Dont (1) and (2) just result in inflation. Whoever is getting the new reward, it is diluting the value of all other coins that are fungible with that reward.
If (the value of the average coin) = (value destroyed in creating the average coin), then this is inflationary.
Why don't they print out a bunch more USD and give us $2000 stimulus every month?
(1) Depends on a hard coded list of kings who get free money to control the rest of us with. It is incompatible with a decentralized and egalitarian system. The hard coded kings are a central failure point.
(2) Seems to be based on the assumption that running a validator is approximately free, and that it is impossible to sybil the system by launching more validators. Which means the current list of validators is a hard coded list of kings, same as (1).
(3) Vitalik's link calculating the cost of attacking PoW is based on the incorrect assumption that ASICS are a perfectly liquid market, and a person can rent 51% of hashpower by the hour. If you instead consider that they need to purchase the ASICS in an illiquid market, you get very different numbers. In an illiquid market, the more ASICS you buy, the more expensive the ASICS get. The cost of the attack increases parabolically as it gets bigger.
His calculation of cost of attacking PoS is inaccurate, because the only attack he considers is people who literally purchase the coins they are staking. When in reality it is always possible to rent control over PoS deposits, by paying the PoS validators to make a second larger security deposit to enforce that they participate in your coordinated attack, and having a smart contract specifying limits of how the rented coins can be used, to prevent them from being slashed. PoS deposits are made up of a liquid market of coins, so the cost only increases linearly as you rent more coins this way.