23

I am recently getting my papers rejected from journals (i.e., TALG) on the mere basis of not having a significant difference between the journal and proceedings (i.e., SODA) version.

The main reasons for me to submit to a journal is its thorough review process. Other than that, SODA's 20 page limit is more than enough for all that I want to say. In fact, David Johnson has repeatedly asked the SODA crowd to not "save stuff for the journal version".

Any advice?

Kaveh
  • 21,577
  • 8
  • 82
  • 183
Algorithm
  • 231
  • 1
  • 2
  • 7
    I am quite annoyed by this policy of additional X% in a journal version because, I think, conferences and journals should provide different services: announcement of a result vs a thorough check of correctness. At some point, I considered sending to conference a final version with 2 pages even when the full version (~15 pages) was already available. – someone Nov 27 '11 at 14:59
  • 3
    Can you provide more detail without breaking anonymity/confidence? Did referees recommend rejection, or was the rejection instigated by the editor? If referees are recommending rejection without novel content, you're hitting a real community value, not just a journal policy; your best best is to either add new material, or to ignore David Johnson and abbreviate your future SODA papers. If rejection was not recommended by the referees, then a discussion between the TALG editorial board and the SODA steering committee is in order! – Jeffε Nov 27 '11 at 15:06
  • 3
    Also, I agree with @someone that a "X% new material" journal policy is stupid. – Jeffε Nov 27 '11 at 15:09
  • 5
    There is a trivial solution: Just re-write your paper for the journal version. 100% new text, 0% new results ≈ sufficiently different. Re-writing from scratch almost always helps a lot with the presentation. – Jukka Suomela Nov 27 '11 at 21:01
  • how about another trivial solution: write the proofs in the conference paper with "(extended abstract)" written after the title, and write all proofs in a very sketchy/informal way. Maybe that won't work with SODA because 20pages double column is in a lot of cases enough. – Marcos Villagra Nov 28 '11 at 10:40
  • 2
    We don't like duplicates, be it here in cstheory or TALG. If you have already published a clean and self-contained version of your paper in SODA, there is no point resubmitting to a journal for the sake of going through a thorough review. Today conference papers are as accessible as journal papers. Also, what you publish in conferences must be correct and on the other hand, the mere fact that a result gets published in a journal does not by itself add to its merits or make it bulletproof. Just spend the extra time producing new research. – Mahdi Cheraghchi Nov 28 '11 at 19:42
  • 9
    MCH: It is often the case that what is published in conferences includes bugs that are found and repaired in journal refereeing. That improvement to the literature benefits us all, but is unavailable to those like the original poster (and he or she's not the only one I've heard of in this predicament) who are shut out of journal publication by having made the "mistake" of putting full details in their conference versions. Conference reviewing is not an adequate substitute: it focuses on interestingness rather than accuracy and often has no process for ensuring that authors fix their papers. – David Eppstein Nov 28 '11 at 22:30
  • 4
    @MCH The institutions using ISI rankings (e.g. most 3rd world countries where the government feel they need a measure more "objective" that asking peers to congratulate or not their scientists) refuse to consider conference papers as publications: in this case, it is not the choice of the author to decide if a SODA publication is "enough" or not. – J..y B..y Nov 30 '11 at 00:01

1 Answers1

1

It depends. If you write a paper and exhaustively include all of the relevant materials, then you should submit the paper to a journal (if you think the journal you choose is more valuable then a conference on the basis of impact factor, reputation and other metrics). If you only describe part of the work, then it may be better to submit to a conference and later, when you have new/updated results, then you may consider submitting again to a journal. What constitutes a significant difference is of course highly variable and depends on your particular work. A very rough rule of thumb is that journal versions differ for at least 30% from corresponding conference versions. However, there are exceptions. And, finally, you should carefully decide the venue. Especially in TCS, a conference may be extremely valuable, in some cases even more than a journal. I do not think that the review process applied in some highly valued conference is worse than the average review process applied in journals. This is strictly dependent on the particular conference or journal, and on the reviewers an editor finds available.

Massimo Cafaro
  • 2,216
  • 20
  • 19
  • 1
    "A very rough rule of thumb is that journal versions differ for at least 30% from corresponding conference versions." — This is not a universal rule of thumb; the fraction of required new material is highly field-dependent. In theoretical computer science (STOC/FOCS/SODA/SOCG/etc), that fraction has historically been indistinguishable from 0%. In other fields (economics, for example), the fraction is much closer to 100%. – Jeffε Nov 27 '11 at 15:19
  • Indeed, I stated immediately after: "However, there are exceptions". – Massimo Cafaro Nov 27 '11 at 15:22
  • 6
    I don't think an entire field of study should be considered an "exception", especially in an answer to a question about publishing in that field, posted in a forum devoted to that field. – Jeffε Nov 27 '11 at 15:25
  • 2
    @Jeffe: I did not want to start a debate about this, and I do not want to start an harsh debate right now. Simply stated, I do not think (this is of course my humble opinion) that we are talking, as you said, of an entire field of study. Please note that, coincidentally, even Dave Clarke stated the same 30% percent. This is (merely) our experience. I understand of course that yours may be different. This forum is a wonderful place that gathers so many interesting but different experiences: the key point, at least to me, is that I can learn from many different people. – Massimo Cafaro Nov 27 '11 at 15:42
  • @JɛffE: I don't want to start another debate in the comments, but does that mean we should always publish a journal version even though we have nothing new to say (w.r.t. the conference paper)? – Anthony Labarre Nov 27 '11 at 21:02
  • 2
    @AnthonyLabarre it means that if the conference version is an extended abstract, does not have full proofs or can benefit from better presentation, you can fix these issues and submit to a journal. Of course you can do that with ArXiv, but a journal provides a review of the paper in its full form. – Sasho Nikolov Nov 27 '11 at 23:31