A text which is not a draft, but expected to be updated - upload to arXiv?

Question

I've been working on a text which you could call a short monograph. Without going into details (which might lead to another question) - I'm considering posting/uploading/"publishing" it on arXiv.

It's not a draft - I've gone over it several times and it seems in pretty good share; but, at the same time, it is not finally done: I expect that, with feedback from others, I will need to make several additions, rephrasings, and possibly even slight correction of certain details regarding existing work.

Now, I would have liked to have received this feedback before ever publishing, but this is unlikely to happen; and I've been "sitting" on this for quite a long time already.

Is it legitimate, and is it reasonable, to upload an initial version on arXiv, expecting that an update will be necessary?

Answer 1

Yes, it’s legitimate. The fact that arXiv has a versioning system tells you everything you need to know: the ability to post updated and improved versions of your manuscript is an intentional feature of arXiv. So your question really amounts to “is it reasonable to use this feature?” - to which the answer is, yes, of course.

Now, keep in mind that the versioning feature can be, and at times has been, abused by people who post way too many versions of their paper with many confusing, incremental updates, usually in an attempt to bring an originally very low quality paper up to an acceptable level. This type of behavior is certainly looked down upon as a form of “noise pollution” and may be seen as a signal of low quality content, so if you think you’ll be posting more than, say, three versions (including the original submission), I’d suggest rethinking your approach. Also remember that the older versions of the paper remain accessible indefinitely. If you care about your reputation, you’d want to make sure that even the original version of the paper is of a high enough quality that you can accept being associated with it far into the future. With that said, your desire to share the work even in non-final form, whether because you think it may be useful to others and/or because you need some feedback to put the finishing touches on it, are quite understandable and acceptable, and shared by many; that is why the versioning system exists. So as I said, it’s totally legitimate.

Answer 2

My normal workflow is to:

1. Post the first version after all coauthors are happy
2. Collect any feedback, and if there are needed edits incorporate them into a second ArXiv post which is exactly the version submitted to a journal (otherwise you run the risk that the referee looks at the wrong version).
3. After any referee reports and responses make a third ArXiv version which is the author-final version.
4. No more than once-a-year update with any post-publication typo fixes.

For any given paper steps 2, 3, or 4 might not be needed, so this is more of a maximum than a required. But I think if you’re not doing more posts than what I’ve outlined you’re certainly ok.

Answer 3

Legitimate but not recommended.

1. My experience is that feedbacks from persons unknown are very rare. Think of the reverse: do you yourself send feedbacks to random authors in your field who submit on arXiv?
2. There is a reputational cost to submitting work obviously not ready for publication. Do you yourself read a lot of submissions of this type or do you stop once you realized this is not finished work?
3. Constantly updating an arXiv submission tends to bad at the time of journal submission.

If you are a few days away from journal submission then nothing is lost by privately circulating advanced copies of your work or posting on arXiv: you might get the occasional feedback especially if there is a glaring omission in the text or in the bibliography, but arXiv is not a peer-reviewing service, and thinking of this as such may lead to acute disappointments.