The author's central argument has been formally proven: https://eprint.iacr.org/2016/919.pdf. PoS protocols cannot offer a client with no a priori knowledge of the history of transactions a way to distinguish between an honest validator set and a corrupt validator set. There's no way to compare chain quality in-band.
> There's no way to compare chain quality in-band.
Just like with PoW where the longest chain and greatest hash power is the "honest" chain (even that is arguable), PoS functions similarly where the greatest staking power is the "honest" chain.
