A recent yellow paper by Vitalik Buterin simplifies and applies a 99 percent Byzantine fault tolerant algorithm for blockchains. ETHNews explains it in language you can understand (minus the math).
Last week, Vitalik Buterin published[1] "A Guide to 99% Fault Tolerant Consensus," wherein he lays out a simplified version of Leslie Lamport's 1982 algorithm to address the Byzantine General's Problem[2].
The algorithm he discusses is not necessarily a replacement for proof of work or proof of stake (though, with some modification it could be, if you're not set on decentralization). Rather, it is primarily meant as a method for observers to monitor block creation in real time for 51 percent attacks, and to organize soft forks to overwrite any such attacks. Under this model, if a 51 percent attack were to occur, the attack will be detected so long as one percent of nodes are honest.
While it is unlikely for a 51 percent attack to take place on Ethereum (or other truly decentralized networks), the algorithm could prove useful under many situations. Even with Ethereum, there's always the possibility of increased centralization of mining pools, and this could both safeguard against that threat and increase participant confidence, potentially bringing more developers, companies, and consumers to the blockchain. Moreover, Ethereum's eventual switch to proof of stake will decrease fault tolerance[3] to 30 to 33 percent, which means those consensus mechanisms can only function if two-thirds of participating nodes are honest. In other words, it will become increasingly beneficial to have additional defenses against attacks.
Further, the algorithm is not specific to Ethereum and could be used by any project, including more centralized blockchains that may come under consumer scrutiny.
