Suggested average-case improvements to reduce capital costs of being a Casper validator

You definitely could offer different buckets and offer more returns for the larger buckets, though that’s not especially useful because the security of the system depends on the withdrawal time of the lowest 1/3 of the validators, so a few validators having longer lockups than everyone else won’t actually help too much.

The objective of my proposal (2) is to try to target having maximally low withdrawal time except when it seems like there might be an attack.