On: When are two proofs essentially the same? (2007)

[naniwaduni]

> If we don't treat almost all integers in an algorithm as fixed size then the analysis gets really messy and annoying in a way that has nothing to do with real computers.

I guess nobody does numerical computing or distributed systems with real computers anymore, huh.

[Dylan16807 → naniwaduni]

Wrong way around. Almost all the numbers in those systems are fixed-size. There's a few types of variable width payload, but all the little ancillary types are 64 bits or whatever.