I personally like Yann LeCun and like this idea. However, the root problem in CS research community is not a bad reviewing/publishing system. From my point of view, it is the evaluation system of the so called "professor" and researcher. The performance of the research is evaluated by the publications (the number or the citation). So here is the consequence: only a part of good "professor"/researcher really care about his research, all others care about the publication instead of research. My personal experience, you can see everywhere the "professor" push his graduate students to publish the paper. They are engaging in making the relationships to make his paper easily be published and make a lot of citations without any meaning. The publication is just a medium to describe the research analysis, results, etc, the true contribution is the research itself. Another point I want to say to learn and teach the knowledge is a very important responsibility of a professor. However, many professors only want to get publications from his students. So my conclusion, if the evaluation does not change, people still will abuse the new good review/publish system to get more publications, instead of doing valuable research.
Another very funny thing is that until now, CS research community always publish paper with improved results, good performance. It is definitely not true and the experience from failure should be also a good material to share.