34

Disclaimer: I can only vouch for my research fields, namely formal methods, semantics and programming language theory. The situation is possibly different in other parts of the discipline.

It seems that TCS has become rather conference-oriented. Researchers aim at publishing in the next conference. Sometimes a journal version appears. Sometimes it doesn't.

In other disciplines (biology, mathematics, and most others I guess) this is unheard of. The effort put into writing the conference papers is a lot lesser, but in turn, the conference papers count a lot less. The "real deal" is the journal publication.

Arguing whether this situation is good or bad could result in a flame war, and doesn't have a precise answer. Instead, let's try a more factual question:

How did we become so conference-oriented? How did conference papers gain so much weight?

Ohad Kammar
  • 2,677
  • 23
  • 28
  • 8
    This question is asked quite often. See, for example, http://blog.computationalcomplexity.org/2009/07/time-for-computer-science-to-grow-up.html – Stasys Jul 20 '11 at 21:26
  • 16
    Airplanes http://blog.computationalcomplexity.org/2004/07/why-are-cs-conferences-so-important.html – Lance Fortnow Jul 20 '11 at 21:34
  • 3
    Thanks Lance, that was an interesting article! I am not convinced about airplanes being the only factor, though. If you look at other new, post-jet, research disciplines --- are they also conference based rather than journal based? – Ohad Kammar Jul 20 '11 at 22:02
  • 1
    @Ohad, if you look at a list of typical undergraduate majors (as representative of "broad" or "large" research disciplines), which disciplines are you referring to? Computer science is really fairly young/recent compared to other options like biology, economics, or math! – Daniel Apon Jul 21 '11 at 03:05
  • 2
    Sure. Here are some examples for new research disciplines: Neuroscience ('60s), environmental studies (independent since 1965, I think), sustainable development (relatively new, I think from the mid '80s). Are these fields conference oriented rather than journal oriented? – Ohad Kammar Jul 21 '11 at 10:35
  • I am not sure how this is true for TCS, but in general I guess that people feel pressure to publish often and it is more work to prepare a (good) journal article than a conference article (based on what people have told me). – Raphael Jul 21 '11 at 11:22
  • 1
    @Ohad: As far as I understand, CS is the only conference-oriented field. It's possible that the other relatively new fields you list weren't subject to the airplane effect because they started as offshoots of more established (and journal-based) disciplines. In some ways CS grew out of logic and EE, but in many ways I think it was more independent of these "parent" fields than the other examples you listed. The idea of CS as the rebellious child fits nicely with another relevant metaphor: http://cacm.acm.org/magazines/2009/8/34492-viewpoint-time-for-computer-science-to-grow-up/fulltext – Joshua Grochow Jul 21 '11 at 22:15
  • 1
    Can these narratives be backed by facts? E.g., particularly rebellious researchers leading the field, etc. We seem to agree that CS is unique in this respect, so what are the unique causes? – Ohad Kammar Jul 22 '11 at 00:32
  • Seconding Raphael's comment, the (much!) faster reviewing process of conferences certainly plays an important role in that perspective. – Anthony Labarre Jul 29 '11 at 15:15
  • Another issue to (maybe) take into account: TCS is very productive and it seems easier to produce more results than in other disciplines such as mathematics and biology – Carlos Linares López Apr 29 '12 at 19:55
  • 1
    Why is that true? How can I evaluate the truthfulness of this claim? – Ohad Kammar Apr 30 '12 at 09:22
  • 4
    I'm not convinced that CS is unique. My impression is that engineering fields are quite often conference-oriented. Certainly, looking online at calls for papers in power engineering conferences, they seem to be highly selective, something you don't often find in non-conference-oriented disciplines. Since CS has feet in both science and engineering, maybe its being conference-oriented shouldn't be too big a surprise. – Peter Shor Jan 23 '13 at 17:08

2 Answers2

15

While I don't know the answer to this question, it seems crucial that the phenomenon isn't limited at all to theoretical computer science. I believe SIGGRAPH plays the same sort of role for graphics, NIPS for machine learning, ISCA for architecture, etc. that STOC and FOCS play for theory. Yet it's true that the emphasis on publication in conference proceedings is a striking feature of computer science as a whole, one that isn't shared by any other academic field that I know about. (But maybe there are other such fields?)

Scott Aaronson
  • 13,733
  • 2
  • 62
  • 68
7

Well, I can only guess, but for me this seems to be some sort of tradition. If you start looking up how the ACM came into existence it all started with a lot of informality. In the beginning it was just a little meeting and a mailing list which then transformed into the ACM[1]. My best guess would be that such "informalities" were the basis for most conferences in the first place. I would argue that the notion of "Special Interest Group" embeds such an informal view.

The reason for having peer reviewed conferences is unknown to me, but having peer reviewed conferences is the unique point separating computer science from other fields as far as I know. For some conferences this review process is even harder (or better...) compared to most journals in other fields. New conferences then use existing (reviewed) conferences as an example. Of course, old conferences do not change (partly due to the flame-war you mentioned in your question). During POPL 2012 there was the SIGPLAN Business Meeting, where there was a discussion about publishing the proceedings as a journal. During this session some sort of flame-war started, but you would have to ask Philip Wadler about who those "rebellious researchers" were.

I hope my guess-work is not completely off and helps you find the right direction for more information on this subject.

[1] Samuel B. Williams. 1954. The Association for Computing Machinery. J. ACM 1, 1 (January 1954), DOI=10.1145/320764.320765 http://doi.acm.org/10.1145/320764.320765

jf2011
  • 71
  • 1
  • 2