Icosian Reflections

…a tendency to systematize and a keen sense

that we live in a broken world.

For whom it tolls

Another may be sick too, and sick to death, and this affliction may lie in his bowels, as gold in a mine, and be of no use to him;
but this bell that tells me of his affliction, digs out, and applies that gold to me: if by this consideration of another's danger, I take mine own into contemplation, and so secure myself...

I was talking with a friend the other day, and the topic turned to vaccines. It's expected that the Sinovac and Pfizer vaccines will become available roughly simultaneously in Hong Kong, and so the question was, which vaccine we'd would prefer to receive.

Two topics that came up were safety and efficacy...


On safety, one can ask whether the Sinovac vaccine should be trusted quite as much as the ones developed in the West. (Hey, one can ask just about anything...)

Well, medically speaking, CoronaVac is a relatively conventional killed-virus vaccine. There's significant trial data available on it, as it has been given to more than a million people in mainland China since it received EUA in July. They've had months to observe side effects.

The issue is that that trial data is in the hands of Sinovac, a state-owned enterprise. And the question, then, is whether the Chinese authorities might have EUA'd the vaccine even if it had side effects that would make you, personally, balk. If it did, and even if it was massively good for society as a whole to get it, you might prefer to pass on getting it in your arm.

But that's just speculation; how likely is it to be true? Well, if CoronaVac's minor negative effects would be suppressed (and you assume they couldn't possibly be in the West), then your Bayesian posterior should assign greater weight to the chance that there is something to be concerned about (compared to the Western vaccines which had Western trials).

Except -- and this is very important -- what do you make of Sinovac's choice to run phase-3 trials in Brazil, Indonesia and Turkey? Well, for starters, the efficacy trials have to be run outside of China, because there isn't enough Covid-19 in China to test whether the vaccine is preventing infections.

But I think you should also believe that Chinese decisionmakers (who presumably have the safety data from the million-and-some Chinese citizens already vaccinated) would not choose to run foreign trials at all if they knew those trials would reveal side effects that would be unacceptable by global standards. If there were any side effects, they would know (even if they chose to keep them quiet), and wouldn't have authorized the foreign trials. To the contrary, the very fact that Sinovac has gone ahead with trials abroad seems to me like credible evidence that they know the vaccine to be safe -- even by the Western standards that will be applied to the foreign trials by the foreign press.

So on safety, I weakly think the Sinovac vaccine comes out in the clear -- even if you don't trust the Chinese regulators at all.


What about efficacy? The CoronaVac numbers aren't yet final, but the early numbers look materially lower than the 95% risk reduction that's claimed about the Pfizer and Moderna mRNA vaccines. Is it worth getting the less-efficaceous vaccine, or should you hold out for the other one?

Here's one thing: That's the wrong question. Vaccination isn't for you.

It's easy to think that it is, because it goes in your arm, and there is some benefit to you. But if you're in the demographic group with <<0.5% infection fatality rate (and I predict that you are, dear reader), then the principal benefit of the vaccine going in your arm is in slashing the expected number of people you might infect if you are infected, cutting R0 below 1, and suppressing this thing out of existence.

And if the most important thing is getting spike protein antibodies into people's bloodstreams, then the difference between a 95% reduction, an 80% reduction, or a 50% reduction is just a matter of "What's the alternative?"

While Hong Kong has purchased 7.5 million shots each from Sinovac, Pfizer, and AstraZeneca (together enough to double-shot the population 1.5 times over), I wouldn't be surprised if there's a significant shortfall of demand for "the Chinese vaccine", especially if it becomes available around the same time the Pfizer shot is.

If that happens, and the Sinovac shots just sit on the shelves while people try to get the "better" Pfizer vaccine, you can sign me up for CoronaVac. Because a shot -- any shot -- in one more arm is better for us all.

Even assuming that the Pfizer shot is more effective, the perfect isn't just the enemy of the good here -- it's the enemy of saving lives.

edit 2022-02-28: I think the thesis and argument above holds, except for one piece outside of my model, which makes the bottom-line recommendation fall apart. Demand for Pfizer way outstripped demand for Sinovac -- by at least 2:1 when last I checked -- but both fell way short of available supply and so there was no tradeoff to face in practice. If you wanted a shot at all, there were enough Pfizer shots for you to have one.

In a later post, I retracted by suggestion of taking a Sinovac shot to ease demand. In the end, I took a three-dose course of the Pfizer mRNA vaccine, between April and December 2021.


We had essentially the same problems with testing last year, because key parts of the medical system (in the West) couldn't deal with the fact that testing also isn't for you.

A self-administered paper strip test has a 50% false negative rate but can be processed at 5% the price? It's effectively useless as a diagnostic instrument, but that's irrelevant -- we're not using it to diagnose you for treatment; we're using it to suppress a pandemic.

To suppress the pandemic, we don't actually need to know with particular certainty whether you, specifically, have the virus -- we just need to catch enough pre-symptomatic cases that enough infectious people stay inside and we cut R0 to suppression levels. If catching half of them at scale (50% false negative!) means that a third of the infectious people mostly stay inside, the test is doing just great (and we should have had it everywhere eight months ago).

But no, the FDA refused to approve tests that were inferior to PCR tests (which reportedly took US labs ~a week to process) because the regulator has no regulatory framework for even thinking about whether we should have a thing that's worse at diagnosing a specific patient but better at stopping a pandemic: (emphasis mine)

Mina: The only pathway that we have to evaluate tests like this are medical diagnostic pathways, pathways designed specifically to ensure that a physician like a detective is getting all the information they need to diagnose a sick person... We have so devalued and defunded public health...that we don't have a regulatory pathway to approve a test whose primary objective is stopping an epidemic versus diagnosing a sick person. And that has held everything up.

All the companies that could be producing these rapid tests in the millions and millions, they have been sitting on these tests trying to hone them so they can pass FDA standards as a medical diagnostic... If we can do the cheap version they can be made very fast but the just won't get through the FDA.

Gladwell: I find your explanation unconvincing. How dumb is the FDA?... If you make the exact argument you made to me...the FDA is not going to see your logic?

Mina: It's not that [the FDA] are not smart it's that this is a regulatory body, they just don’t have a pathway. You can't apply for approval for a public health test tool... In our country the medical establishment is extremely strong, you can't go to get a cholesterol test without getting a prescription from your doctor. Why can't we know that? It's all through this very heavy medical lens and changing that, getting that big ship to turn is turning out to be a very, very difficult task but leading to potentially tens or hundreds or thousands of deaths that don't need to be happening. (...)

The only thing that the FDA even has tried to do is regulate things that are for you. We can talk some other time about whether it does that task well, but the regulator isn't. even. trying. to make sensible policy for the things that look like medicine but aren't for you at all. Things that aren't for you just aren't on their map.

Oh, except that if it looks even vaguely "medical", then it's banned by default -- unless we can prove that it's the best possible thing for you, out of all the possible things. For your protection, of course.


It's the same yet again with "first doses first". We can give 85%-efficacy treatment to 20 million people, or 95%-efficacy treatment to 10 million.

Doctors, with a duty of care to the specific patient sitting in front of them, are loath to administer an inferior treatment that hasn't been blessed by Serious People. Understandable; doctors, by custom, have an explicitly special relationship with "their" patients -- and no corresponding special duty to humanity-at-large.

And for a regulator who only thinks about regulating medical care for you, there's no question at all. It's unthinkable to do the inferior thing. The arguments for it just aren't in scope.

But one of these choices kills tens of thousands of people who desperately want to live. The other will see them with their families at Christmas. And we, as a society, are choosing death.

Everybody knows that everybody knows that there's no other choice.

A circle of stones reads "UNLESS", from Dr. Seuss's "The Lorax"

"It's a lot more complicated than that—"

"No. It ain't. When people say things are a lot more complicated than that, they means they're getting worried that they won't like the truth..."


I have another post to write about epistemology and the philosophy of science, as practiced by "experts" and institutions in the year 2020. It has much sharper words than this one. (preview: 1, 2)

And I have another, more personal reflection on intellectual humility and my trust in public consensus in 2021. It's much angrier than this one. (partial preview: 1, 2, 3)

And while I'm signal-boosting things, Zvi Mowshowitz is the single most useful source to be reading about Covid-19 in the US. (12/31 post)

But that's enough for now, I think.

No man is an island, entire of itself; every man is a piece of the continent, a part of the main; if a clod be washed away by the sea, Europe is the less, as well as if a promontory were, as well as if a manor of thy friend's or of thine own were; any man's death diminishes me, because I am involved in mankind, and therefore never send to know for whom the bell tolls; it tolls for thee.