There are many services in this world which require a trusted intermediary between users. The fees these services charge are _not_ a function of compute cycles. They are, however, often a function of pricing power derived from competition (or lack thereof). Where decentralized computation is less expensive than the fees charged by incumbents in the trusted intermediary business, there is the opportunity for a new entrant to leverage the tech to profit on the spread.

Because it's something that doesn't otherwise exist. It's actual innovation, where else in the world can you run trustless global peer-to-peer applications? It will be a while before it's commercially viable, sure, but it's something unique.

One tangible example is how there is significant legal and accounting overhead for banking contracts, particularly credit default swaps and other risk-mitigating things. Some of these can be automated in Ethereum.

I very much agree with you. Its hard to believe that this will rise to the bitcoin trust level. If anything I believe in sidechain.