot in many places recently, especially the Pacific Northwest, but that only happened in the past few days and it was sufficiently dramatic (both record breaking and >100F) behavioral patterns likely went straight to ‘don’t go outside at all for any reason’ rather than a bunch of meetings indoors.

Also worth remembering this from last year, which tells a very similar story so far.

Date (2020)	WEST	MIDWEST	SOUTH	NE ex-NY	NY	Total
June 4-June 10	35487	24674	55731	16622	6071	138585
June 11-June 17	41976	22510	75787	12905	4986	158164
June 18-June 24	66292	26792	107221	10922	4524	215751
June 25-July 1	85761	34974	163472	11890	4413	300510
July 2-July 8	103879	40139	202863	13376	4850	365107
July 9-July 15	108395	53229	250072	15199	5077	431972
July 16-July 22	117506	57797	265221	16037	4880	461441
July 23-July 29	110219	67903	240667	21301	4707	444797
July 30-Aug 5	91002	64462	212945	19152	4632	392193
Aug 6-Aug 12	93042	61931	188486	17091	4478	365028
Aug 13-Aug 19	80887	63384	156998	16358	4499	322126

Last year at this time we saw an explosion in cases. This year we are seeing a halt in the decline in cases. That’s not good, but it’s also not a reason to panic. Nor is it reason to attribute the shift entirely to Delta. I am curious why there is so little discussion elsewhere of these obvious patterns this time around, when seasonality has been a talking point in the past.

(Actually, I don’t wonder all that much, because the official Very Serious Person narrative wants to worry as much as possible about Delta, so it would ignore alternative explanations.)

Vaccinations

We picked up an additional 0.8% of the population getting their first dose, and 1.3% of the population got their second dose. The uptick from last week gives hope that we may be in something approaching a short-term steady state, or perhaps talk of Delta has convinced some hesitant people to get the vaccine. That makes sense, since the selfish value of being vaccinated is no longer rapidly approaching zero. It’s becoming increasingly clear that there will be social benefits for some time.

If we can sustain this pace of vaccinations, that is excellent for our ability to close things out. It’s happening in spite of a lot of headwinds. Vaccine misinformation, misrepresentation of information and hesitancy continues to frustrate. There has been quite the abundance of self-inflicted wounds, and giving them attention is always a double-edged sword.

Thus, while I hesitate to give the topic attention and the results could easily be misinterpreted, this seems like useful data:

Looking at the comments, it’s clear that this did not sufficiently disambiguate ‘severe’ and that the category still mostly covered people who got knocked on their ass for two days.

Plus we have to worry about anti-vax people voting combined with the lizardman constant. I’d similarly be unsure whether to say none versus mild for my shots. So this seems consistent with ‘essentially zero side effects other than some people being knocked on one’s ass for a day or two, which occasionally got considered severe.’ It also is clearly inconsistent with the conspiracy theories anti-vax people (and the person everyone needs to stop asking me about) are throwing around.

Hopefully that will be the last I need to say on that matter in any form.

In other good news, mix-and-match vaccines, as one would expect, work quite well, and we finally have a study on that (study).

The control was one shot of AZ for completely insane You Fail Mathematics Forever reasons.

But that’s fine, because we don’t need a control when we’re measuring antibody response, and the whole control group was fetishistic Science(TM) rather than a source of meaningful data.

As an alternative, we have at least one health official in Australia, who happens to be the QLD chief medical officer, talking obvious nonsense:

The idea that AZ poses a greater risk than getting Covid makes zero statistical or mathematical sense, and framing the issue that way indicates that blame avoidance and Sacrifices to the Gods of responsibility are what matters here rather than any attempt to do math on a physical world model.

It would be one thing if they had plenty of mRNA vaccines to use instead. They don’t.

It would be a somewhat different thing if you couldn’t mix and match shots. You can.

That’s all before concerns about population-level effects.

Thus, this should come as no surprise:

I think we can lay to rest the hypothesis that Australia did better than other countries because it was more sane and has wiser systems for making decisions. Australia did better for other reasons, including being an island, that led to a different equilibrium. Now that we are in the vaccination phase of the pandemic, Australia is utterly failing.

Delta Variant

This Nature post provides a perspective of how many are thinking about Delta. As usual, there’s talk about whether the new variant will take over and whether it can be prevented by some magical force, rather than how quickly it will take over and the need to accept that reality, but mostly the reality is being accepted here.

They are worried about Africa:

I continue to not be as worried, because I do not expect that we are approaching the limits of Africa’s control systems. I expect there to be enough slack to absorb Delta without things going critical. I’m not super confident in that, but I do think it’s a solid favorite (~75%).

As estimates of prevalence of Delta go, using this seems like a reasonable method:

Alpha was previously on the rise, so if it declined from 70% to 42%, it’s safe to say that variants that are substantially more infectious than Alpha are replacing both it and the remaining 30% (to the extent that it wasn’t already such variants), so if we think Gamma is mostly similar to Alpha as we did last time, that provides a lower bound for Delta of 40% by mid-June.

Contrast that with last week’s calculation from another source via taking the numbers from Delta directly, which had it at 30%. That’s a reasonably big gap, since here 40% is a lower bound, and if the previous mix included a bunch of stuff less infectious than Alpha we’d expect this to be closer to 45%-50%. And that was two weeks ago, so if it was 40% then it’s presumably more like 70% now. If that math seems quick, keep in mind that we didn’t even break 1% until April.

In other news, look at where that other source is after its last update:

This continues to be consistent with the sequencing data there being accurate but effectively delayed. Note that P.1/Gamma is now declining rapidly as well. Over the last 60 days they have Delta at 11%, but over the last 30 days they have it at 21%, which means it was something like 1% in the previous 30, and 30 days from whatever ‘now’ is in that calculation it implies Delta will be above 80% barring geographic barriers slowing things down.

If we estimate that the infections observed now are happening in something like a 65% Delta 30% Alpha/Gamma 5% Other world, with the remaining others largely similar to Alpha, then we’ve absorbed two thirds of the transition from Alpha to Delta, and three quarters of the transition from the old strain to Delta. Going forward, one should expect the share of old strains to be cut in half every ten days or so, and Delta to be almost all infections in most regions by August 1.

Olympics in Japan

The Olympics are here, and Japan is not exactly fully vaccinated, as they’ve been delaying things quite a bit, with Moderna and AZ not even being approved until May 21.

Once again I’m going to come out in favor of doing things that scale, are central to the experience of life and bring lots of people joy, even when they’re not fully ‘safe.’ Any athlete going to the Olympics who wants to be vaccinated, one hopes, has already been vaccinated – if not, it would be trivial to take care of that. Same goes for those who need to be in the Olympic village, it’s not that many people.

What’s crazy is allowing a bunch of spectators who aren’t vaccinated into indoor events, and it looks like there will be 10,000 spectators per event, whether indoors or outdoors.

The idea that something not being ‘the safest way’ means someone is ‘not on board’ continues, and reflects a mindset we need to move past. The safest way to do most everything is to not do it at all, yet things must be done.

One way to think about whether the spectators are providing enough value is to look at the ticket revenue. There are 330 events, so assuming everything sells out, tickets are going to average $121 per full event, many of which are over multiple days. That’s not that much money, so my instinct is they need to charge a lot more at least for indoor events, to justify the health risks involved.

It’s always important to think about base rates…

…where Japan continues to outperform the United States. If Delta wasn’t involved I’d say that this is all basically fine, but it remains to be seen if Japan can handle Delta in its current state, and this could take substantial time ‘off the clock’ for vaccinations to catch up.

The Olympics cost about $12 billion to host and can leave a lasting legacy. If one thinks that the Olympics will be remembered for a long time, I understand taking the full risk, but I still wouldn’t have substantial crowds at indoor events, especially minor ones. Why ask for that kind of trouble?

In Other News

A bunch of people this week mentioned how bad it was that people were making base rate errors, and using ‘some vaccinated people got infected’ as the latest scare tactic in full How to Lie With Statistics mode. Nothing to see here that we didn’t expect.

Thread in praise of the excellent Microcovid project. I disagree with some of the Microcovid calculations but having any plausible calculations at all was the important element here, giving certain types of people a way to both do any of the things ever and not do many of the stupider things. That doesn’t mean the paralysis before that wasn’t a failure mode that requires investigation and correction. The general solution might be ‘if microcovid does not exist it would be necessary to invent it so people have a Schilling estimate to converge on’ and that’s not great but better to know it than not know it.

A good way of framing what Zeynep and I both point out about the question of the origins of Covid-19: What matters is how such an event could have occurred, rather than the way it actually did occur. If the plane could have crashed due to a faulty part, but it turned out something else caused this particular crash, the part is still faulty and we need to fix the root cause of that failure.

If Gain of Function research puts us in danger, that’s enough reason to ban it, whether or not it actually did kill millions of people this particular time around.

Zeynep also points us to this paper about potential lab leaks explicitly warning against exactly the things we frequently do… from 2015.

Abu Dhabi drops the hammer.

No, we can’t know for certain that vaccines provide years-long immunity without waiting those years, but I do think the extrapolation process here is mostly reliable, and a much less misleading statement than the typical ‘provides protection for up to X months’ where X is how many months we’ve had the time to check, which then gets quoted by media as if X is an upper bound rather than a lower bound.

Marginal Revolution previews/reviews the new Michael Lewis book about those preparing for a pandemic, and what they did during the early days of pandemic they’d prepared for. The book, titled The Premonition, if Alex’s summary is accurate, holds the CDC (Delenda Est) in even lower regard than I do, as our heroes who see the pandemic coming are stymied every step of the way, preventing them from having most of their potential impact, while we don’t so much as quarantine or even test those returning from China, let alone take reasonable precautions. As Alex notes, it’s weird to have a (true) story of lone heroes fighting the good fight against the system to stop a global catastrophe, who then completely and utterly lose. A story at least as worth telling as when they win, but definitely hard on the cliche structure.

Paper about long term past effects of pandemics in Sweden over 220 years.

Potential good news in a path towards an all-coronavirus vaccine, I don’t know if it’s meaningful or not but passing it along.

Not Covid

If your goal is to avoid claims that our elections are full of fraud and tabulation errors, and/or you don’t want to ruin ranked choice voting for the rest of us, may I suggest perhaps making more of an effort to not do things like this, as happened in the NYC mayoral election:

Thus, the Evergreen Tweet: I hope we will hear the end of this, but fear that we will not.