35

I submitted a paper to a top tier conference and it did not get accepted. The paper contains an original idea, and I am afraid that the paper could get plagiarized between the time I send it to another conference after incorporating reviewer's suggestions and making some modifications. Such things have happened in the past to a few students in my department.

I had considered posting the paper on arxiv as a means of timestamping before sending it to the aforementioned conference, but didn't, as this would have violated the double-blind review requirements of the conference. (The conference chair replied to my query as follows:)

The submission of a paper in the arxhiv.org system is not strictly speaking a double-submission, but rather a violation of a double-blind submission required in this conference, as the reviewers will be able to find your name (and those of the coauthors) if they do a simple search in Google. Thus, I am quite certain it will be rejected, and I do not consider it an acceptable practice in this conference's submission to be fair with all the submissions in order to handle them equally during the review process by the reviewers.

Hence I had not posted it on arxiv. Now this paper has been rejected and I would like to timestamp this paper now, while still being able to send a modified version to some conferences in the future (including those that follow a double-blind review).

Is there a mechanism by which I can timestamp my paper that does not violate the blind review requirements?

Edit: The paper was in computer science, if that matters. A related question "How can I time-stamp my data without publishing it?" asks about ways to timestamp while keeping the information in the paper private or hidden to avoid plagiarism. However my question is about ways to timestamp such that it is still possible to publish the information while not violating blind review.

  • 30
    The argument from the PC Chair was bad. Even without posting it on arxiv, it's usually relatively easy to find who is the author of a blind submission using Google. The blind submission process should mean that the reviewer cannot guess the author just by looking at the paper. – user102 Nov 2 '14 at 8:20
  • 6
    Who are you afraid will plagiarize you? The conference organisers can definitely certify you submitted that paper at a certain date. – Davidmh Nov 2 '14 at 8:31
  • 7
    Most conferences don't make a fuss about posting papers on arxiv, even if they have double-blind review. I think this particular conference's reply is unusual. As others have said already, finding the real authors is easy if you really want to. Some universities, like mine for instance, require posting tech reports on university pages prior to any submission (unless there is an embargo in place for some major journal). – Marc Claesen Nov 2 '14 at 9:48
  • 1
    @NehaKaranjkar Since you are probably talking about CS, very few extra-top conference have double bind reviews. This work has already been rejected once. Can't you submit to a lesser (still A) conference that does not require double-bind review? – Alexandros Nov 2 '14 at 12:01
  • 21
    Peer reviewers in your field generally reject papers if they figure out who the author is? I have never heard of such a practice. – Sverre Nov 2 '14 at 15:05
28

The most typical way to time-stamp a piece of work is to place it in a public repository: arXiv is a good example when it applies; many institutions also have a technical memo or technical report system in place. Although this may technically violate certain interpretations of double-blind submissions, as noted in the comments blinding of authorship is pretty iffy in many cases. Moreover, in practice this will only be violated if the reviewers actively go looking, particularly if you put it in an institutional repository rather than a global one like arXiv. I would thus advise that if you are worried about date-stamping, just do it and let the double-blind nitpickers complain if they even notice.

One other consideration: certain publications (particularly certain journals) do have a policy against accepting material that is already available online. While this is in my view an insane policy, if you find yourself dealing with such, one way to handle with it is to put up a shortened version of the work, i.e., extended abstract, as the date-stamped pre-print.

  • 10
    "The only way to time-stamp a piece of work is to place it in a public repository" I don't have any real life experience here, but I wonder if that is actually true. The history of mathematics is replete with stories about sealed envelopes in the French Academy of Sciences, for instance. Or you could electronically publish an encrypted version of the article, if you really wanted to... – Pete L. Clark Nov 2 '14 at 13:34
  • 12
    @Pete: It is definitely not the only way. Cryptography certainly offers ways to prove that you possess information without revealing, until later, what that information is. See stackoverflow.com/questions/6726382/… for much more. – Nate Eldredge Nov 2 '14 at 13:39
  • 3
    @jakebeal why don't you recommend it? If it helps making OP feel safer, I don't see any arguments against using an online timestamping service. There is no need to disclose it to editors/reviewers; it is completely anonymous unless the author decides to disclose it in case of plagiarism. – Federico Poloni Nov 2 '14 at 22:59
  • 5
    @jakebeal: Why would it make people think you have a bad attitude, since you don't even have to tell anyone you're doing it? – Nate Eldredge Nov 12 '14 at 21:41
  • 2
    @NehaKaranjkar: Note that a reviewer could presumably also recognize your paper from an extended abstract, so that may not solve the present problem. – Nate Eldredge Nov 12 '14 at 21:42
8

Are you really afraid that it will be plagiarized? The only people who have seen your work are the programme committee and any reviewers who saw the paper when you first submitted it. It would be very surprising if the programme committee or reviewers of what you describe as a "top-tier conference" rejected a paper and then plagiarized it.

Or are you worried that you'll be scooped, i.e., that somebody else have similar but independent work accepted by another conference or journal before you manage to get your paper published? If so, the fact that you already submitted it to a conference should be enough to demonstrate your priority. It's not just you and your co-authors who are claiming that your paper already existed in October 2014 but also the programme committee of the conference can confirm this.

  • 1
    There are other examples in which people have submitted the paper, only to find a publication, with their verbiage, but different notation, on a different website, journal, or book. Some have even gone so far as to present other people's material at a conference. Several "related questions" are on this topic. – ABrown Jan 15 '15 at 20:54
  • 5
    @ABrown Yes, it occasionally happens but it is very rare, especially at top-tier conferences. Of all the things to worry about, the referees rejecting the work so they can steal it is very low on the list. – David Richerby Jan 15 '15 at 22:19
4

Here's a version that focuses on

  1. making the proof watertight for legal matters
  2. not publicly revealing the content
    (which follows from the double-blind requirement)

The legally provable way of time-stamping something (and also to legally prove the state it was in at that time) is to have a notary attest it, and/or give a copy to the notary for keeping (which allows to prove that the content was not tampered with afterwards). This kind of service is the main business and duty of a notary.

The "poor man's version" of a sealed and unopened post-stamped envelope is much cheaper, but unlike the notary attest there is no legally binding guarantee that a court would accept this.

As has been pointed out, making the content public, preferrably with a time stamp (aka publishing (!)) is the yet cheaper alternative - however IMHO it is not compatible with the double-blind requirement of the question: It is not possible to have the content + your identity + a timestamp publicly (which are the 3 pieces of information you want to connect) available to everyone but the blinded reviewers.


As for cryptographic methods, at the moment I'd consider them at a similar level to the post-stamped envelope. The legal weight of this is quite unclear to me -- if you're really concerned I'd say an encrypted email exchange with someone who states "I got this email at this and that date" is at the same level as having someone testify that they saw or got the content from you at the stated date.

I'd imagine that a notary could do the encrypted signing, but AFAIK this is a techonolgy that yet has to fully emerge:

  • There has been a project called "Datennotar" (data notary) at the Fraunhofer FOKUS and the University Kiel - final reports in German are available at ISPRAT (funder).
    AFAIK there are no data notaries so far in Germany.
  • Wikipedia says such services exist in the US. A quick search brought up some companies, but the first two web pages I tried at least to me looked incomplete in some important details: one has a contact page without any kind of address, the second has email and phone only. Over here in the EU this wouldn't be legal for any kind of business ...
  • I think this answer somewhat undersells the cryptographic methods. One avenue is to produce a pdf of the document, take a strong cryptographic hash of the file, and make that hash public in a way that gets timestamped by some trustable third party. The legal weight is probably comparable to a post-stamped envelope, and the academic acceptance will probably increase the closer the field is to CS, but there's plenty of DIY options one can consider that may well be sufficient. – E.P. Sep 10 '18 at 19:37
  • (My apologies for butting into an old thread, though.) – E.P. Sep 10 '18 at 19:37
2

A partial workaround that was often used in the previous lab where I worked was to tamper a bit the article you put on arXiv by changing the title and the abstract, or even publishing an incomplete or old version of the article to also tamper the full-text.

This is surely not foolproof as reviewers can still find the article if they really want to do so, but at least you honestly tried to avoid violating the double blind process while still retaining guarantees of authorship.

And personally, I'm inclined to say that if reviewers still find the article with these precautions, then they are to blame for violating the double-blind process, not you.

  • 2
    People can very easily find articles on ArXiv without actively seeking them out. If you subscribe to the mailing list for a topic, you get notified of all new articles in that area. – David Richerby Nov 19 '14 at 23:23
  • @DavidRicherby maybe, but I still think this is unlikely without an active search: first even if you're notified, there are so many papers everyday even within a single topic that it's unlikely anyone read them all, even less probability for one person in thousands (your reviewer). Secondly if you tampered your arXiv article, it still would be pretty hard to establish a connection between the article you're reviewing with this one arXiv paper you glanced in the past, and furthermore, even if you remember some of the text, you certainly won't remember the authors, thus not breaking anonymity. – gaborous Nov 20 '14 at 1:53
  • 1
    Remember the conference chair argument: "The submission of a paper in arxiv.org [...] is a violation of a double-blind submission required in this conference, as the reviewers will be able to find your name (and those of the coauthors) if they do a simple search in Google.". It seems that the concern here is about easy access to authors names for a given paper. It doesn't matter that they've read some version of it in the past as long as they can't easily find the authors names of the article they're reviewing. A tampered article probably clears most of the concern for the reasons above. – gaborous Nov 20 '14 at 2:00
  • 2
    One major problem with this is that it compromises the usefulness of the arXiv. I would be angry if I read a paper on the arXiv and later learned that the author had deliberately posted an incomplete or out of date version, and even deliberately changing the title would really annoy me. – Anonymous Mathematician Mar 7 '16 at 5:10
1

You can make a copy of your paper, seal it in a brown envelope and go to the post office, have the envelope hand stamped and mailed back to you. Do NOT open your the envelope but keep it in a safe place. Then if you ever need it, you can bring the sealed envelope that has been postmarked with the earlier date which is your proof the original material was yours.

  • 2
    While this is certainly an interesting idea, I doubt that it would be reliable or convenient in any potential legal scenario. – user3209815 Mar 25 '15 at 15:44
  • @cbeleites I think your comment above (notary attest the paper (and maybe keep it) is worth an answer. Why not post it as such? I think it's better than putting the paper on arXiv. Because putting it on arXiv risk violating the double Blind process. – scaaahu Apr 11 '15 at 6:08
  • @scaaahu: done. – cbeleites supports Monica Apr 11 '15 at 12:45
0

The informatics way to do that would either be a digital signature with a timestamp. Such a thing exists e.g. for PDF files, but there are some disadvantages:

  • you need to obtain a digital certificate, which is expensive
  • you need to know how to digitally sign the document
  • you need to understand how timestamping works and enable it

A simpler approach is a cryptographically secure hash of your paper. If you don't exactly know, which one to choose, simple choose all of them. There are free tools to calculate the hashes. On Windows, try HashCheck.

In the next step you publish those hashes on a website. Twitter is quite common, Pastebin should work as well, maybe also a website of your university. After that, you archive that website, e.g. on Archive.org.

Whenever someone wants to verify whether you wrote the article first, you can proof by letting him recalculate the hash on (a copy of) your file and compare to the hash you published earlier.

Make sure you never modify that version of the paper. Not a single byte.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.