I can see the use of blockchains and networks like ethereum, but it seems like the majority of use cases could run on a private network. I still don't see the appeal of a single public network. Why not a dozen or a hundred? That's why the value or Ether seems so useless to me. There was a backlog of transactions lasting hours yesterday and this is supposedly just the beginning.
So far, there hasn't been any killer use case for ethereum yet. Everything it does, can be done more efficiently outside of the ethereum network. Some say that its "decentralized" on ethereum though, but this isn't true, because a lot of the inputs to ethereum contracts originate from a centralized source. For instance, if you want to make a smart contract that sends you 100 eth based on the outcome of a basketball game, the centralized source where you get the trigger on who won the game, is the centralized source your contract has to trust. It doesn't really matter whether the contract was executed on the ethereum network, it is still vulnerable to centralization. Whats the difference between that and just making it work on bitcoin by having a service sign a multisig transaction sending you btc if the basketball team wins? The only way something can truly be decentralized, is if all the actions in the smart contract are primitive actions based within the network itself. For instance, if y address receive 1 eth, send x address 2 eth. Neither action requires trust outside the ethereum network.
I think the future may well be lots of blockchains talking to each other via something like Polkadot (http://polkadot.io).
There may also be use cases for private blockchains, and there are startups pursuing those, such as http://chain.com (though I'm not sure why regular databases wouldn't work for most of those use cases), but I think there is also a need for a global decentralised network (or networks) for other use cases - global transactions/contracts, decentralised companies, probably other use cases yet to emerge.
Current scaling/latency issues with blockchains - there is ongoing work to solve those via off-chain networks, sharding etc - it's still early days, it will take some time to develop.
I think one use-case that makes some sense is Mysterium (https://mysterium.network). VPNs actually have material benefit from decentralization, and the ability to actually cover costs in operating exit nodes is somewhat of a game changer relative to something like TOR (I do realize they have different security profiles, though).
Data storage is also a moderately interesting application. Decentralizing data persistence, and being able to take advantage of the enormous amount of excess hard drive space that exists in the world is an interesting concept. Things in this category are Sia, Storj, and MaidSafe. While i'm not sure the economics and logistics will actually work out, I do think there is something there.
