A famous mathematician Gowers wrote that it is needed “a positive strategy of actually setting up a new system might work rather better” for publishing math articles.
In this post I suggest a way for the community to improve quality of self-published (or self-archived as scientists say) articles.
I suggest to add the following feature to arXiv:
Anyone should be able to submit patches to arXiv. There should be choice for patch submitter whether he creates a patch in unified diff format himself (provided he has necessary technical skills) or just provides his version of LaTeX source for the article and arXiv should create a patch automatically.
It is important to be able to manage patches not just new suggested versions of an article, because there may be several patches for the same article.
Hm, well we need some sophisticated conflict resolutions system (for patches which may interfere with each other).
We should think about usability of this system. For example new PDF files corresponding to every patch should be built automatically for the article author to be able to view the exact changes, before he approves or declines them.
Note that the patches may be both scientific improvements and improvements of formatting. The patch submitter should be able to select one of these two categories (scientific improvements and improvements of formatting) for his patch. Patches should be also accompanied with a short text (in PDF format?) describing the exact changes.