Algorithms/protocols to solve the consensus problem in the presence of arbitrary "Byzantine" faults have been around for a long time (since the 1980s, at least e.g. [1]), and they keep getting better. Make the network closed-access (i.e. where the owners of all nodes are known/approved), and voila, you've also got resistance to Sybil attacks.
The end result is something with blockchain characteristics, but at lower computational cost than Proof-of-Work. Maybe that's not something everyone wants, but banks do.
The article explains why this is not true: back-end financial technology is boring and banks weren't willing to spend enough money or risk on upgrading them until overwhelming blockchain blockchain blockchain hype pushed them over the edge.
* Better BFT algorithms (e.g. lower network load, more reliable in practice)
* Faster computers
* Better computer networks
* Cheaper storage
* Better software development and deployment practices (faster, more reliable, more secure)
* and yes, the recent interest in BFT spurred by Bitcoin.
For example, the original algorithms had network load that was exponential in the number of nodes, O(2^N). Some newer non-POW algorithms are O(N*polylog(N)).
The end result is something with blockchain characteristics, but at lower computational cost than Proof-of-Work. Maybe that's not something everyone wants, but banks do.
[1] http://research.microsoft.com/en-us/um/people/lamport/pubs/p...