Wednesday, December 29, 2010

Publication Venues in Computer Science

In academic Computer Science, there are basically three publication venues that "count": peer-reviewed conferences, peer-reviewed journals, and peer-reviewed workshops/symposia. There are of course many other perfectly credible ways to publish one's work (e.g., technical reports, books), but these are the top three.

Unlike in most scientific fields, including our closest cousins Engineering and Mathematics, journals are not the de facto place to publish papers. I can't speak for all subfields of CS, but basically everyone I know only publishes in journals because they feel they have to (e.g., multi-disciplinary tenure and promotion committees that expect journal publications, research rankings organizations that still don't seem to 'get' conferences, etc.). Some subfields this is not the case, such as in interdisciplinary fields like Bioinformatics and CS Education, but for most major areas of CS, conferences are where the action is.

For these fields, the top conferences have extremely low acceptance rates, many less than 17%. The program committees are comprised of the top scholars in the field. And in some fields, anyone who is anyone attends these conferences, so managing to get a paper accepted is a pretty big deal that gets a researcher much visibility.

Some journals have similarly rigorous standards of review and are known for their quality, for example the IEEE Transactions and ACM Transactions family of journals are highly regarded. There are occasionally other journals that are good, but the vast majority are either decidedly mediocre or utter rubbish. We don't really have any comparable C/N/S type journals.

Workshops and symposia generally have a much higher acceptance rate than conferences and journals, but they are still peer-reviewed and are often archival (e.g., ACM Digital Library, IEEE Xplore). They have quite a few advantages. First, they are usually co-located with a conference, which means you can often go to both on your University's dime. Second, they present a fantastic opportunity to float half-baked ideas and get one-on-one feedback from your peers. And third, oftentimes workshops are the only place you can meet other scientists interested in the same area of specialized research as you, which nearly always leads to good things.

But the most useful thing about conferences and workshops over journals is that you have the opportunity to tell potentially hundreds of people about your work. These are all people who learn your name and face and start to match it to a research area. This is invaluable, because it leads to professional relationships that will see you through your career - jobs, funding, tenure/promotion letters, etc.

Journals don't really get your name and face out there as well, unless some news outlet picks up on your article. People do read journals, but I suspect most readers associate papers more strongly with an institution than a name, particularly if it's a new name on the research scene.

So I think it's nice to have a diverse portfolio when it comes to publishing. It's good to have a mixture of papers in the top conferences with low acceptance rates, in journals that are well-respected, and in workshops that are useful to the researcher.


  1. How many conferences do you go to per year? The IEEE people I know go to about a dozen. Which kind of blows my mind and is way more than my 1.5 per year. I don't think I could do that much travel.

    In my field conferences don't "count" the way journal publications do. They are more like icing on the cake, as opposed to the actual cake.

  2. No, that's the case in theory, too.

  3. @Bashir: Depends on the year and where the conferences are being held. I have one annual conference and one biennial conference I always try to attend, regardless of location. There are a few other conferences I try to attend if they're reasonably close by (i.e., less than 6 hours of airplane travel). Workshops are the same way - depends on the location mostly, but there is one I really like to try to make it to. So I guess on average maybe 2-3 per year?

    Often I actually enjoy publishing in journals more, both because no travel is required and because the reviewing process is usually more cordial and relaxed.

    @Jeffe: Ah, thanks for letting me know - I'll update the post. My theory colleagues were more math people really.

  4. Thanks for the post, FCS!
    In my university's CS department there are a number of people who do control theory/optimization/signal processing and they do publish in journals much more than at conferences. (But I am not sure if these fields fall under hard-core CS?)

    I am in a physics/engineering field and I go to about 4-5 conferences per year, depending on other obligations, plus a few additional trips to give talks at univeristies, panel reviews or grant progress reviews. There are people who travel much, much more -- it's a matter of choice in how you run your group really and what the rest of your life looks like (e.g. small kids or not). When you have a group of 20-30 people, you are likely not very hands on and must travel for networking and fundraising. Some people really enjoy this aspect of science, I do it but it's not my favorite thing; I err on the side of making sure every paper is written up and submitted as soon as it's ready and work closely with students, so I have really high papers-per-grant-dollar and papers-per-graduating-student ratios and can usually get away with having to raise less money than most people in similar fields to do the same amount of work. I think there is a fair bit of leeway in how the academic game is played, depending on where you set your sights. But if you want to be a really big player, sooner or later you must give in to all the schmoozing and traveling and fundraising (at least in my field). I am planning on changing my game (and conquering the world ;) when the kids no longer need me as much as they do now and I can be away from home more.
    [Sorry for going off on tangents, FCS.]

  5. Ah, see, control theory and such reminds me much more of Engineering than CS. But it's all splitting hairs, really.

    No worries for tangents, GMP! What you have to say is interesting, and as I think I've said before - mi blogga es su casa. Go to town :)

  6. While I enjoy going to conferences, I have to wonder how much CS as a field is contributing to global climate change, what with all the air travel involved. I live a pretty "green" day-to-day life, but because of all my air travel I end up with a pretty big carbon footprint per year.

    Though as we like to say in my subdiscipline, a bad baseline means it's hard to get worse! Maybe CS could contribute to solving the problem just by reducing conference travel :)

  7. I'm with you, Anon! I think about that every time I get on an plane. But I also appreciate the value of interacting with people face-to-face. (Funny how we are one of the most anti-social disciplines and we value conferences higher than journals).

    A few conferences I attend rotate between several parts of the world (e.g., US West Coast, US East Cost, Europe, East Asia), and I think that helps a little bit overall. But really we should engineering something better to be sure! :)