If you published a paper book you couldn't make libraries give it back to you, even if you gave it to them for free. On other projects like Wikipedia, there is no special privilege to remove your own contributions. Even on HN you only have a few minutes to delete comments. Why should this be any different?

None of these seem to have any logical reason to prevent deletion or undo. Perhaps the most justified among them is the paper book in a library because it needs physical effort or logistics to remove and return it.

That's not true for digital data, though. Reddit/Twitter/Facebook have no trouble letting users delete posts and comments though their data models involve inter-dependencies between data. Why should a repository for papers, that too with no inter-dependencies, be any different?

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in the case of Wikipedia, if your contribution meet the standards to be included in the first place and has not yet been replaced by other edits, then presumably the article is better with it than without.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for academic study as well as preserving the history of the field/academic progress, even if the papers themselves are no longer current or accurate.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.

> ArXiv won't let authors remove their paper from the website

This is an essential feature of arXiv to maintain academic integrity and encourage serious academic submissions.

If you realize that your paper contains an error, then you can mark the paper was "withdrawn". People have limited time so if you do this no one read the paper (everyone makes mistakes sometimes).

As long as you do not make a habit of uploading and withdrawing papers, then it will not matter.

I am not a real scientist, but I really want this kind of content to be eternal, especially when somebody has linked your work in his one.

One thing is to edit some typos and another thing is causing the situation which JS community has observed when Lefpad's author decided to remove his work because of some emotional reasons, causing some obstructions to all the JS community.

Do you really need to remove, or would it suffice to withdraw it?

https://arxiv.org/help/withdraw

I don't know much about academia, but I'd expect journals would force authors to remove preprints given the opportunity. Better for this to be impossible, imo.

> Is this really "open" if the author has no right to remove their work for reasons other than what's mentioned above?

Yes. In fact, if you could arbitrarily require them to stop hosting it, then it wouldn't be open. As a point of comparison, consider that all FOSS licenses have to be irrevocable.

archive, noun, a place in which public records or historical materials (such as documents) are preserved

arXiv requires, at minimum, an irrevocable, perpetual, non-exclusive license for arXiv.org to distribute the article (https://arxiv.org/licenses/nonexclusive-distrib/1.0/license....). That is why their response largely revolves around the possibility that the submitter didn't have the right to grant the work under that license in the first place: if it was granted under that (or a more open, CC license), then there's no copyright-based legal reason why they would need to remove the work at the request of the submitter.

> You could possibly get a result by hiring a lawyer

The submitter agreed to indemnify ArXiv. So they have to hire ArXiv's lawyer too.