I think blockchain will end up like PageRank. Both are amazing algorithms with very narrow use cases. And maybe a few companies can generate enormous value with blockchain like Google did with PageRank, but that will be it. That's my prediction.
It just needs a concept that has transitive properties in order to shine. For web pages, the transitive part is that a high quality page is more likely to mention another high quality page. It uses the indifference one might have about a lower quality page, it uses the author's implicit contract with the reader that he should be showing (and linking to) high quality content. It builds on top of man.millionYears of painstaking content filtering.
In a graph, you can run it about any property that has that transitivity. There is a paper running it on citations in physics papers [1], and it can identify some Nobel Prize winners, even if the paper has a low number of citations; the paper just has the important ones.
I deeply believe that corporations could greatly benefit from continually asking their employees "who in your coworkers had a positive impact on your work this week?", and running PageRank on top of it. You could have natural leaders emerge out of that.