It is a common and acceptable practice to present a preliminary version of a paper at a conference and then to submit the full paper to a refereed journal. The Foreword of a recent STOC proceedings states, for example,
The submissions were not refereed, and many of these papers represent reports of continuing research. It is expected that most of them will appear in a more polished and complete form in scientific journals.That is the ideal. In practice, many of us have been in the habit of putting complete versions of accepted papers in conference proceedings. This is not a wise strategy. Assuming that the publisher of the conference proceedings holds copyright of the papers that appears in the proceedings, the paper that one submits to a refereed journal must be substantially different from the one that appears in a conference proceeding in order to avoid copyright infringement. I asked my contact at Springer, the publisher of the journal Theory of Computing Systems, how different the journal version of a paper should be from the conference version, and was told to use 25% as an acceptable guideline. This is a reasonable response that should not too difficult to manage for theorists, and can be taken up by proofs and explanations that were not included in the conference version. I should add however, that Lance Fortnow asked the same question of IEEE, the copyright holder for papers that appear in the Conference on Computational Complexity, and the response was far more onerous. Lance's contact at IEEE called for 30% to 70% new material, and material from the original paper should be there only to support new ideas. I would rather abide by Springer's recommendation than IEEE's. The essential point is that the journal version must be essentially different from the conference version.
Don't most journals anyway have the requirement that the journal version be substantially different from the conference version (I mean to maintain quality of the journal and to avoid having people "pad" their publication counts, not due to any legal issues involving copyrights)? At least this was the case for some ACM journals I have submitted to in the past.ReplyDelete
I think anything over 25% is an unreasonableReplyDelete
goal for many papers in STOC/FOCS/SODA. When
most people in our field are lax in submitting
journal versions, the advice to "hold back"
in the conference version does not do the
community any service. Shouldn't we be
worrying more about disseminating our work
in the best possible way instead of constantly
being straight jacketed by the for-profit
So how is one supposed to publish an eight-page paper? Should we artificially elide details for the conference version, avoid conferences altogether and submit directly to a journal, or submit only to a conference and forego the benefits of proper refereeing?ReplyDelete
Couldn't we avoid the copyright ingringement issue by declaring our conference submissions to be in the public domain? Nothing would then legally forbid the proceedings and journal publishers from publishing identical work.
If disseminating your work and getting proper refereeing is the goal, these are easy to achieve by (1) posting your paper on the web (at ECCC, for example) and then (2) sending to a journal. Since I suspect most of us would find this objectionable, let's at least not pretend that these are our only concerns...ReplyDelete
I believe that conferences are better ways toReplyDelete
disseminate ones work than simply posting the
paper on ECCC and then sending it to a journal
that takes a looong time to publish the paper.
And sure there are other reasons why we want
to have conference papers - we are part of a
system and we got to make a living. No one is
pretending that this is not there in the background
but making superficial changes for the sake of
arbitrarily set journal guidelines doesn't make
sense. The essence of most scientific papers is
in the ideas and simple text based similarity
measures don't have a chance of capturing the
overlap of papers.
First, everyone should read Jeff's insightful analysis over at 3d Pancakes! Now, after that...ReplyDelete
The system, as I see it, is currently broken. My starting point is determining the difference between conference and journal publications. In the old days, conference publications were less accessible, and arguably less carefully archived, than journals, so this was a distinguishing characteristic. Not so in the Internet age. Another possible distinguishing characteristic is prestige. This is arguable -- these days, a conference publication (in computer science) can be as or more prestigious than a journal publication. The only remaining distinguishing characteristic I see is the refereeing process. Conference articles (no matter how much we might like otherwise) are not refereed; journal articles supposedly are. This is really the service that journals are now providing. (No wonder so many believe that journals in their current form are destined to disappear -- it's not clear they're offering that much for the price...)
If refereeing is the distinguishing characteristic, then there should be no compelling academic reason not to submit a conference paper verbatim to a journal. With good refereeing, appropriate changes will be made before the paper is published in journal form, and the two works will necessarily be somewhat different.
There may be a compelling legal reason not to submit a conference paper verbatim to a journal -- copyright. I agree with Jeff, that as much as possible, we should try to prevent conferences and journals from copyrighting our work -- they should be disseminators of information, not owners of copyright, and copyright should reside with the authors. But that is not the current system. Like most others, I ignore the issue as much as possible.
I was once extremely ticked off by an editor who took several months to tell me that my journal submission was too similar to my conference paper, and could not be considered, even though both the conference and the journal were IEEE publications. This just shows what happens when the system is broken and mindless rules start to apply. I see no compelling academic reason for such an attitude, and since the IEEE were the ostensible copyright owners in both cases, there should not have been a legal reason either. But this is the sort of stupidity we've opened ourselves up to by having the publishers establish a non-sensible set of rules...
This comment has been removed by a blog administrator.ReplyDelete
Wow. If any publisher actually tried to enforce such rules, my response would be simple: "Go @#!$& yourself. From now on I won't submit to your conferences or journals."ReplyDelete
Putting aside the copyright issues, I think you are all missing a broader point: do we really want multiple copies of essentially the same paper being published? If a result takes 8 pages to prove in complete detail, then why do we need to see essentially the same paper in both STOC and SIAM J Computing? Yes, refereeing can be important for some papers but not necessarily for all.ReplyDelete
With regard to an anonymous comment above -- I do not mind at all having 2 copies of essentially the same paper around, provided one is a conference version and one is a journal version.ReplyDelete
A conference version is essentially unrefereed; you can count on the program committee having looked it over, but little more. This is just the way program committees are run; there is nothing wrong with that, it is just the nature of the beast. In this sense, a conference publication should be treated as "preliminary" (I hesitate to say "a draft") in most regards.
A journal version supposedly has been vetted, generally by at least two referees, often three, and/or considered by an editor. It is of a different type completely; there is therefore no problem having a conference version and a similar journal version.
If a journal paper ends up being almost exactly the same as the conference version, that means that the author did a great job on the conference version, and that the referees did not find that anything substantial needed to be changed or added. More power to the author -- great job! -- which is properly acknowledged by acceptance to the journal. (An alternative interpretation is that the journal reviewers aren't doing their job properly -- in which case, that's a problem the editor needs to deal with.)
In any case, to summarize my response, no, I don't see any problem with having two versions of a paper, conference and journal, that are substantially the same when that is the natural outcome. Since I feel that you didn't make it clear, perhaps you might explain more fully why do you have a problem with it?
In response to Michael's question (I was the anonymous poster from before) I think the main problem with multiple publications of the same paper is that it seems to me like the author(s) are simply using this as a way to pad their CVs. Why do I care? I guess fundamentally it doesn't affect me one way or the other, but it bothers me for the same reason seeing weak papers (or a stronger paper broken into multiple "least publishable units") bothers me. And I don't think it's a simple matter of the author being "rewarded" by having such a perfect proof that the referees couldn't find a mistake: for the most part, I think authors *know* when a result is proved in full or not, and if something can be proved in-full in 8 conference pages and not require further elaboration then the authors must know that referees are not going to find a mistake. So why not reserve journals for papers that truly require more difficult arguments --- that need to be independently verified --- or require additional elaboration beyond what a conference version allows.ReplyDelete
Re-thinking things a bit, I guess like anything else the process can be abused. So, I have no problem with a strong/important result appearing as identical conference and journal versions. My issue lies more with authors of weaker papers trying to do the same...ReplyDelete
That's my 6 cents on the topic!
In response again to the anonymous commenter:ReplyDelete
I understand and agree with the issue of CV abuse. I think Jeff highlights this issue in his comments on 3D pancakes. This is certainly a fair concern. My responses to this concern would be 1) weak papers can (and should) still be rejected; 2) as you comment, like anything else, the process can be abused.
I hope, by the way, I'm not guilty of CV abuse -- I generally list conference and journal versions separately, but try to make clear when a journal version is the "derivation" of a conference version. Though I don't attempt to quantify the newness in the CV...
I'd also like to slightly disagree with the comment:
"...I think authors *know* when a result is proved in full or not, and if something can be proved in-full in 8 conference pages and not require further elaboration then the authors must know that referees are not going to find a mistake. So why not reserve journals for papers that truly require more difficult arguments --- that need to be independently verified --- or require additional elaboration beyond what a conference version allows."
1) This seems to disagree with my assumption of what a journal is for -- to offer a refereeing process and serve as a stamp of approval for the paper's arguments. You seem to be suggesting that only "difficult arguments" -- the working definition being arguments that don't fit into 10 conference pages -- should be given an independent verification and such a stamp of approval, and I disagree with that. I guess I also believe that a lot of conference results need some tweaking, and I'd hate to have the assumption be that a conference publication is sufficient independent verification for a result.
2) I also believe that except in extremely rare cases, the reviewing process should add to a conference paper. I've written many reviews that say roughly, "This is a nice result, and was a nice conference paper. It should eventually be accepted, but the authors should really do the following to improve the paper for the journal version:..." I've also gotten reviews of a similar flavor. That's one of the benefits of the reviewing/journal process; the journal paper is the "final form", and the reviewing process should improve it. It's rare I see a conference paper that shouldn't be improved in at least some ways before it makes it to a journal.
By the way, I'd like to thank the anonymous poster for interesting discussion!
I seems that the whole confusion is created by the current trendReplyDelete
of computer science. In a way we are saying that journal papers
are most important because of the refereeing process. On the other
hand, it is given very high importance to conference acceptance.
This leads to people being forced to write conference papers
that are basically in final version. Then, there is not much to improve
for the journal version, beyond trivial extended exposition.
In my view, conference papers should be really extended abstracts,
with just enough to show that the result is correct -- sometimes this would even be a favor to the conference program.
The current version of the ACM copyright policy is quite explicit about the 25% difference standard for republication. However, the policy does not distinguish conference from journal publication and this is misguided. Even if there were nothing changed there is a significant value to the fact that a refereeing process has taken place for a journal version.ReplyDelete
The prime goal of 'copyright' for conference publication should be to guarantee access to the work in perpetuity.
Here is a proposal for conference publication 'copyright' (or license): The sponsoring organization (e.g. ACM, IEEE, SIAM, etc) retains the right to reprint the work as it sees fit (including digital media) including charging for it but must be explicit about what conference it is republished from. This organization is free to license others to disseminate this content unaltered as well. However this license is not exclusive and does not extend to preventing authors from disseminating the content in any other way they see fit.
I'm having to deal with a strange situation. I presented our work at a conference and thereafter had the material accepted in a journal. One of my collaborators and a co-author in both papers, without letting me, the corresponding author know, approached the journal editor and told him that the 2 versions are very similar and that there might be a violation of copyright. Actually the 2 versions DO differ but not substantially. What bothers me, other than the sneakiness aspect, is that there is not much I could have written differently because the basic material i.e. the ideas and analysis, is essentially the same. How do I defend myself to the publsiher?ReplyDelete
I have a question. How can conferences claim copyright to one's papers if they haven't given consent?ReplyDelete
You have to sign over some publication rights before a journal or conference will publish your paper. See the ACM form for an example.Delete