A system could arise where things start small, and trust and reputation is built up over many transactions, similar to how money works post gold-standard. Then there's the idea of private blockchains, where actors already kinda trust each other anyway (transactions between banks) which may apply at some point. Gains would be mostly efficiency.

However when I tried some Ethereum development my impression was that most cryptocurrency people I met just hope that the value of their bitcoins goes up, or that they can sell related products or services. Many things seem to be financed by cryptocoin-money (not necessarily bad in itself) to keep the hype going, but it doesn't feel like a movement that wants to change things.

It's not like with open source where Stallman, Linus and many many other dreamt a dream but also built a lot of really high quality and important software. I don't see that kind of movement going on with Ethereum.

Courts run with the predictability of law/codes...it's one of the reason 90% of cases settle without trial.

If an arbitration clause isn't held up, it's not because of unpredictability of the court it's because the clause runs afoul of the actual law/code the court applies which is predictable. Moreover, smart contracts are not going to permit parties to circumvent law, in other words a smart contract arbitration clause would not be enforceable in a jurisdiction they are not otherwise enforceable.

And with all the reliability and security of code. See, for example, the DAO fiasco.

Interestingly this can be considered both an argument against and in favour of smart contracts.