The ``Peer Review'' system has been used in its current form for roughly one century. It has recently started to show its limits, from systemic failure such as the lack of repeatibility of scientific publication in various fields (Economics, Psychology, etc.), to isolated personal failures such as scientific frauds, and more generally to dangerous trends in the evaluation of academic careers, departments and universities, forming long term incentives potentially harmful to Human ingenuity and ability to progress scientifically.
We postulate that such failures are a symptom of the growth of Academia, and that the pressure on the member of academia is stretching the current implementation of the peer review system beyond its limits. Accordingly, we list modern technologies, currently unused, which could be used for a new implementation of the peer-review process, we discuss the potential cost and advantages of each separately, and which ones and how they can be combined to form a soft transition to a new healthier peer review system.
Among those techniques, we focus in particular on the concept of "Boot Strapping Databases", where all agents are taking turn at being author, referee and reader, and where the quality of one's good work as a referee at each tier is evaluated, and used as a condition to be allowed to participate as an author and as a reader. Such systems have been tested to correct student essays in "Calibrated Peer Review", and to incentivate the collaborative development of a database of teaching material at the University of Waterloo. We will open the discussion about how to introduce to manage the peer-reviewing of academic research via services such as the publishers Arxiv and the Cryptology ePrint Archive (http://eprint.iacr.org/).