Many research papers include mistakes, ranging from typos to incorrect proofs. It is my understanding that currently major mistakes are generally (though not always) communicated to the community by publishing a companion 'note about the paper'.
However, even minor mistakes, or hidden assumptions, can be misleading to the non-expert (in particular when trying to implement a published algorithm, or trying to understand quickly a complexity bound). From my own experience I have a feeling that sharing this kind of information might spare a lot of time when trying to understand a paper (and this might improve the 'quality' of the available literature).
As far as I can see, tcs and cs currently may not really be the place for, e.g., correcting typos in equations or detailing information about implicit assumptions. Is there any website that supports this sharing of information/collaborative edition of papers a posteriori? Or would that kind of site be mostly useless, or raise issues (ex: avoiding putting shame, dealing with 'ownership' issues...)?