That's not what the Pfizer paper showed. There was a confidence interval from between 20-70% iirc for the first shot. Maybe if we're lucky it's on the higher end, but I would stick to proven science if we're going to combat this thing effectively.
That 20-70% efficacy number includes the entire period between dose 1 and 2. It looks way better 10 days after dose 1, btw-- MLE around 80%.
e.g. single dose -- 50% efficacy (to be pessimistic); double dose -- 95% efficacy. 10M doses "in flight" monthly. 100 deaths/million/month. Ignore all effects on transmission (to be more pessimistic).
Scenario 1 (no logistics disruption, half held back): 5M dosed with 1 dose at day 0, 5M dosed with 2 doses at 1 month and 5M dosed with 1 dose at 1 month. 250 saved in first month, 725 saved in second month.
Scenario 2 (no logistics disruption, 25% held back): 7.5M dosed with 1 dose at day 0, 7.5M dosed with 2 doses at 1 month and 2.5M dosed with 1 dose at 1 month. 375 saved in first month, 837 saved in second month.
Scenario 3: (100% production disruption, half held back): 5M dosed with 1 dose at day 0; 5M dosed with 2 doses at 1 month. 250 saved in first month, 475 saved in second month.
Scenario 4: (100% production disruption, 25% held back): 7.5M dosed with 1 dose at day 0; 2.5M dosed with 2 doses at 1 month; (5M late for dose 2); 375 saved in first month, 612 saved in second month.
With a 25% reserve, you only get into the unexpected-only-1-dose regimen with a supply disruption of more than 50%... and even then it's still probably better than a 50% reserve.
Lowering reserves would still increase administration rate immediately (though not proportionally). Actually, we might even enter phase 1B in some areas and have a mostly parallel administration system start to come up...
It's unclear to me how much of the slowness so far in administration (13.5% of doses allotted are reported as administered so far)... is A) (fixed?) delays in reporting, B) (fixed?) delays in logistics getting doses to administration sites, C) limitations on rate of administration, D) holiday-related slowness. Only C really matters in the long term-- if it is ramping poorly compared to production than that's a problem. But I think it's premature to assume C is limiting when A, B, & D are surely large right now.
If you look at the cumulative incidence curves (for Moderna see [1] page 28), you can see that COVID-19 occurences drop approx. 14 days after the first dose with no measurable effect of the second dose after 28 days. This pattern is exactly the same for both vaccines.
[1] https://www.fda.gov/media/144434/download
PS: My parent comment was downvoted more often than any other HN comment I ever made, but nobody cared to elaborate why my line of thinking is unethical or why i am reading the data wrong. That's slightly disappointing.