I remain open to convincing arguments in favor of reserving second shots but all the ones I’ve seen seem very weak.

Prof Bergstrom is a smart and informed person so it's unsurprising that he would offer thoughtful introspection on his intuitive aversion to one dose. All the opposition I've seen is less articulate versions of this - just a gut sense that it's risky. https://twitter.com/CT_Bergstrom/status/1345205648854716416?s=20

But I think this view is incorrect. Double dosing is extremely risky, just in a less intuitive way. The risk is wasting extremely scarce, highly valuable vaccine doses on low uses. That is a *catastrophe*. It's not as obvious as dumping vaccines in a river, but amount to same.

My view is this: right now the value of immunity is at its peak, and vaccine availability is at its nadir. Even in the worst case scenarios where first dose efficacy wears off, I think it's worth it to shift immunity forward.

Value of immunity is at its peak right now because:
1) there are already a lot of cases out there right now, so a lot of infectious people
2) the newer, more infectious strain is about to let loose in the US. It is letting lose already right now.

Vaccine availability is low because:
1) production is still ramping up
2) we only have two approved vaccines. more are on the way including, hopefully, J&J's one-shot vaccine.

We know for sure that the first shot gives a lot of immunity. Close to as much as with the booster. This is for Pfizer, but the Moderna graph is similar. https://twitter.com/GarettJones/status/1345271157667241985?s=20

It seems likely based on evidence from other vaccines that at the very least, the first dose will remain effective for weeks beyond the 21-day mark https://twitter.com/sandyddouglas/status/1344949269535608832?s=20

But even if you think the first dose will start to wear off some after 21 days, surely it's not going to just drop off a cliff and become ineffective on the 22nd day. But guess who definitely has no immunity on that 22nd day? Someone who hasn't gotten any shot yet.

Production is going to be increasing exponentially. It would be *tremendously* more efficient to wait, say, six weeks instead of three. Eight weeks, even better. Vaccines are going to start getting less scarce quickly.

At that point, the opportunity cost of giving the second dose will be much lower, both because the person you're taking the shot from is lower risk and because, due to all the vaccinated ppl out there, transmission should be lower.

So I think by far the greater risk is in being too conservative, giving booster shots to people who don't need them while the virus rampages across the country, continuing to kill thousands of people a day.

It would be bad enough, in my view, if we rigidly stuck to the schedule followed in the trial (chosen, btw, afaict, only to speed up the trial results, not for any clinical reason). That would take shots at the 21-day mark from ppl who should then be getting first shots.

But it's even worse than that. Shots are being *held in reserve* to give to people in three weeks. Shots that could be first shots *today* for someone else are instead being held for the person who's getting their first shot today.

Given the coming increases in production, to me this seems completely unreasonable. You can take those second shots out of future production batches rather than taking them out of batches available now.

Again, bringing vaccination forward in time is of massive value here - case numbers are high now, vaccines are going to be more widely available with each passing day. Why you would instead shift vaccine doses away from the present into the future is just beyond me.

Btw, apparently it's the federal government doing this:

"The federal government will hold back product initially to ensure second doses are available." https://www.tennessean.com/story/news/health/2020/12/20/us-rushes-give-covid-19-shots-tennessee-builds-vaccine-reserve/3981650001/

The absolute last thing we should be doing right now is "holding back product."

https://twitter.com/emilybazelon/status/1345233974394843136?s=20

Gonna try to illustrate w a few graphs here. In each case, green line represents total expected infections between two people. In this case, the treatment and control. This is the red and blue lines added together from the Pfizer data (as approximated by my fat finger in markup)

Here’s if the second shot changed nothing for the person who had one shot, and that shot were instead given to the “control” person. In this case, total expected infections between the two is just twice the blue line*

*in reality this blue line is probably pessimistic bc not only solid infections between these two go down, but they would be less able to infect other ppl, including each other

Here’s with a significant but modest falloff in immunity (new blue line). Total expected infections is still the green line. Notice that once immunity is entirely gone, slope, not level, returns to that of the red line.