It is not just about where the papers are published. A system needs to be established that allows peer review to happen in a meaningful way, archives must be established and maintained, etc. Right now, journals are facilitating peer review and archiving papers, but that carries the cost of papers being locked behind paywalls with outrageously high prices that ensure that the general public has difficulty accessing that work. I would suggest a system in which university libraries act as archives, providing HTTP links for people who want a handful of papers or BitTorrent for people who want large batches and for distributing the archives to other libraries. Peer review would be facilitated in software, using some threshold signature scheme to assert that a paper passed review and using some secure computation protocol to protect the anonymity of authors; people would be invited to review papers by being given a share of the signing key.

My point is that we have a legal system in place now that protects and promotes the existence of a publishing system that is grossly out of date, and that we have a chance to do better (and to abolish the old system).