Should we post on arXiv only papers in publishable shape (or very close)? Question: Should we post on arXiv only papers in publishable shape (or very close)?   
This question should be distinguished from the following:
Should one post a paper on the arXiv if it is not intended to be published?
in the sense that a paper which has not a publishable "contents" can have a publishable "shape".
Moreover, a paper which is intended to be published can be not yet in a publishable shape.
Sometimes it happens that we start to write a paper, we develop some interesting new ideas, but then we do not continue the paper for several possible reasons:    


*

*change of the research subject   

*loss of motivation for this problem    

*too difficult   

*$\dots$  


Sometimes also we just want to share the current state of our work through a draft, even if all the proofs are not yet complete.
Whatever the reasons, the ideas contained in such drafts can be interesting and useful for the community, regardless of the state of advancement of the paper. So after this explanation:
Should we definitely not post such a paper on arXiv, even if its state is clearly specified at the beginning?  
 A: There could very occasionally be a case for making available plausible seeming methods for attempting to prove interesting results, which turn out not to work.
I was involved in one such situtation in which at least three different people, spent a long time trying in vain to do essentially the same thing, and they would all have preferred to have been told early on that their efforts were almost certainly doomed to failure. All of them were subsequently thought that the result in question was probably false.
A: The documentation of arXiv has this to say (my emphasis):

Inappropriate format. arXiv accepts only submissions in the form of an article that would be refereeable by a conventional publication venue. This excludes abstract-only submissions, submissions without references, book announcements or reviews, reports that do not contain original or substantive research, papers that contain inflammatory or fictitious content, papers that use highly dramatic and mis-representative titles/abstracts/introductions, or papers in need of significant review and revision. 

Source:  http://arxiv.org/help/moderation 
