Ideas for a new kind of scientific publication

At LinkedIn, Alex Zhavoronkov wrote that he thinks "it is time we move from regular publishing in peer-reviewed journals presenting individual disconnected papers to publishing thematic research papers that automatically update themselves with new data, dynamically integrate new results and link new papers on the theme".

A lot of people responded to his statement. IMHO, the most interesting suggestions are to make a Wikipedia for research articles and not to publish texts any longer but data and code that can be easily processed by computers. I agree that the concept of research articles as human-readable essays is somewhat outdated and impractical.