> potentially the costs for publishing are [zero]...
Potentially the costs of any product or service one simply renounces are zero.
But we don't want to renounced peer-reviewed journal articles.
And the costs (to the journal) of providing peer-reviewed journal articles
are not just the costs of dissemination and storage. They include the
costs of implementing peer-review (and have been discussed in the annals
of the American Scientist Open Access Forum since 1999!). See references
below.
> services like Arxiv, SSRN and RePEc are free to authors and readers:
They are merely means of dissemination. Peer-reviewed research journal publishing
is not and never has been merely a dissemination service.
(A Postgutenberg journal may in fact turn out to be a peer-review service-provider
and certifier of its outcome.)