And here lies the innovation of proof of work schemes, preventing one party from casting arbitrarily many votes by making votes expensive. This also is no innovation of Bitcoin, Bitcoin just put all the pieces together. But if you don't need pseudonymity and can know the involved parties there is no need for proof of work, i.e. you can easily replace mining with an identity verification process. And you get a system that is in some sense better because you remove the possibility to obtain more voting power by buying more hardware.
It doesn't work without that centralized arbiter because then you'd end up with situations where each party has their own separate list of other parties that they consider vetted, and this falls out of sync, and the blockchain completely falls apart as there is no longer consensus on whether any given block is valid or not.
I will agree that the banks can probably come up with a centralized arbiter that works for them, and thus they don't need blockchain technology, but that isn't a counter to my assertion that the blockchain's key innovation is that it allows decentralization in a way that was never before possible.
You could come up with something like the process when verifying a domain for a certificate, creating a special DNS entry or serving some specific content at a specific address. This doesn't really work because nobody stops you from registering arbitrarily many domains, it is just meant as an example that the process does not necessarily require a central authority.
But there is a hard problem here, who is a party and gets one vote? Could a bank just create ten shell companies and then obtain eleven votes? Bitcoin kind of sidesteps this problem by implicitly declaring you a separate party if you possess some mining hardware and weighs your voting power by the hash rate of your hardware.
So you are definitely right, Bitcoin achieves a degree of distribution that is probably really hard to do without a trusted central authority that has the power to decide who is a separate party and who is not.
If the number of parties is small you could maybe still do it in a distributed way with some manual work. If a new bank wants to join they broadcast a request, use something like the mentioned DNS record to prove that the request is genuine and then all existing parties have to manually decide whether to accept or reject the new party and cast a corresponding vote. But this obviously doesn't scale well.
There's a chicken-egg problem there that requires a central arbiter to bootstrap, but the central arbiter then immediately becomes powerless.
An immutable data store of data would be a new type of storage system the world has never seen before. Up until now, things stored in a database have always had a decent amount of probability they wouldn't stay there forever.
It would be nice to be able to pay a decent amount of stored value to keep some truths as universal truths, where the probability of losing those truths approaches zero. I think there are truths that are worth keeping around and ones that are not.
Figuring that bit out is the next hard thing.
That sounds like a very anonymous way of distributing messages. Bitcoin's anonymity is more like having an unregistered bank account whereas BB (before bitcoin) all bank accounts were registered in one way or another. You don't even need an 'account' to accept a payment, u just need a magic number, though there is arguably not a difference. I feel like you're talking about network layer anonymity and everyone else is trying to talk more basic protocol.
Real breakthrough of Nakamoto is devising a scheme where whole network is protected against Sybil attack. I.e. there is no need to don't trust or register parties who participate in the network.
All elements of Bitcoin, such as proof of work, crypto ledger, merkle trees, etc were already known. Nakamoto combined them in a way which solves an unsolved problem.