I think that role of for-profit publishers is largely to provide a ‘certificate’ that a paper meets their quality criteria. Granters and the ones hiring academics want to be able to look at a candidate’s publication record and quickly assess whether their work is high-quality without having to sit down and do a thorough analysis of their work - at least not until the final stages of the granting or hiring process. The review process is centralized in the sense that the journal’s editors are the final gate keepers, and journals build their reputation over time by the quality of work that they publish.

If we want to transform the publishing industry we need to come up with ways of simulating the peer-review process without a centralized authority. In principle this might appear easy, the peer reviewers are already doing it for free! But the the initial filtering the work, selecting the reviewers, editing the work, and gate keeping what gets published does actually provide a great value and is not so easy to emulate.