“A crowd-sourced long-term reputation system? “
Once you construct a system of rules the system will be subverted and gamed. See any internet ranking system ever devised for examples.
“Check out http://hypothes.is for possible solution to the problem you highlighted. It IS being worked on and it’d work for all content, not only scientific papers …”
This looks a prime example of a system just waiting for the aforementioned subversion.
Do you think the future of science should be thinkprogress and WUWT directing people to vote up their favourite papers? Talkorigins and creation wiki voting down the credibility of each other’s positions?
Until software is able to independently judge the quality of work it will simply act as a tool to publish the criticisms of others. The wider the net you cast for those criticisms the lower their average quality will be.
Computers can allow hundreds of millions of people to share their views on the latest blockbuster movie. It can’t tell you which of those views are any good, which of them are relevant or which of them will appeal to you.