This is very thick, but potentially revolutionary work in terms of blockchains.
The Ethereum devs are working to replace mining (proof-of-work) for blockchain security with proof-of-stake in the form of betting on which blocks are included in the chain. The crazy thing here, is that the blocks that are included in the chain are those that most parties wager will be included.
Imagine a race with 3 horses (H1, H2, and H3) and 4 bettors, where there are sequential rounds of betting as the race is run, and where each bet made on a horse moves the horse forward. Say the first bets are: H1, H2, H2, and H3. As a result, after the first round of betting, H2 is in the lead. Another round of betting begins. As H2 is in the lead, there is a higher probability that H2 will win, however the payout on H2 has decreased. The next round of betting gives: H1, H2, H2, H2, and horse 2 wins.
As Buterin writes:
In short, the blockchain is secured by ETH stake holders (validators), taking bets on the evolution of the chain. It is costly to place bets because: 1) ETH is not free, 2) transactions cost gas (ETH), 3) there is a minimum bet. The chain is secure because it is exponentially more costly to create an alternate valid chain than to extend the current one.
Two big upshots of this, if they can pull it off are: 1) no mining, 2) very fast block times (~5 sec).