Bitcoin "solves" the problems behind Byzantine fault tolerance, quorum systems, etc by completely ignoring the past 30 years of research on the topic, and introducing a very simple construction that bypasses all of the issues entirely by using the concept of proof of work. Don't get me wrong, Bitcoin is a brilliant idea, but it's the sort of brilliant idea which is actually more likely to come to you if you were NOT bogged down by existing research on how to do things. Satoshi's primary gift was not deep knowledge, it was a fresh perspective.

And I am not saying these things because Ethereum is a super-magic-brilliant protocol that involves deep knowledge about thirty years of development in multiple fields; it's not. It's also an ultimately fairly elementary idea that I still remain surprised that nobody tried to seriously push before me. In fact, at least two other groups got very close in 2012-2013, and a few weeks ago at the payments innovation conference in Boston I learned that apparently the concept sans blockchain was around in the 1990s; but for some reason they did not take the idea to its logical conclusion.

Also, our implementation of GHOST does not in any way compromise the concept of global state; blocks are required to specifically include uncle headers in order to benefit from them.