In the current journal system, editor-votes have absolute veto power, and reviewer votes have large upvote power. This has led to much progress, but it is unclear that it is the best system possible.
Then there is the whole segmentation question, which is a big tradeoff between S/N and coverage of the space. An example: Physical Review A, B, C, D & E used to be one journal -- Physical Review. Phys Rev had a green binder, and people would keep these dead-tree objects on shelves for reference. Well, the publication volume kept growing: Some wag calculated that rate of growth of the volumes on the shelf would cause their edge to exceed the speed of light. (He also pointed out that no information would be exchanged, so there was no violation of physics.)
I agree. Right now the reviewers (= designated experts) have all the votes.
As I think about it, the reviewers achieved that status because they wrote articles that got upvotes from previous reviewers. The first reviewers were historical and they set the standards.
You could end up with a situation where the reviewers were poorly chosen and so upvote complete bullshit (http://en.wikipedia.org/wiki/Sokal_affair), but fields of science have a reproducibility clause that provides good amounts of self-correction.
Hmmm. I didn't think your boycott post would lead to an examination of the way science and engineering is published, but the time might well be ripe. That said: My recommendation is to narrow the scope and focus on the boycott. It's enough to keep your plate full for a while.
This would give more sway to old, established academics, but that's already the status quo (and roughly what you are trying to achieve anyway).
We don't need journals for that. They could be replaced by services like http://f1000.com/