I do believe I might disagree, both philosophically and technically. GNUpedia and Nupedia both failed because of the formal review process, maybe you can learn something from the history of these projects?
By combining Mediawiki with a directory service, wiki view and page write access, and a formal review process that requires reviewers to have the article under review's RSS-feed on their bookmark bar in their browser, then I think you can get quite far - actually writing the articles and reviewing them.
The catch is formatting, though LaTeX and a lot of other things can be supported, and LibreOffice with wiki export should probably be investigated.
@Erkan Yilmaz I would love your feedback about this issue. My main reason to have such feature is for trying to make a book!
The book would be open for anybody to read and to "suggest" modifications (fix typos, or even comment about something wrong). But the author needs to approve them. It would be great also to have such suggestions visibles too until rejection/approval, so that 2 people don't loose time suggesting the same thing.
Perhaps doing a book colaboratively with an "editor" role "releasing" the "official" version.
I can't help thinking about something 'git' based. But, if MediaWiki has a plugin, it's better to follow the flow.
for #Mediawiki: I think these 2 #extensions could be for you:
a. Extension:FlaggedRevs (1) - more features, used in some #Wikipedia
b. Extension:Approved Revs (2) - simple, but effective I think
the git editing: I am currently doing also with #Openadvice #book (3)
works for me with someone adding teh pull request, but I guess for newcomers would be not so satisfying experience
Scenario:
1) I read an online book (fork) and find a typo.
2) I fix it (patch), and, in my version (branch) of the book, this should show instantly.
3) My fix should be available (pull request) to be "incorporated" (merged) into the main book (master branch).
I think this is the kind of work to a backend based on git. Git relies on hardlinks and it's very "economic" about harddisk space. And it's fast.
(bare) Git is too hard for "normal" people. It has to be a good system (good interface, and everythig) based on (at least like) git.
This merge problems can be circunvented with a well designed system. But, I think this kind of system doesn't exist yet. So let's keep an eye on the options. Thanks for your tips.