The idea of decetralized publishing (or better to say decoupled from greedy publishing houses like Springer) has been circulating in the science community for a while. arXiv and bioRxiv have solved many problems, but not all. There have been good attempts to decommercialize the scientific publishing business, for example, the eLife journal. But also eLife is slowly drifting to the conventional publishing scheme and has changed its publishing strategy and fees several times already.
The major problem here is how to verify a scientific publication. I very much like the analogy to #bitcoin mining where a network of clients (scientists) would submit their manuscripts (transactions) to publishing agencies (relays), with the latter initiating a decentralized review process (mining) and finally appending it to the global knowledge database (blockchain). But to me there seem to be two large obstacles:
(1) Who is going to validate those? In the bitcoin universe, you as a user don't need a specific expertise to validate bitcoin transactions or to solve cryptographic hashes to have a chance to win the block reward. Some computer knowledge is of course needed, but the validation/mining is done by soulless machines. And it's a very objective thing — no misunderstandings, no ambiguity. The question is then: How can we formulate a mathematical problem such that its solution corresponds to verifying a scientific claim? It’s absolutely nontrivial.
(2) How do we make this new publishing system self-sustained? Again, in the bitcoin universe, this is achieved by block rewards to the one solving the cryptographic puzzle correctly. Would the same scheme work in science? How do we ensure the system is not misused to mine rewards insead of validating scientific arguments?
All in all, the question that you raised has immense importance in my opinion. Despite (1) and (2), trying to approach this ideal is definitely worth working on it.