3

Did anyone produce updates on existing published papers later on?

Or is it always that when it's submitted, then "that's the way it is"?

5
  • 2
    what are you trying to achieve? Sep 6, 2021 at 15:44
  • @aaaaasaysreinstateMonica I've thought that having updated journals might allow "results users" to dig into results of some paper earlier on, while the other standards of the paper might still take longer to complete. Thus one could push a "results, but unfinished quality check" paper for "early adopters" while continuing to hone the more aesthetic parts of the paper afterwards. After all, errors such as those related to grammar etc. are quite minor compared to withholding results, because the grammar etc. is not flawless.
    – mavavilj
    Sep 7, 2021 at 5:26
  • 2
    @mavavilj What you've described is a typical use case for preprint servers.
    – jakebeal
    Sep 7, 2021 at 9:12
  • @jakebeal Well, preprints omit peer review too, his proposal seems to be more like "arxiv + peer review now, make it look nice later". So, you can trust these results, it just doesn't look polished yet. I am unaware of anything systematically filling that niche; but I don't think this niche is relevant when the peer review will take a while anyway (and it could even delay the process if images or wording is hard to understand). Also, once you publish you tend to not really bother with all the improvements you could have made to that paper - you focus on your next one that improves results too. Sep 7, 2021 at 11:12
  • 2
    @ZizyArcher What you're describing is called a "postprint", and they are already made available by some journals, e.g., as a category of "just accepted manuscripts". These are manuscripts that have gone through peer review, but not yet gone through the copyediting and formatting process.
    – jakebeal
    Sep 7, 2021 at 11:23

4 Answers 4

8

Actual updates are unlikely. Subsequent work is frequent and valued. The problem is that an "update" assuming content is changed and not just form, could make citations made in the interim invalid.

Some publishers will publish notes or errata on papers found to be flawed.

Books get "corrected" editions published, but it isn't a general practice for papers. Even for online publication, changing the text makes some subsequent work (quoting) obsolete. It requires work that leads to as many problems as it solves, even for typos.


Note, however, that the paper published is almost never the same as the paper submitted. It goes through a review and editing process that can result in quite a lot of changes. But once published it is pretty much cast.

9
  • I find it odd that actual updates are not done, because the internet and computers specifically allow this. Compared to when it had to be printed and each print cost a lot.
    – mavavilj
    Sep 7, 2021 at 5:03
  • 2
    @mavavilj Optimistically, the published version is still the version of record, and freezing that keeps us honest, allows others to build on the work, and means citing a paper for a specific point won't be invalidated by that point changing. Pessimistically, online publishing is paper publishing without the printing, and we're bound by old habits. Preprint servers attempt to fill the gap between these views
    – Chris H
    Sep 7, 2021 at 10:30
  • @mavavilj It's not that it's technically unfeasible, it's that making post-publication edits difficult adds robustness to literature: the authors must be very sure before they publish; if I cite a paper, everybody who follows the cite reads the same paper; I can be sure the version I'm reading is the version that passed peer-review. If there is an actual error in a publication, a correction or retraction can be published, but this is unusual and generally doesn't look good.
    – Cai
    Sep 7, 2021 at 14:40
  • 1
    @Cai Why cannot publications behave like updates to code? Someone pushes an update -> people using that project will get informed that there have been changes. They don't need to update, but at least there's a mark that "future revisions differ". This does not invalidate current projects, because when they used those references, they did draw usefulness from them. I.e. existing publications retain their functionality in some sense.
    – mavavilj
    Sep 7, 2021 at 14:45
  • 1
    @mavavilj Then I guess the answer to the question comes down to: conventionally academics consider the quality of research to be important. And part of that quality means that when you read a paper, you have some trust that the authors have honestly presented the best version of the work in an effort to contribute to collective knowledge, and that through peer review it has been vetted by qualified experts. Part of this trust comes from the fact that edits can't be made. Not that there's nothing to gain by making every publication a living document, but it wouldn't be a strict improvement.
    – Cai
    Sep 7, 2021 at 15:01
8

While @Buffy's answer holds true in most cases, it is worth noting that a number of disciplines have a system of two-stage publications that are explicitly designed around publication of an updated version.

For example, in many computer science fields, papers accepted in a conference are encouraged to later send an extended version to a journal. Both enter the scientific record, and the journal version is intended to be a more authoritative and final version, citing and superseding the original, and typically with at least 30% more material than was in the original.

I have seen this beginning to appear in some other fields as well, e.g., with extended abstracts in a conference being invited to publish a full paper in a journal.

5

In large areas of physics and mathematics, we are reading papers on arXiv and paying little attention to their published versions. (Assuming there is a published version, which is not always the case.) When there is a correction or improvement to be made post-publication, it is often done on arXiv only, as this is much easier than sending an erratum to the journal, which nobody would notice anyway.

Example: an arXiv preprint and the published version. In this case the title and format differ. And the preprint has corrections post-publication.

Example: this article has 9 versions on arXiv including 4 prior to publication in a journal, and 5 after publication. Between the first version post-publication (v5) and the last version (v9), 4 years have passed, and the article has grown from 36 to 55 pages.

2
  • 1
    Agreed this is an important point, though worth mentioning that usage/views of this vary between and within fields. In the fields I know (logic, pure maths), minor revisions like your first example are common (especially for errata), but extensive post-publication revisions like your second linked example are rare, and frowned upon by at least some people since they make clear + consistent citation much harder.
    – PLL
    Sep 7, 2021 at 11:31
  • @PLL: Good point. As far as I know, extensive post-publication revisions are rare in any field, and reserved to some types of articles. After all, the average research paper is quickly superseded or forgotten, and not worth revising. But the original question was about "anyone", not "most authors". Sep 7, 2021 at 12:13
3

There's a journal series that does exactly this - the Living Reviews series published by Springer. An example is Living Reviews in Relativity. From their description:

Living Reviews is unique in maintaining a suite of high-quality reviews, which are kept up-to-date by the authors. This is the meaning of the word "living" in the journal's title.

So one can expect articles published in Living Reviews in Relativity to be up-to-date. Most other papers, however, will not have been updated and so can be old or even completely superseded by later data.

3
  • These are reviews, of course, not the published papers themselves.
    – Buffy
    Sep 7, 2021 at 0:17
  • @Buffy: Really? If I’m understanding right, this series publishes review articles in the sense of full-length articles surveying research topics, not just reviews in the sense of short reviews of articles.
    – PLL
    Sep 7, 2021 at 10:57
  • 1
    @PLL, they are not the original research, so their stability/finality is much less of an issue.
    – Buffy
    Sep 7, 2021 at 11:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .