I discussed similar ideas with Peter Storm and Tamar Ziegler about a year ago. It is not that hard to build an abuse safe system - everything gets some score:
Any review has two scores:
- The review reputation, dependent on votes from other reviewers, and the reviewer reputation and the score it assigns to a paper.
- The reviewed paper score.
Reviewers have scores which are a function of the scores of reviews they write, and the papers they write.
Papers have scores, which are a function of the reputation and score of the reviews they get (yes, a paper may have several reviews).
This system assures that important papers will be reviewed very fast and very thoroughly. The downsides we saw are:
As Peter put it: it would be an excellent system for Vulcans, but mere humans may get hurt.