On a post-publication peer review platform. Some notes after online conversations. These are quasi random thoughts, comments welcome. I personally feel that we have reached the stage where we should actively design the system and seek ways to implement it, large scale. Sorry the typos, misspells but it is late :)

First we need a paper repository, with full proof back-up and always on capacity.On top of that repository we will implement a collaborative reviewing layer. This really is first and foremost a platform enabling users to comment on papers in the repository. 

First about the comments: they should not be text only. We are talking about rich comments, with pictures, support for latex formula (à la mathjax, something really easy to use, maybe even offering a GUI of symbols for those who do not master latex). I believe comments should also include code, data. A simple way would be to allow users to upload material attached to their comments. Code is particularly important: it should be treated just like a paper. Reviewed, tested. Comments can be rated and the most highly rated comments will be promoted as reviews.There should be a robust system to clean offending comments, spams, robots. This could actually be managed via user authentication.

Second then is user authentication. First users should be allowed to comment anonymously or not. However, users should log with an identity that can be checked. This is where maybe societies can play a role, as certification bodies. Just like you log in using your google or Facebook login on many websites, you could log with your IEEE or AMS credentials. To post papers or to log comments, you should be accredited. As a side effect, your activity (posting papers but also comments, the values of your papers and the values of your comments) will be available. Some may think this is too much of a "big brother" thing. I think it would avoid alienating societies. 

Third, there should be a recommender system. The ability to receive a "promoted papers" based on what a user has read and commented on. The ability to enter keywords and be guided to relevant papers. And, why not, a social network layer among users: it is interesting to discover new content by way of others, if you feel they have similar taste, expertise.

The user experience is key. The system should be very simple to use, but rich enough. All users should perceive a sense of trust in the system.

Note that the backend contains a lot of valuable data. What users read, what they think of papers, what is trending, the amount of papers on a subject, relationships between subjects, true measures of multidisciplinary research, effects such as the half life of a paper, of a subject, how fast it becomes mainstream, community effects. But also user-related statistics that can help assessing a reviewer. Warning: with big data, come big responsabilties. Who will have access to this: in my opinion it should be everybody or nobody.

Note that the technologies necessary to implement and deploy at large scale everything described above ALREADY exist. They are used on a daily basis by Amazon, Google, FaceBook, Apple and many others. 
Shared publiclyView activity