I think desire to have everything free is a fantasy of the open source community. Not everything can be run like open source software project. Sometimes people need to get paid.
I think you have a fundamental misunderstanding of how academic journals are structured and managed. The editing, typesetting, handling of communications and such are all done by the academics themselves. And those academics are not paid by the journals.
In short the publishing houses do sod all, and get paid a lot of money for doing sod all.
The editors, most of whom are minimally paid and do it for prestige. I.e., the same people who currently do it.
...who's gonna edit,
Unpaid referees/editors, the same people who currently do it.
...typeset...
Authors, the same people who currently do it:
\usepackage{freejournal}
Readers, the same people who currently do it. Or are you imagining that there is some pressing need for physical journals to be printed and delivered to libraries?
The only thing Elsevier does is archival and distribution. Editing and typesetting is done by the authors themselves, who provide camera-ready copies. I do not believe the publishers ever get involved in managing mail. The reviewing of paper is done for free by academics. It is coordinated by an academic on the editoral board of the paper. Even these editors mostly work for free or for some small compensation.
While Elsevier do provide some service (archival and distribution), their service is in absolutely no relation to the cost. The difficult parts of the business are already done by unpaid volunteers.