I really think that a federated system for scientific publication could sense. Each "journal" is a separate instance, federated with all others. A department, a school, a research institute, whatever could host it. You manage submission, editing, peer review, etc. through the system.