> looks like duplication. why not arxiv? could lead to multiple submission pipelines

Arxiv is for preprints, this is full peer-reviewed publishing. It replaces academic journals, not preprint repositories like arxiv.

> as I understand researchers still pay to publish, no?

They mention Diamond Open Access, so probably not

> initiative could lead to centralization of publishing power

Publishing power is already incredibly centralized in hands of a few oligopolistic publishers

> the problem is not just in access to papers (what about connection to business and real applications)

Not sure what this point means and how it's relevant to this initiative

> the article is published by CERN and is promotion (vague on details, buzzword-heavy)

Idk, seems pretty specific to me