How does Casper’s sharding work? Tendermint is a building block for introducing parallelism. Maybe it’s compatible.

One thing I never could figure out about Casper… it seems that Casper could be vulnerable to a “tug of war” attack with less than 1/3 of Byzantine validators. Imagine a malicious subset of nodes that have infiltrated the network, that can sense when the bets are about to tip towards one or the other block. These malicious nodes can introduce timely bets that keep the balance in the middle, with honest nodes presumably betting with increasingly more “certainty”, yet never being able to decide towards one or the other because the attackers maintain the balance. In short, it seems that a small number of attackers can economically halt consensus. (In Tendermint, temporary gridlocks are resolved by moving onto the next round)

Looking forward to future results from Casper. I hope to see some proofs!