Ethereum's Casper shit is more of the same proof-of-stake (nothing-at-stake or centralization by economic weight, e.g. DPoS) nonsense. The betting stuff enables what Vitalik refers to as "dark uncles" or "dunkles", which Vitalik incorrectly thinks will solve the nothing-at-stake problem. Also Casper has the problem that all deterministic finality PoS and Byzatine agreement systems have, which is a 33% liveness threshold which if that many validators balk or stop processing, then the chain can't move forward without a hard fork.
Also casper does not assume rational attackers, so if a 51% attack happens in POS, at curent prices, the attackers would have to burn 100-500mil $. In which case they would rather do it now than later because it would cost them 10 times less. With ~30$ mil you can even attack bitcoin.
I think it better to increase the block size and let the miners earn more so that they will better protect the network.