What are the downsides of the arXiv?

The title above was a recent (somewhat contentious) MathOverflow question by Igor Pak. While I still think this is a slightly problematic framing, it’s a discussion people seem to truly want to have, so I thought I might as well create a space for it.

Let me start discussion with the biggest concern (I don’t know if it is a downside) I have about the arXiv:
What is the arXiv’s policy on what can be posted and what can’t? I ask this as an honest question (with some trepidation about getting flaming from both sides). On MO we have an FAQ that tries to lay out as clearly as we can what sort of material should be put on MO and what should not. I cannot find an analogous statement from the arXiv about what they will accept and who they will accept it from. I doubt I disagree very much with what they do in practice, but the lack of an easily located statement of what that practice is actually disturbs me a bit. I apologize if such a document is publicly available somewhere on the arXiv website, but I maintain I shouldn’t have to hunt for it.

After reading the discussion below, I’m even more convinced that the idea of “downsides” is so tied up in one’s values and personal experiences that its impossible to come up with a list that makes sense to everyone. For example: one serious worry seems to be that if you put your papers on the arXiv, people will write follow-ups to them before you’ve had a chance to fully process your ideas. I’ve had this (sort-of) happen to me, though from a talk, not the arXiv; I gave a talk about my research program, and about 6 months later got an email from a graduate student saying he and a collaborator had solved one the problems in my talk. And you know what, it was great. He’d found a reference I hadn’t that made it possible to do lots of other stuff he hadn’t thought of, and I got to farm out that part of the research program to his paper. It really was a problem I wish I had a lot more often.

84 thoughts on “What are the downsides of the arXiv?

  1. I can (snarkily? snarXily?) confirm that if you typeset a private copy of Mem AMS 127, stick in lots of disclaimers that it is a private copy including many corrections of typos and the occasional update, and submit it to the arXiv, then it will – probably justifiably – be rejected as “not an original contribution of research or exposition”.

    On the other hand, if you claim that the reals are countable then this seems to be fine.

  2. As far as I know, the only arXiv policy statement of the kind you are looking for is

    “arXiv is an openly accessible, moderated repository for scholarly articles in specific scientific disciplines. Material submitted to arXiv is expected to be of interest, relevance, and value to those disciplines. arXiv reserves the right to reject or reclassify any submission. Submissions are reviewed by expert moderators to verify that they are topical and refereeable scientific contributions that follow accepted standards of scholarly communication (as exemplified by conventional journal articles).”



    By the way, there’s also an interesting question of what their policy on trackbacks to discussion elsewhere of arXiv articles is (these are now heavily dominated by MO trackbacks), with the only formal policy:

    “We reserve the right to reject trackbacks for any reason.”

    In practice this actually means: pretty much anything is allowed from anywhere, as long as its not from someone who has been critical of the research program promoted by certain arXiv moderators.

  3. Ah ha! I was hoping for this.

    I share Ben’s worry. But, ultimately, you can still post things to your own webpage; or to university eprint servers if they exist. The point is that, with the internet existing, _something_ like the arXiv was always going to exist– personally, I think the fact that we have _one_ arXiv, and not loads of competing outlets, is really useful, and out-weights the problem that some people/preprints won’t get past the gatekeepers.

    For example, I (and other people I know) worry about “loosing citations”, where a colleagues print paper only cites your arxiv paper. This is a pain; but now the arxiv has added the ability to cross-link to printed journals. And with the arxiv being by far the biggest game in town, I hope it won’t be too long before software can automatically see an arxiv citation, and then check to see if this links to a printed paper. It would be much harder to do this if everyone used an eprint server run by their university, say.

    Responses to Igor’s “problems”:

    i) The arxiv is public, but so is sending an article to a journal! If you’re really worried about someone picking up your ideas before you’ve fully explored them, then why would you send them off to a journal, to be read by the experts in your field– exactly the people best placed to exploit your ideas before you can! So, if you’re willing to submit a paper to a journal, then there seems little danger in also sending it to the arxiv. And if the paper isn’t ready for a journal? Then, it probably isn’t ready for the arXiv!

    ii) This, if it really happened, seems pretty insane– would the PhD advisor have acted in the same way if the paper had appeared in a print journal? Or if it had arrived by (e)mail from the author? I would hope not (in which case, why behave differently with the arXiv?); but if so, that strikes me as a problem with the advisor, not the arXiv.
    (I _will_ agree that there is some problem here– in the “old” days, it was easier to prove that you’d worked independently, by submitting to different journals without actually knowing another journal was considering “the same” work by another author. I can see that this is harder with the arXiv. But, it is just a risk you run, and again, I do think the benefits of increased communication outweight the risks).

    iii) Many papers are wrong! (And a similar problem can occur when someone moderately or very eminent “announces” a result at a conference, say, and then doesn’t publish– it’s pretty hard to write a paper which follows up, or maybe repeats, the work). Again, I acknowledge this as a slight worry.

    iv) Well, that’s a risk you run. I’d hope people might forgive youthful enthusiasm! Mistakes (even big ones) get published; see my comment above about, if you submit to a journal, you’re “going public” to the people who matter most– the experts in your field.

    v) As above.

    vi) You don’t have to read the arXiv! And you could spend 2 hours a day reading journals (and beware that Physics might have rather different research styles to Maths).

    vii) I’ll accept this, a bit. But, the pressure to publish, to my mind, is more driven by universities, and funding bodies.

    viii) I’ll give you this one!

    Sorry, bit tedious to rebut each point. I guess my general point, one Mel made, is that have these issues can be solved by simply not using the arXiv! The other issues are societal, but again, you cannot change the world (i.e. remove the internet) and all these problems seem to simply be slight extensions of problems which already exist in some other form.

  4. Thanks, Ben, for starting this thread. My own opinions are pretty far out on the pro-arXiv end of the spectrum, but I am nevertheless very interested to hear about what various people consider to be downsides of the arXiv. I would have preferred that Igor’s original question stay open on MO, but having a thread here is a reasonable substitute.

  5. I have issues with the timestamp on versions. Topology papers have lots of pictures. If I uploaded pictures which were jagged, then redrew the pictures to be nicer, it would be a whole new version. This prevents me from doing so in practice. So I have 2-3 papers on the arXiv now which I’d like to “minor edit”, but won’t, because it will generate a whole new version.
    I actually see this as a huge downside of the arXiv (maybe print journals are even worse). The lack of a “minor edit” feature is extremely bad. You should be able to tweak after it’s been posted without generating a new version.

  6. Daniel- I agree 100%. I’m curious, though; do you think that the determination of which edits are minor should just be on the honor system?

  7. @Matthew’s point #1:

    With respect to worrying someone will steal/supersede your ideas: There is a clear difference between sending a paper to a journal, which will have probably only one person look at it, and which has a good chance to get it formally into refereed print vs. putting it on the web for anyone in the world to see, in a non-refereed venue.

    I could see (and have seen) the following happen. Someone will see the paper, improve the results in a way you didn’t foresee, and publish those improvements before your paper is accepted. Now, when the referee finally gets to your paper the referee says your result is superseded, and you cannot get it in a decent journal.

    On the other hand, suppose you submit it to a good journal, it is published, and then that other mathematician again writes a paper improving your result. You don’t have to worry at all about her paper preventing yours from appearing in a good journal, you appreciate the citation she will give (which is now to a “real” paper, rather than a preprint), and all is well.

    Or how about the following: You spend years working on a very difficult theorem in topology. The paper you finish is clearly tier 1. You distribute it, and by some stroke of sheer genius another person realizes (without having to cite your work, but from insights gleaned from your work) that the class of spaces you are considering is empty. This person immediately puts his ideas on the arxiv. Now, when you submit your paper, you must put the disclaimer that this genius has made your work, in essence, vacuous. Is it still as likely to be published in the same venue?

  8. Or how about the following: You spend years working on a very difficult theorem in topology. The paper you finish is clearly tier 1. You distribute it, and by some stroke of sheer genius another person realizes (without having to cite your work, but from insights gleaned from your work) that the class of spaces you are considering is empty.

    This is what I mean when I say that “downside” is a debatable term: obviously this sucks for you (the hypothetical writer) personally, but I would say is a success for mathematics as a whole. We really don’t want papers which are vacuously correct going into good journals.

  9. I could see (and have seen) the following happen. Someone will see the paper, improve the results in a way you didn’t foresee, and publish those improvements before your paper is accepted. Now, when the referee finally gets to your paper the referee says your result is superseded, and you cannot get it in a decent journal.

    I will spot you that this is a serious issue. I would be curious to see if this is something that people think happens a lot; it would require pretty fast movement on the part of the other person.

  10. Let me begin by saying that I have no patience for the “stra men” typo of comments, like “oh, you are worried about car safety, you don’t like cars driving 120m/h – well then you can just sit at home and don’t drive.”

    About the Ben’s question: “I would be curious to see if this is something that people think happens a lot; it would require pretty fast movement on the part of the other person.”

    Yes, this does happen periodically. As an editor, occasionally I get desperate emails from my colleagues asking if I can help. Things are a bit more complicated though. Here is a typical case, the one I dealt with last year.

    A person X with student Y write a pioneer paper which proves the first result of a kind (still failry weak) and submits to Inventiones (not an actual journal in that case). X goes to a conference, hooks up with an expert Z in adjacent field and together they extend the result to a somewhat stronger, writing a paper maybe 6 months after the XY paper. A year after submission Inventiones rejects XY paper with “results are not strong enough” rationale, and a less prestigious journal accepts XZ paper at about the same time. XY submit to a another journal where the paper is rejected as weaker than another accepted for publication paper. So the student Y goes from the breakthrough and Invetiones submission to now publication at all. Ooops… Then X starts sending emails asking XY paper to go somewhere as otherwise she/he feels that she/he accidentally just destroyed student Y career.

    My view: X and Z made a mistake putting XZ paper on arXiv to begin with, and updating the status of teh paper on the arXiv (or her/his webpage). This was a very avoidable mistake, as they both feel it is the XY paper that should get a widespread recognition as a pioneer work, not the XZ paper.

  11. Hi Igor, it’s not clear to me that’s a problem with the arXiv — the arXiv is in the story but it’s not a central player. In particular, precedence is frequently based on when the idea first appears in the arXiv so to some extent publication in a journal isn’t relevant, except for when it comes time to apply for jobs or grants where paper counts are a factor.

    People choose when to talk about results and when to publish. Your example sounds like someone who could have perhaps sat on their ideas for a little longer and made them more ripe. Then again, maybe someone would have scoooped them entirely if they had done that. You could argue if they didn’t post on the arXiv perhaps they would have never received any credit at all for those ideas.

  12. Ryan: I think Igor is saying that it is X and Z who made something public before they should have; I get the impression from his story that student Y did not post on the arXiv.

  13. Indeed, it seems that XY posting to the arxiv in the first place would have largely solved the problem — everyone in the immediate field would then have known that this was the breakthough paper, and hopefully the referee and editor at the second paper would make their decision taking this into account.

  14. Igor – I don’t see your anecdote as a problem with the arXiv (I’ll note, you don’t actually mention the arXiv until the end when you sum up). It sure sounds like the problem is that X and Z got their paper published and that they didn’t include Y as an author.

    It’s also very far from clear to me that Y’s career is ruined; I’ve had a lot of papers rejected from two journals (often neither of them as good as Inventiones) and my career seems to be doing OK. Furthermore, if Y made a really important contribution to the field and X and Z recognize this, presumably they will write very good letters for her/him, which are more valuable on the job market than +/-1 publication.

  15. Ben – that’s right, this story happened not because of the arXiv per se, but because of the culture that arXiv fosters where everyone is expected to post their latest thing and update it as soon as possible, without thinking. I was just trying to answer your question whether these things even happen because of the timing. What happened I believe was that X just acted in haste posting XZ, not thinking that it might have an adverse effect on the future of XY paper. In the good old days when the referees of XY would not have read at XZ on the arXiv, both papers would have been published without a bit of trouble.

    Again, I am not saying these kind of things never happened before, or that we should stop using arXiv in fear of competition. However I do think posting on arXiv should not be viewed as “a must” and that we should fight the culture above.

    P.S. Scott, there was indeed a semi-happy end to the story. Eventually XY paper was published in an ok journal and after the initial scare Y did get a job.

  16. So, I think it’s pretty clear that “some but not all papers are posted to the arXiv when they’re submitted” is not a good stable equilibrium, and that in this situation the people who don’t post can get screwed over. As such, if I were someone who didn’t post to the arXiv I’d be annoyed about the arXiv. Nonetheless, I think the “everything goes on the arXiv” equilibrium is far far preferable than the “nothing goes on the arXiv” solution (if only because the arXiv is publically accessible and journals are not).

    In the particular example Igor discusses, I think if you want your referee to think something like “XY is the breakthrough paper which later lead to small improvement XZ” then you should just *say that* in the introduction. It sounds like this is just an example of a referee being misinformed, which is certainly a problem, but hardly one that seems special to situations involving the arXiv.

  17. So I have a lot of sympathy for the position that people are in too much of a rush to write too many papers. But I don’t think the arXiv is the driver here, instead I would blame the postdoc system. As long as you have people in 2 and 3 year positions they’re going to be in a big rush to get stuff out.

  18. Noah, somehow your binary choice “all/none of papers on arXiv” doesn’t feel right. For example, as I mentioned on meta thread, we *can* fix things, including having arXiv datestamp papers to be revealed at a later date, or sooner if the author wants it (say, a competitive paper is posted). This would give people more options is using arXiv. Other potential improvements include for example, the “minor change”, removed restriction on unaffiliated authors, remove restriction on the size of the files (I had problems once with a 3Mb paper), allow .pdf even if this is “made from .tex”, linking of arXiv to MathSciNet author datebase (aXiv does a poor job combining and distinguishing authors) and perhaps many other. I really don’t know what other problems people have with arXiv – that’s why I asked the question.

  19. As for the “postdoc system” – you are right here, but that’s not the fight I am prepared to have. This problem is deeper though – it’s really rooted in the tenure and “tenure track almost sure to become tenure” system which makes high volume publishing during postdoc 2-3 years so valuable.

  20. On the “minor change” front, the recent arXiv upgrade did improve things in that direction (you can change meta-data without reposting). I agree that there might be room for further improvement in that direction (though I can see “minor edits” introducing new problems as well). The size thing doesn’t seem to me to be a problem. The current system of “if you have a good reason to go over the size limit you email them and ask them” seems to work to me. For both the size restriction and meta-data linked to MathSciNet what you need here is more money and time invested into the arXiv. Correct me if I’m wrong, but the arXiv doesn’t have the financial resources that MathSciNet does. I think the priority of funding agencies are wrong here, and they should be giving more to the arXiv.

    The delayed timestamp thing seems really weird to me. I guess this is that I have no attachment to journals and have never been involved in the pre-arXiv time-delayed world, but I still don’t get it.

  21. @ Daniel and Ben, re “minor edits”: I guess I don’t understand the problem with minor edits resulting in a new version. There’s an intrinsic advantage for others being able to refer to a specific version of an arXived paper, as even changing a single word or adding an equation could affect pagination or numbering of various elements. The “comments” field on the arXiv effectively lets you (on the honor system) flag which when changes were minor. E.g. I have a paper (arXiv:1002.3423) where the comments field reads

    22 pages, 4 figures. V2: Corrected Table 1.9; V3: Added Table 1.10; V4: Minor edits; V5: Corrected Figure 2.1. To appear in AG&T

    and I don’t think there’s any shame in having made several rounds of minor revisions.

  22. Off-topic, but I think MathSciNet is as deserving of funding as the arXiv, though perhaps not as disproportionately as Noah mentions.

  23. I’m with Nathan — what’s the big deal with minor edits being timestamped? I’ve frequently seen people hesitant to post new versions. Does it really look bad to enough people to worry about?

  24. Also, regarding Igor’s suggestions for improvements above.

    Restricting non-affiliated authors submissions is a compromise to limit the amount of moderation work required. I think one shouldn’t suggest such things without proposing an alternative mechanism for moderation, with a commensurate workload.

    Also, allowing PDF uploads, even when the TeX is available, is potential dangerous to the longevity of the arxiv. Already, very old PDF files do not render properly on some modern software! The ability to regenerate the typeset version centrally is very important for the long term preservation of the literature.

  25. At the risk of misinterpreting Igor’s position or his points: it seems that two related issues are being conflated. One is “do people know of negative aspects arising from people’s use of and promotion of arXiv?” and the other is “what things about how arXiv currently works are, on occasion, not good”?

    Premature announcement of results seems to me to belong in the first category; inconsistencies of moderation would seem to belong in the second one.

  26. I’ve frequently seen people hesitant to post new versions. Does it really look bad to enough people to worry about?

    I guess it depends on community norms (of one’s own research tribe, or of the expectations one might be trying to live up to). To be grossly simplistic: some people, in my experience, really don’t like the impression of sloppiness and tinkering caused by frequent edits to what should be — in their view — a considered, semi-public piece of work.

    I also know of one case where an eminent specialist’s renumbering & rewriting of things between drafts of preprints caused some grumbling from people making use of his results and relying on them being correct as stated. That said, the fact this thing actually was on the arXiv was probably beneficial to them in the long run.

  27. @Nathan Dunfield: In that case, perhaps it makes sense to be able to mark a replacement “minor edit”, which doesn’t include it in new listings or generate a new version, but is available in a “history” tab refering to the paper, so that somebody who needs the information can find it, and one can reference the text as it was on a specific date.
    I don’t know that it looks terrible, but I have the tendency to have second thoughts about this or that tiny thing after having posted it (maybe the lines in the image don’t curve quite right, or maybe I want to slightly modify a sentence, or maybe to correct a typo), and I won’t do it if it will generate a whole new version.
    So the obvious and essential fix would be to be able to flag something as a “minor edit”, not generate a new version, but have that information recorded (in a not-immediately visible place) for reference purposes for the few people to whom it is really important.

  28. Sorry, I didn’t actually mean “relative to MathSciNet” (which I believe is funded by subscription, not grants). I meant the NSF should fund the arXiv instead of spending its money paying me…

  29. @Noah Snyder: I’ve had horrible experiences with the file-size thing. What should really be done is that the software should strip down eps files to the basics as a default option- it’s eps bloat that accounts for most file-size issues. But editting hundreds of eps files by hand is no fun.
    I had one paper- it drove me bananas- where the file size was 3.5 times their limit. I e-mailed them, they let me post it (they should have stripped it down), and then I stripped the eps files down to size- no visible change at all to the reader- asked to resubmit, and the moderator posted it to make it a whole new version without asking me! I got so angry.
    You’re right that money would solve the problem- if they had more funding, these issues would probably go away. That’s a definite downside of the arXiv- they had insufficient funding, so their software is poor and so is their user support, despite the best intentions of the people working there.

  30. I guess everyone’s milage varies, but I’ve found the arXiv’s software user support quite good. Certainly much better than the software or user support of any journals that I’ve interacted with. (Which is not necessarily to say that it’s better than the AMS, which seems to do a very good job.)

  31. Couple of minor comments.

    On the main topic, Igor’s anecdote seems to me a problem more with the editor of the journal. I would regard it as the editor’s job to sort out issues of priority and that when there is a doubt to contact the authors involved to determine the correct sequence. In this case, this sounds like it would have been easy to do. So perhaps there ought to be a guide for editors (many of whom started their careers in the pre-arXiv era) on how to take into account the existence of the arXiv.

    Timestamping is easy. Simply send a copy of your document to a bank. Or post a copy somewhere very public but encrypted.

    Minor edits is a feature of what I regard as the major problem with the arXiv. Namely that it tries to be two things with the result that it doesn’t do either as well as it could if it did just one. It tries to be the mathematical equivalent of both sourceforge and freshmeat. For those that don’t know, sourceforge is where you keep your code and freshmeat is where you announce it. I don’t see why we couldn’t have a system more like FOSS where we have a “stable” version of a paper and a “development”. Only new stable versions get announced, but if someone wants to look at the latest version with all typos and so forth corrected, then they can get it very easily.

    So we need to launch MathForge (and yes, I have already registered the domain name).

    (Incidentally, I welcome the move of this from MO to here. Of course, it would have been better in a forum where discussions could be properly threaded and started by any participant, but here’s better than there.)

  32. This is a really interesting debate. I’m sorry Igor thinks I made a straw man comment– but as I tried to argue, there is a difference between “societal problems” (the sort of culture which the arxiv contributes– a real problem for us all, if you do think the arxiv is to blame) and “personal problems” which _can_ just be solved by ignoring the arxiv.

    A lot of the debate above has been about how ideas can be “stolen” (or, rather development more quickly by others). If you’re worried about this, then just not posting to the arxiv until, say, just after a paper is accepted by a journal does seem to solve the problem. (Maybe I’m wrong, but I think this is an overstated problem). Maybe Igor and others fear that eventually, we’ll get to the point where you’ll just have to post to the arxiv to be taken seriously– yes, that’s maybe a risk; but we seem very far from being there yet.

    It seems that the debate maybe comes down to whether to be more or less open. I think the arxiv is incredibly _useful_ in making things more open– it gives anyone access to the sort of information that maybe once you’d get via rumour, or you’d have access to if you were sufficiently senior and people sent you preprints etc. The arxiv allows people who are geographically remote (e.g. unable to get to conferences) or who don’t have access to journals to still keep up-to-date with current work. I like to move between different areas, and I’ve found the arXiv _incredibly_ useful for finding out what’s going on in areas where I just don’t have the personal contacts. If I lived and worked in the USA, and could attend lots of conferences, maybe I’d need it less, but I work in the UK, and find it hard to attend conferences.

  33. On re-reading, maybe I’m misinterpreting Igor’s main point– the arxiv fosters this “race” to publish quickly, and it’s this which can lead to problems. Yes, I guess I have some sympathy with this. But, at least here in the UK, we get judged on our research outputs by the government. You _have_ to produce a certain amount of work of a certain quality in a certain timeframe (it’s actually not that short a timescale; I don’t wish to moan too much). This is just the culture we’re in. I don’t like it, but it seems a stretch to suggest the arxiv is anything more than a peripheral cause.

    But if the argument is– please, step back and _think_ before you rush to push the next preprint out the door (especially if it might hurt your own grad student or postdoc!!) then I agree.

  34. re: Matthew Daws (33): To me there ia a much simpler interpretation of Igor Pak’s point, which I think is this: To him (and thinking about it, I agree) it seems relevant to point out to people, in particular to those just starting to publish, that there is a risk (or risks) in making your work available rightaway (by posting to the arXiv). A risk is that somebody else will pick-up your idea and develop it in a direction in which you actually wanted to go yourself; and this other person will do so faster than you manage to do so yourself and thus ‘destroy’ or ‘harm’ your research program (and to avoid a misunderstanding, I do not mean the ‘destroy’ or ‘harm’ as a willingly destructive or unfriendly act, but just as a side effect of normal scientific development). In contrast, if you did not post a preprint and just submitted the paper, you would have had, say, perhaps a year or at least a couple of months to continue your research program before the public saw your work and in this way you minimize the risk of being ‘overtaken’). Of course, there are also advantages of making your work available right away (say, somebody might suggest something you would have never thought of and in this way help your research program, and of course you want this help as soon as possible).
    Or, perhaps, you have no intention/idea to continue with a line of research, or various other reasons.
    Whether this potential problem resulting from posting to the arXivis relevant in a given situation and/or is more important than potential benefits of posting, of course has to be decided by the aothor in the individual situation. But advice of the form: ‘Before you decide to post to the arXiv, consider whether this [insert reason] could be a problem for you. And then make your decision’ seems good advice to me. To collect examples for [insert reason] to me seems to be a good idea. Sort-of to assemble a check-list to go through before posting; if everything is a non- or minor-isuue, then post if not rather don’t.
    [Of course there are other ways to make your work available before formal publication in a journal, but for various reasons, mainly but not only its visibility, I think the arXiv plays a special role; and it is not the same or even similar, say, to sending out preprints by email or giving talks on the subject.]

  35. re: Noah Snyder (16): I do not really see ‘…in this situation the people who don’t post can get screwed over’;
    assuming the person at least follows what is going on, on the arXiv. (Perhaps I am missing something.)
    To me the situation is this: the worst thing that can happen to somebody (say author N) that (typically) does not post to the arXiv is that somebody else (author P) post something to the arXiv that N has already ready (perhaps submitted 5 months ago). But, then N can simply post (as an exception) the next day his/her work pointing out that this is already submitted since 5 months. (In case this needs to be ‘proved’ in a subsequent discussion, the journal submission provides a, I belive sufficiently credible, ‘time-stamp’) For N this is perhaps a minor annoayance to do and discuss this; but for P the situation is potentially way worse; and if N had posted right away he/she might have saved P a couple of months of work that are now lost or at least somewhat devalued. And, if P had also not posted, the publication procedures might have taken there course and perhaps P, the editor, and the referre would have stayed all unaware of the work of N, and P’s work would have appeared (all parties acting in good-faith) perhaps even before N’s. And, then months down the road it would eventually be established that N and P did this independently and N did it somewhat earlier.
    (How this ends regarding ‘credit’ say in form of citations, will I believe depend more on other circumstances then these 5 months.)
    In short, I cannot see how regarding priority posting submission-ready/submitted manuscripts to the arXiv can be a real advantage. (Of course, if somebody posts ‘annoucments’ things are a bit different.)

  36. I wonder how much the arXiv really changed things with regards to issues of priority, others improving someone’s results faster, and the other negatives that Igor raised. Mathematics had a “preprint culture” long before the arXiv existed, and it was common practice for people to mass mail a preprint to everyone they thought might be interested at around the same time they submitted it to a journal. (There were also various preprint series associated to universities and institutes which had more formal distribution mechanisms.) In contrast, in some lab sciences giving out preprints ahead of actual publication is simply not done, for whatever reasons.

    I’m not quite old enough to be able to really be able to answer my question — the arXiv broadened to all mathematics when I was in grad school, but I suspect that the arXiv may not have changed things that much…

  37. I’m just barely old enough to comment of the “preprint culture” to which Nathan refers. My overall impression is that most the problems/dilemmas discussed above regarding the timing of making one’s work widely available existed pre-internet and pre-arXiv. The arXiv affected these issues in a quantitative, not qualitative, way, and the quantities, while significant, are not huge. I would say a factor of 2 rather than a factor of 10.

    As a grad student, I was fortunate that my advisor (Rob Kirby) was one of the central nodes of the preprint network. He always had a large stack of recently arrived preprints on his desk, which I would skim through periodically. I recall, in the early quantum topology days, that people felt pressure to circulate preprints as soon as possible, lest they be scooped. (There was a lot of post-Witten/post-Jones low hanging fruit to be collected.) It was common to see an initial quickly written preprint, followed a couple of months later by a more fully worked out revision.

  38. @an_mo_user(36): I’ve thought about this a bit, and actually, this convinces me. This _has_ worried me in the past when I’ve put a preprint on the arxiv. And I’m not quite sure what I’d do in the position of person (P).

    On the other hand, in the position of person (N), it would have been better to have posted on arxiv at the time of submission. It’s the asymmetry which is the problem.

    And, again, I don’t think the arxiv caused this problem– you same issue could arise by speaking about your work at a conference. Indeed, I suppose this must have occurred in the past– does anyone have anecdotes of how this worked itself out (i.e. (P) speaks about recent work at a conference, and in attendance is (N) who says that they submitted much the same work to a journal a few months back…)

  39. Let me make a couple of general points. I agree that “preprint culture” is older than the arXiv. I still remember some people (esp from Russia) travelling with suitcases full of their preprints, leaving them at every math department they visited. I am glad this is over. The an_mo_user (35) is right to take lessons from the NP story.

    But again, this is not black/white – there are actually soft ways for N to let P know she/he is working on the problem and proved the result. To see how this is done, we really need to look at other related modern cultures.

    In theoretical CS, they have a “conference culture” when “publication” is very rapid, but things are not public. What they do is publish a conference proceedings where every paper is pruned to 10-12 pp. This way everyone in the community knows who is working on what and claims to have proved what result, but the technical details only (if at all) appear in print. Or earlier if the authors want to, but the cultural norm does not force them. This system has its own fairly obvious weaknesses, but it does serve as a guide to potentially improving the arXiv. Simply allow the preprints in the “extended abstract” form, with “appendix” with hardest most innovative technical parts of the proof timestamped to appear at a later date. Of course, 99% arXiv contributors might want to simply ignore this option and make their whole papers available, but I really don’t see why not allow letting people use this option. More choice is good, and in the NP story would help both, I think.

  40. @ Andrew Stacey (32), “imply send a copy of your document to a bank.”

    Oh, please… You are not serious. Tools are made to help people. Rota tells a story about Gibbs who worked in American obscurity (back then) and became famous in Europe by sending everyone his preprints. ArXiv is made to simplify this. I am suggesting arXiv simplifies datestamping as well. This would cost next to nothing but expand people’s options.

  41. @Igor: Do I understand that your main point is that you think that people should be able to:

    1) Work on their projects at their own pace without being pressured by other people working on related questions.

    2) Retain full credit for what they’ve done privately before they publish it.

    Basically you’re saying you want math to move at a slower pace and be a more private activity? Or have I misunderstood?

  42. @Noah: Sort of. I don’t mean for “math” to change in any way. But I do want people, in particular my own students, to have these 1) and 2) as options, which they can use or not depending on a particular paper.

    The students sometimes need time to learn quite a bit of math before they can make further progress on their work (esp if in the neighboring field). In this case both options are not great: posting their latest work on the arXiv, or holding out until the these results are generalized and form a good thesis. This is all field specific, or course, but “joining the race” is not a good option either, as graduate school is exactly the time to learn fundamentals of a broad spectrum or mathematics rather than concentrate on latest developments.

  43. Igor, what I mean is that if you want to timestamp your work without publishing it then that is already possible with current technology. My wider point is that the arXiv does not need “feature creap”, that will only cripple it. Indeed, as I said, I would split the arXiv in to two.

    It’s the same issue as putting questions like this on MO. What makes MO great is its specialisation. Trying to make it something broader will dilute its effectiveness, in my view. The arXiv is the same. There are already effective ways of timestamping documents – I merely mentioned two, one technological and one “from the dawn of time”. There are more.

    GNU/Linux works so well because it is a lot of little things that fit together to do big things. MS works so badly because it is one big thing that fails miserably to do little things. We should emulate FOSS, not MS.

  44. Basically what you want is a more formal means for the current informal process where people know that some problem is being worked on by a graduate student and so avoid working on it?

  45. Andrew, everything is *always* possible with “current technologies” – the point is we can and should try to make things easier. By this logic, clearly arXiv as a technology is unnecessary – you can simply post a paper on your own page and send a mass email to all members of AMS, EMS, etc. about this new fantastic article you wrote. Do you really want to do that? Alternatively, why have email – snail mail technology works almost as well, right? Same with me going to the bank, encrypting my paper somehow, etc.

    Here is a bigger problem that I am hinting at – the arXiv is NOT a community but a (free) service, and that’s a bit of a problem. A community, even as large as English Wikipedia has various forums with a discussion on how to improve it, has rather remarkable pages with criticism of Wikipedia, has an elected body of bureaucrats, etc. Eventually the problems are hashed out and an effort is made to improve them. My “datestamp and late release” is just one idea that may or may not be important to many others in a similar situation (Noah – yes, I would recommend my graduate students to use it). But with arXiv barely changed in the last 10-15 years I really think there are potentially many other problems/downsides of using it. But only once we collect them all, discuss these problems at length, maybe then we can start lobbying to have arXiv change some of its rigid policies.

  46. But only once we collect them all, discuss these problems at length, maybe then we can start lobbying to have arXiv change some of its rigid policies.

    Incidentally, my view on this is exactly the opposite: one of the best things about the arXiv is its stable and rigid policies. I don’t want the arXiv to expand or change its mission, to experiment with new policies, or to try to reshape the mathematical community’s norms for how results are announced or credit is established. In particular, I don’t think the arXiv should make any changes unless they are near-universally agreed to be harmless or necessary.

    This isn’t to say nobody should experiment, just that it shouldn’t be done on the arXiv. One danger is that there are many things some people wish the arXiv would become: a discussion forum for comments on papers, a recommendation engine for suggesting related work, a rating system for quantifying popularity of papers, etc. There’s no technical reason why any of this needs to be part of the arXiv itself, and the only real reason for integration is that it would guarantee a large user base and would take advantage of the arXiv’s excellent reputation. However, that’s an awfully tempting reason. This means there will always be pressure on the arXiv to implement various ideas, but I sincerely hope they continue to resist the pressure.

  47. I strongly agree with Henry Cohn. All the major suggestions people make for changing the arXiv that he lists are controversial and would likely result in at least some people stopping using it. For instance, I very much doubt that I would put my papers there if there were some place on there for people to comment on or rate my work.

  48. The archive is a good service. However, over the years it has almost become a requirement that you be connected with an established institute making it hard for independent researchers to post a pre-print there even if have prior to rule changes.

  49. This may be a problem that only high energy theoretical physicists encounter, but some people invest nontrivial effort to compete for the first spot in the daily mailings. Apparently, Ginsparg has accumulated some statistical data indicating that people near the top get more citations.

    Also, I agree with Henry Cohn concerning the benefits of ArXiv stability.

  50. I also agree with the desirability of stability.

    That doesn’t stop us discussing what _else_ we could do or have. But as yet, I haven’t seen any proposals here for anything that doesn’t already exist in some form.

    To return to the software idea, I like GNU/Linux partly because it is made up of lots of little pieces, each of which contributes some functionality. I, as the end user, can pick and choose which bits I like and can ignore those I don’t. If there’s something I like that I don’t yet have, I can go and find it if it already exists. The only difficulty is in figuring out where to look, and in ensuring that the bits I use fit together. That’s where distributions help, but they aren’t necessary, and distributions’ main role is in fitting the pieces together, not in building them in the first place.

    So, yes, I can see the need for timestamping (to take one example). But I don’t see the need for it to be a part of the arXiv.

  51. @Ben concerning a minor quibble: You said: “We really don’t want papers which are vacuously correct going into good journals.”

    In some ways I disagree. Firstly, with your label. The paper in my example was correct, not for vacuous reasons, but for highly nontrivial reasons. It is important to realize that things which are vacuously true can be true for other reasons (and that the vacuousness can even sometimes be harder to prove!).

    Second, I disagree that a paper shouldn’t appear if it has an alternate vacuous proof, *if* the vacuousness is only shown later (and *especially* if the vacuousness came from ideas gleamed from your paper).

    But in any case, you prove my point. People would be less likely to publish your paper even though its quality (in terms of how much time you spent, and whether it was a viable area of research at the time) does not change between the two scenarios.

    It seems to me the trade off is the following: Someone has access to your paper a few months in advance vs. you lose prestige and thus the ability to do more good math. Sure, by putting the paper on the arxiv earlier you may have contributed to the faster advance of mathematics (and that’s a big “maybe”) but you may have cost mathematics in other ways (you getting tenure, having the paper read more seriously by getting it in a good journal, etc…).

  52. Pace-

    This is a tricky value judgement (what is interesting and what should be published is always a value judgement), but I also think its especially hard to make when discussing in the abstract. It’s very hard for me to imagine the kind of paper you are discussing; I don’t think I’ve ever encountered one in my life. I think it’s basically impossible for me to jump into a world-view where the possibility that I am studying an empty set of objects is something I worry about and a consideration which guides how I conduct my professional life.

    Trade-offs have to be weighed against their probability; an unpleasant outcome which is very low probability doesn’t carry a lot of weight. Do you really think this is more likely (or even in the same order of magnitude) than a valuable collaboration or interesting followup work arising from putting your work on the arXiv?

  53. Do people really spend years of their lives proving theorems about objects without first knowing some interesting examples of those objects (aside from cases where the (non)existence of those objects are subjects of famous conjectures — for instance, I could imagine proving theorems about purported exotic smooth structures on the $4$-sphere)? I find that totally bizarre. Certainly as a referee I couldn’t imagine recommending that such papers be published even in pretty middle of the road journals…

  54. Andy,

    I think your disclaimer “aside from cases where the (non)existence of those objects are subjects of famous conjectures — for instance, I could imagine proving theorems about purported exotic smooth structures on the $4$-sphere” is more than a disclaimer, but excludes most examples I had in mind.

  55. Ben,

    Are you unfamiliar with papers on the existence of Siegel zeroes? Or zeroes off of the half-line? Or computations related to counter-examples to Goldbach’s conjecture? Or odd perfect numbers? I can give you many, many more examples. Before Perelman there were partial results about what exotic smooth $4$-spheres looked like (as Andy mentioned).

    To answer your question: I think that interesting follow-up work can be done just as easily if you post your paper to the arxiv *after* acceptance (if, as most do, the journal allows the posting of a preprint). There is the possibility that your original paper could have been improved by collaboration, and if you want to seek for collaborators then arxiv might be a good place. But, in my experience, collaboration is not always a positive experience. Whether one would want to collaborate depends on how original the ideas are, how much work has already been put into the project, how much the proposed collaborator can really contribute, whether you can work well with the collaborator, and so forth. I think most collaboration regarding papers which are complete enough to post on the arxiv could be done in follow-up papers just as easily.

  56. Pace-

    Basically yes. Those all sound plausible, but I’d think I’ve ever read such a paper.

    I will admit though, when I read your comment, I was assuming that you were referring to a paper written from the perspective that the objects probably exist. Probably I was just misunderstanding you, but that was the idea I was having trouble summoning.

    I don’t see your examples as bolstering your point though. If you publish a paper which helps someone find the insight to prove the Riemann conjecture or that there are no odd perfect numbers, I doubt you will have a lot of trouble getting it published. You will feel like an idiot for not getting there yourself likely, but that’s life.

  57. Yes, I’d been meaning to come and mention that one! Curiously, I don’t think that Akbulut pulling the rug out from underneath us really hurt. Quantum Topology happily took the paper, and it’s already received about a dozen cites.

  58. @Ben: “I don’t see your examples as bolstering your point though. If you publish a paper which helps someone find the insight to prove the Riemann conjecture or that there are no odd perfect numbers, I doubt you will have a lot of trouble getting it published. You will feel like an idiot for not getting there yourself likely, but that’s life.”

    Maybe so, and yet I know of situations where even small generalizations (which did not make the previous results vacuous) made it so a decent paper (maybe not tier 1) could no longer be published in the same venue. Your doubts on the matter do not convince me. I think we’ll have to agree to disagree on this one.

  59. Like I said, this stuff is just bound up in a very serious way with one’s experiences. I can’t bring myself to worry about the possiblity of my work being made redundant by being putting on the arXiv for the same reason I don’t worry about my ideas being stolen: it’s never happened to me, or in a dramatic way that I noticed to anyone I’m close to. The example Noah cites above is probably the best example, and as Scott notes, it sort of illustrates the opposite point. (I do take seriously the concern that one will put insufficiently worked out things on the arXiv and thus be embarrassed, because I have done that; on the other hand, my feeling is that this has not had too many ill-effects on my life). Furthermore, in my own life, I feel like the returns to openness have been huge.

    I don’t say this to invalidate anyone else’s experience, but to point out that we all come at this from different perspectives.

  60. I hope you don’t mind me picking apart your last post. It will bring the discussion a little bit off topic, but I want to flesh out your comments a bit.

    Ben: “Like I said, this stuff is just bound up in a very serious way with one’s experiences.”

    The same could be said of smokers. Or of those who leave their front doors unlocked. Some smokers have only ever had joy in the fact they smoked. Some people have never experienced a break-in, and don’t view it as likely, so they leave their front doors unlocked. Do their experiences give them sufficient knowledge on the subject to suggest to *others* that they should similarly smoke, or leave their front door open?

    I understand that you believe there is a very low probability that ideas will be stolen, and that the possible negative impacts of the small possibility of having your paper rejected because another author used the preprint to supersede it is outweighed by the benefits of earlier openness. You see Scott’s (non-)example as evidence to bolster your view, along with your own lack of examples, while the experiences I relate don’t dissuade you.

    And frankly, what is the probability that if I leave my front door unlocked for the rest of my life that my house will be burgled? Or, if it is burgled, will it be because I left the door unlocked or simply because the thief wanted to break in? Would hearing of a neighbor’s burglary convince you that you should start locking your door? I imagine that the chances that you would be burgled are quite low in any case.

    But now imagine that, as a society, we all start leaving our doors unlocked. Suppose you advocate for openness, and those in thief-ridden areas take your advice and find themselves burgled, not realizing that your advice was probably meant for those in safer neighborhoods. But even then, would it be wise for *everyone* to leave their houses unlocked? How many burglaries does it take before we need to start locking our doors?

    So here are some of the points I hope I’ve made. First, it is one thing to use your own experiences to decide upon a course of action, and another to *advocate* for all/most others to do similarly. Especially when you are familiar with *contrary* experiences. Second, it is important to qualify statements. For example, you might mitigate the possible damage by only advocating for posting your preprint before submission to a journal to those who already have tenure. Third, while we all “come at this from different perspectives” it is important to take all available data and put it into a coherent whole. We wouldn’t want a chain smoker, who hasn’t suffered from cancer, to ignore the experiences of others. Similarly, while in my view I recognize that there are situations where the behavior you are suggesting often (and maybe even almost always) does not result in negative consequences, you do seem to (contrary to what you said) be invalidating the experiences I have related that there can be a real danger occasionally.

  61. Pace-

    I think you’re right that we’re going to have to agree to disagree. I will mention, it’s not just that this hasn’t happened in my life; it’s not a concern I hear about from other mathematicians (with the exception of you and Igor Pak). In my experience, most mathematicians have a lot more trouble with people not reading their papers than with people reading them; when other mathematicians I know start to express the same concerns as you, then perhaps I still start to worry about them for my own purposes.

    (I’ll just clarify, since Pace used the word “stealing” above, that stealing ideas is not what I am talking about or what I thought this discussion was about; I thought we were talking about the danger of people building on your papers in a way which is legitimate and ethical, but still disadvantageous to your career because of the effect they have on your publishing prospects).

  62. On the question of whether it’s appropriate to give advice to others on these matters, I think it’s important to decide whether you think “mathematics” as a whole benefits from earlier dissemination (I do). I think it’s reasonable to say “there’s a small risk this approach will backfire on any given person, but on the whole we all end up better off, so I think we should all try to take this approach”.

  63. Ben: “In my experience, most mathematicians have a lot more trouble with people not reading their papers than with people reading them;”

    But that isn’t the issue, because posting your preprint on the arXiv after being accepted for publication will get you the same readership as if you post it beforehand. Your issue was with inviting early collaboration and getting extra commentary on preliminary versions; which is where the arXiv really shines. [Note: To put things into perspective, if you look on the arXiv you will find that I have put some of my preprints there long before they were accepted for publication. I do this for the reasons you have expressed previously, to get some comments back and invite collaboration. I think this is a great idea, just not *universally* great for all papers for all mathematicians.]

    “…when other mathematicians I know start to express the same concerns as you, then perhaps I still start to worry about them for my own purposes.”

    It appears that you are getting caught up in the specifics of the example I gave. It might not be the case that a large number of mathematicians are concerned specifically about their ideas being made vacuous; but from my experience a large number do recognize that there is a danger to making your ideas part of the “ether” before you have published them. My example was meant to just show one specific case where revealing your ideas might have a negative impact on the later publication of your work. In papers not dealing with non-existence issues, my example is probably moot.

    But your response also misses the point of my previous post. I personally don’t care if this changes your personal behavior. You can do whatever you want, and you can disregard the examples Igor and I gave you. [And, depending on your situation, I might even *recommend* that course of action!] However, if you start advocating for *others* to behave the way you do, you better be darn sure that the negative affects are minimal. And you better provide a warning to them if you are familiar with cases where posting a paper early led to hardship.

    “(I’ll just clarify, since Pace used the word “stealing” above, that stealing ideas is not what I am talking about…”

    I should have pointed this out myself. Thank you for making it explicit.

    If you make your work public and someone else uses your ideas to do something you were planning to do, or generalizes your work, or makes it vacuous, or whatever else, you cannot call that theft (even if you feel cheated).

  64. Scott,

    That seems to me a one-size-fits-all approach. Earlier does not always mean better. One should also ask:

    Do I want collaboration on this project?
    Is the paper ready for feedback yet?
    Would this embarrass me or my institution in the state it is in (e.g., have I distributed it to a few friends first)? [You all might remember a colleague of mine who is a very very good mathematician, but put up a preprint on the arXiv about the Riemann hypothesis before vetting it. From what I understand he thought that putting it on the arXiv *was* the vetting process. Others thought otherwise.]
    Is there a chance I could use the ideas in the paper to write another paper? Do I need a follow-up paper?
    Do I work in a field where others would use my ideas before I have a chance to explore them myself?
    Do I need more publications for tenure?


  65. To just follow-up on the following thought in another way

    Scott said: I think it’s reasonable to say “there’s a small risk this approach will backfire on any given person, but on the whole we all end up better off, so I think we should all try to take this approach”.

    This is exactly what we do with children, when we vaccinate them. There is a small but sizable risk to every vaccination. Deaths do happen. And yet we are all better off if we all get vaccinated.

    I contend, even in this example, it is still important to clearly state the possible side-effects (especially since this can help us mitigate them–we can take a child to the hospital if she reacts badly) without trying to pretend the risks are not real. It is also important to make exceptions from time to time. For example, if your child is currently sick, you should reschedule the shots for another time. (Or, in other words, if you are working towards tenure, and have an ongoing long-term research project, you probably shouldn’t reveal all the details of what you’d eventually like to do at the end of your first paper, unless you really wouldn’t be harmed if someone else uses your ideas to solve the problem first.)

  66. Here is how I think about the arXiv: In my field, the arXiv is to be the main platform to advertise your paper. Many people get the daily subject mailing, others check periodically for new articles in their area.

    This has a couple of consequences. The arXiv v1 is the version of your paper that will be read much more frequently than any other version of your paper, including the final published version. In particular, it would be a mistake to post “preprint”-versions of your paper – any bad or badly written paper posted on the arXiv will harm your reputation (your colleagues will be less likely to look at the next paper you post on the arXiv).

    So nobody should post articles to the arXiv before they are ready to submit to a journal. But once they are ready to be submitted to a journal, I don’t see any reason not to post them to the arXiv. If the paper contains any new ideas, people are more likely to attribute them to you than if they first read about them in another arXiv posting. They are more likely to invite you to give the most prestigious talk at this famous conference. They are more likely to use your ideas – which is always good for mathematics and 98% of the time good for you. If you wait the 3-20 months until your paper is accepted, maybe some of your ideas are standard by now, similar results have been proved by others, etc.

  67. The idea that autism is caused by vaccine has been thoroughly debunked. There has never been evidence for a connection, and the main person involved in the initial claim was shown to have invented data, and to have done other unscientific and unscrupulous things. All other coauthors on the paper have since completely rejected the paper.

    On the other hand, anyone who takes their kids to get a shot is told that there is a chance their child will die. I forget the actual numbers, but the doctors told them to us.

    Ben, the way I’m reading you is that you are comparing me to these unscientific people who try to link autism to vaccine. That my claims of a danger in revealing your work too early are unfounded, and invented. That they are not “scientifically supported.” If not, I don’t understand why you brought it up. If so, shame on you and I am done here.

  68. Pace-

    That was not my aim. I’ll just make my general disclaimer that statements made on the internet lack the usual context of human interactions and thus are easy to read as being rude when they were not intended to be. For example, above, it sure looks like you compared people who do follow-up work on other people’s papers before they are published to burglars (I’m not complaining about that analogy, just pointing out that it could be interpreted in an uncharitable way).

    My point was just that people have trouble weighing risks in a way that really balances their payoffs and probabilities (another example is that people move to the suburbs to protect their children from crime, but car accidents kill many, many more children, and of course are more likely when you drive everywhere). Now, I’ll note that’s not a claim that the risks you’re talking about *are* overblown, just that I don’t feel like I have the data to judge how they balance against the benefits of the arXiv. Anyways, clearly this discussion is not being productive, (and anyways, Arend basically summed up my feelings above better than I could manage to) so I’m quitting on this thread before I say something stupid(er).

  69. Thanks for the clarification Ben. Sorry for over-reacting.

    Honestly, I don’t know how serious the problem is. I know of a few examples, but none from my own work.

  70. Regarding Pace Nielsen comment (67) on RH and vetting process: First, while I am not even completely sure which paper you mean, I am almost sure which it is; yet in general I would prefer that my comment is read abstractly (there are a couple of instances like this I saw from a distance), and not as one on a specific paper or even person that I do not know. Yet, if whoever wants to use the arXiv or also another venue (actually I’d say essentially whatever venue) to get first feedback on a quite extraordinary claim (or I would even think so for great but less extraordinary claims), then personally I do not understand why this person is not very explict about this. There are people who put preliminary versions of, let’s call them, everyday-papers on the arXiv with a comment like ‘preliminary version, comments welcome’ and in exceptional circumstances I’d say one could even consider an explicit disclaimer along the lines ‘I am aware this is an extraordinary claim, and chance are that despite my best efforts there is a problem with the argument I did not see.’ Yet, for a couple of such extraordinary claims I saw from a distance, I did not have the impression that something like this was done. By contrast, my impression rather was that the person intended to present the result with a splash.

  71. an_mo_user,

    Concerning your suggested disclaimer sentence: What would it’s purpose be? In the case in question, the paper was not a preliminary version, but a finished draft. Unfortunately, the paper also contained a hole and had not been vetted. Every paper may “despite [our] best efforts” have a “problem with the argument [we] did not see.” Adding such a sentence does not distinguish between competent mathematicians and cranks. It seems an unnecessary burden to simply demonstrate (perhaps unfelt) humility.

    Further, adding such a disclaimer would do nothing to mitigate the “splash” effect, nor the negative publicity from not vetting it elsewhere first.

  72. Pace Nielsen,

    of course adding this is merely a matter of style, and has little or nothing to do with crank or competent. But, in view of your parenthetical assertion, I am surprised that you seem to not see my point. You can read my comment as the assertion that I do not understand this, in my opinion, lack of humility.

    It is true it likely would not mitigate the splash-effect, yet why you are so convinced it would have no effect on negative publicity is not clear to me. I am convinced it would have an influenced on my — which, if I am the only one with that opinion, is admittedly totally irrelevant — perception of such a matter.

  73. If I’m reading you correctly, one of the reasons you would like authors to add such a sentence is so it would be easier for you, as the reader, to see that they are humble in presenting their work to be reviewed by the world. Furthermore, that many of the negative perceptions (in the abstract case you present) are related to perceiving a lack of humility.

    If this is correct, I would say that your sentence would probably achieve the outcome you want. Such a sentence would give a perception of humility. But I think a better measure of true humility is in the response the author gives to feedback. Or, in other words, perhaps we should reserve judgment about whether an author is open to criticism and correction until we see their response when such criticism is presented to them.

  74. Pace Nielsen,

    First, I am in complete agreement with your statement that the reaction to (critical) feedback is considerably more important. And, yes, in the end, if this is handled well by the author, this should be considered as sufficient and it is certainly sufficient for me. [Sorry, this is a possibly strange formulation, but I don’t find a better one.]

    The ‘disclaimer’ perhaps could serve two purposes:
    On the one hand, it could be taken as a signal of a critical distance to ones own work and thus as a preemptive sign that if it should become relevant the author would react well to criticism.
    On the other hand, yes, negative publicity you mention is about perception of a person and/or situation, so how the presentation is carried out has, I believe, some influence.

    To give a not too fitting analogy: if a group of people tries to do whatever, say, change a wheel of a car and fail, and then somebody else shows up to participate in the effort (and in the end also fails) after which introducory phrase will the reactions of the others be better towards this failed attempt:
    a. Everybody else step back, I will now show you how to do this right in 30 seconds.
    b. I believe I know how to do this, perhaps let me give it a try.

    In some sense it is irrelevant whether the person said a. or b., and it is only of practical relevance whether, say, the car does (or does not) get damaged in the process or the person is willing to give up the try after a reasonable time to let still somebody else try [the reaction to failure]. Still, I’d say it can make a difference to how the person is perceived whether a. or b. was said at the start.

    Finally, my suggestion of the disclaimer was prompted by your remark that an author thought something was the vetting process and others thought otherwise. So, why not make (in exceptionally critical cases) explicit what is the status of the preprint: draft; final from aothor’s point of view, yet not checked by anybody else; final form au. p.o.v., circulated among selected experts;…

    Perhaps this does not change anything at all, but maybe it would have a small influence on certain aspects. At least I would take a preprint claiming a proof of a very important result more seriously if it came with information and/or disclaimers along these lines.

    Completely finally, personally, first I tend to forget who the actual person involved in a failed attempt of a famous result was pretty quickly (even if I still remember the incident itself); and second if I have any thoughts regarding the (to me unknown) individual behind the proof at all it is mainly that I think it has to be really difficult to deal with a situation where one thought to have achieved something truly great and then it falls all apart.

  75. Arxiv is an archival document server. Its purpose is to ensure that publications and reports remain accessible and citable, nothing more. It’s not a paper review site, or a discussion site, or a rating site. By design, there is only minimal quality control.

    Please don’t use Arxiv as part of a crusade for truth and justice on the Internet. If you are concerned about the veracity of what’s on Arxiv, then simply assume that everything on it is false, all the people who post papers on it are cooks and cranks, and get on with your life. That way everybody will be happier.

  76. “At least I would take a preprint claiming a proof of a very important result more seriously”

    If you are a specialist in a field, you should be able to determine quickly yourself whether you should take an important result seriously. If you are not a specialist, just wait for the peer review.

    It’s called a “preprint” because it’s printed before the peer review, so if you require everything to have peer review, then you’d be outlawing preprint servers.

  77. Imagine X (a noob) submits a paper to ArXiV. It is its first submission and he is not affiliated to an University that provides automatic endorsement; therefore X is asked for endorsement.

    X contacts with a renowned expert in the field (who got his PhD with a famous Nobel winner, has hundred of papers, books and monographs published by Wiley, Springer, Dover classics… scientific coordinator of European scientific network…), who reads the paper, discusses it with X and finally decides to endorse X. Then X submits the paper to ArXiv and when ready to be announced, a moderator stops the whole process, deletes the submission, and asks X to not submit any paper more before answering a series of questions done by the moderator.

    The questions include a list of previous publications by X and extra information about affiliations.

    A perplexed X contacts with expert for advice and this says him that moderator actions are totally unacceptable, that those requirements are nowhere at ArXiV, and suggests to send a letter to Cornell head.

    I would add that moderator own work, in computational sciences, is totally unrelated with the field of the paper.

    What are your suggestions?

  78. It sounds like a frustrating situation, but it’s hard to know what to make of it based on the information presented. For example, Brian Josephson is a famous scientist (he even has a Nobel prize), but his current views on science are very far outside the mainstream, and I don’t think his endorsement would or should count for much. If the expert you have found is someone controversial, or someone who has a history of endorsing controversial arXiv submissions, then I would not be surprised to see the moderators decide to investigate. The same would be true if your paper appears to be very unusual or implausible, regardless of who has endorsed it.

    I’d suggest answering the moderator’s questions and seeing where the process leads. I do not think sending a letter to Cornell would help you achieve your goals (it is extremely unlikely to change the general policies or to help with your particular case).

  79. Thank you Henry,

    If the endorser was someone controversial then administrators could cancel endorsement and ask for the search of some other endorser.

    Administrators do not ask for new endorsement, but accept the endorsement status of X.

    After of some email correspondence a new moderator ask only for a list of publications only (the other requirement vanished). Maybe I would provide such list, but it would be fair to know where at ArXiV policies is established that moderators can stop a submission and ask for author publications.

Comments are closed.