"a lot of America's ridiculously expensive healthcare is a result of inefficient use of resources, and dodgy incentives, and such things. "

Given Alcor is based in the US, and has to rent time and labor from the medical profession writ large, I'd suggest that overcoming those obstacles would indeed result in a fairly substantial savings for them. Alcor is small enough to suffer rather heavily from inefficient use of resources - they do maybe 10 vitrifications per year, but have to have the staff and supplies ready to go 24 hours a day, 365 days a year.

Also, the idea that doubling your size decreases costs by 10% is plausible, but dubious over 17 doublings.

I'd honestly conclude "insufficient information" here. The history of computers in the last few decades makes a mere tripling in productivity look excessively pessimistic. On the flip side of the coin, equipment and labor costs run in to hard limits. I can certainly see a 200% productivity increase being optimistic, but I can also see it being pessimistic. Since I don't know, it felt safest to just run the numbers as current research suggests, without adding new assumptions :)

That said, the best defense I can give for those numbers is that I was trying to be optimistic, because my starting assumption is that there's no way you're going to finance universal cryonics in today's political climate. I didn't want to throw myself up against an easy version of the problem :)