I disagree that the logistical costs have anything to do with why libraries would not remove a book simply because the author didn't like it or want people to r

I disagree that the logistical costs have anything to do with why libraries would not remove a book simply because the author didn't like it or want people to read it anymore. If the work is still relevant/important/frequently accessed enough to warrant space on the shelves, then it is a thing of value and the library would have no reason to throw it away just because someone else asked. Moreover, libraries generally see themselves as having a historical and preservation mission that would not support retracting books. Likewise in the case of Wikipedia, if your contribution meet the standards to be included in the first place and has not yet been replaced by other edits, then presumably the article is better with it than without. Why would they agree to go back to a worse version of the article? Although digital, I believe a repository of academic publications is far more similar to a physical library than it is to a social networking site. Social networking sites simply need people to look at their ads to make money. Whether or not people can delete tweets is not that big of a deal one way or the other to their business purpose. The purpose of arxiv is to permanently archive open-access versions of academic papers for academic study as well as preserving the history of the field/academic progress, even if the papers themselves are no longer current or accurate. In any case, the site is clear that contributing a work to their library involves the grant of a perpetual, non-exclusive license to publish and redistribute it.