Following on from a request to start such a discussion. :)
I'd suggest a number of things - feel free to disagree!
1. Panel of reviewers - each piece to go to three reviewers, but not always the same ones. There's little point in sending me something which is based on examination of minutae of ancient Greek as I don't speak the lanugage, for example.
2. Clear reviewing guidelines, published widely. I see there's another post about those so I won't go on here - but if all reviewers know what those are, it'll make life a lot easier.
3. Clear time scales. ADHERE TO THEM. Reviewers who can't get stuff back in time are not used again. I'd suggest deadline --> publication being three months; that allows time for reviewing, commenting, resubmission, etc.
4. Mentoring for reviewers. Those who have done this in an academic context to introduce others to the sometimes arcane process... :} This from someone with one too many scars from academic reviewers who really never met Mr Manners... there are ways and ways of dealing with things. (Why yes, I am seeing this as a developmental opportunity for people. So sue me). :)
5. Reviewing to be seen for what it is - not a popularity contest but a process of seeing that content is reliable, well resourced, readable, etc.
Discuss, throw popcorn, ask questions... ;)