Tuesday, January 27, 2009

A free CG journal?

For the last few years I've been discussing the possibility of setting up a free computational geometry journal with my colleagues. For me the two main reasons of having a free journal is the price we pay for a subscription (CGTA costs EUR 1,041/year) and the copyright transfer.

The idea would be to make the journal completely free which would require it to be as low maintenance as possible. The aim is to make the journal belong to the community and hence minimize the workload on individuals. How can this be achieved and what are the issues that must be considered?
  • The journal should work as a traditional journal with the difference that papers are only published on-line and are free to access by anyone.

  • The editorial board (including the Editors-in Chief) should be changed regularly. I believe 3-5 years is a reasonable time to serve as an editor.

  • Fairly large editorial board, say 30 (reliable) researchers. Try to distribute the papers among the editors.

  • The editors have to take greater responsibility throughout the process - from submission to publication. For example, the editor has to check that the final version is in the required format. One should also allow the editors to have more power, for example, if authors don't follow the stated procedures then the editor should have the right (within reason) to reject it, all to minimize work for the editors.

  • Aim to only accept high quality research since this will decrease the number of submissions (at least in the long run). Since the journal isn't making any money it doesn't have to publish any issues if there are no papers accepted.

  • Where to host it? If we buy a domain name then it doesn't really matter where we host it and we could move it if necessary. A domain name is cheap so this isn't a problem.

  • Free Journal management systems are available. I played around with Open Journal Systems which might be adequate (needs to be investigated further).

  • How to get ranked? If the journal is successful it will get ranked sooner or later. Is this a problem initially?

  • Name of the journal? Suggestions?

It doesn't sound hard, but I'm sure I forgot lots of issues. Comments and ideas are highly appreciated.


  1. Have you considered hosting it on arxiv.org? Some other journals are heading in that direction, including I think JAIR.

  2. By the way, I think hosting does matter. Specifically, authors want to be reassured that the host won't disappear at some future date leaving their papers inaccessible (as happened I think once for CCCG). That's why joining a bigger site like arxiv might be attractive.

  3. arxiv.org is an excellent idea. I'll look into it. Btw, where does arxiv.org get their resources from?

    > By the way, I think hosting does matter.

    What I meant was that if we we have a domain then it doesn't matter where files are hosted since they can easily be moved.

  4. Long ago it used to be run out of Los Alamos but nowadays it's operated by the Cornell University Library. According to http://www.library.cornell.edu/staffweb/june2004status/8.html the university itself has kicked in a lot of its operating budget; the fine print at the bottom of the main arxiv.org page says there's also partial funding from the NSF.

  5. perhaps it would be enough that an open journal like dmtcs would have a section in DCG section...

    In any case, the idea of open CG sounds great!

  6. It's a possibility. However, looking at the webpage of DMTCS it doesn't seem to work very well.

  7. I was about to tell you that you were probably at the wrong URL, then I went the DMTCS website...

    Here are a few suggestions:

    A) Always have an open-ended issue on the go. When an editor accepts a camera-ready version it immediately (as in the same day) goes into the current issue and gets assigned the next available page numbers. When an issue is finally closed, the date should include the entire time the issue was open so you get a bibliographic entry that looks like

    A. Fricklesnitz, Improved flavours for pickle chips, Journal of Pickle Chips (3)4, Dec. 2008-Mar. 2009.

    B) Find a way to make (A) work with things like special issues. Accept the fact that you will have special issues that come out in the middle of an open issue. Maybe just assign issue and volume numbers based on the closing date of the issue.

    C) Don't let people submit to individual editors. Have the journal system assign the paper to an editor. The scope is small enough for this to work and it prevents the problem of people who are efficient editors or very well-known being rewarded with more submission to edit. Let the author know which editor was assigned their submission.

    D) Have the journal system send automatic, and fairly frequent, reminder emails to referees and editors. Many editors feel bad about having to constantly nag referees, so don't make them do it.

    E) Make a sample submission for authors and editors to use as a guide for formatting. The editor will do a quick pattern match between this and a camera-ready paper to decide if its ready to go. That should be the only form of copy editing that gets done. Authors are responsible for checking their own grammar and typsetting.

    F) Try and set up the system so that nobody ever has to send an email to anyone else. All communication between authors, editors, reviewers, and the editor in chief should be done through the online system and the system should preserve them. Lost (or ignored) emails lead to publication delays.

  8. By the way, JAIR is a great example of what the journal should look like from outside world:


    Although I don't see any mention of the arXiv. According to the webpage they're still being published online by something called the AI Access Foundation


  9. Pat, all your ideas sound good although (D) and (F) depend on the journal management system. I'll try to take a closer look at it this week to see what kind of features the software offers.

  10. I very much like the idea of such an open CG journal for reasons that everybody knows. And also this discussion in this blog is good to get some feedback.

    As for the hosting, a system using mirrors might mitigate the problem of a host disappearing. JGAA seems to have a nice system.

    As Joachim and also Pat pointed out, the management system should be as low maintenance as possible. I also see security as one aspect of low maintenance. The more features a website has, the more likely it might be hacked/spammed etc. Open Journal Systems address this question in their FAQ, but one should look into this a bit more closely.
    From the outside, JGAA has a pretty "simple" website ("simple" = only links to other pages and pdf files). From an administrators point of view this is probably a rather secure setup with low technical maintenance.

    I agree with Joachim that the initial ranking (or lack thereof) could be a problem to get submissions, as rankings and other counters seem to get more and more important for a personal career.
    So yes, a rather high standard right from the beginning might be a good choice to get ranked highly in the long run. For the beginning, one could hope that there will be scientists that would volunteer to be an editor of the journal, where these scientist should be reliable, respected by the community and well-known for quality. This might help to kick the ranking in the right direction, and it will ensure a wider dissemination and hopefully a high quality standard.
    Anyone? ;-)

  11. I found this in the OJS documentation:

    Open Journal Systems supports the LOCKSS (Lots of Copies Keep Stuff Safe) system to ensure a secure and permanent archive for the journal. LOCKSS is open source software developed at Stanford University Library that enables libraries to preserve selected web journals by regularly polling registered journal websites for newly published content and archiving it. Each archive is continually validated against other library caches, and if content is found to be corrupted or lost, the other caches or the journal is used to restore it.

    so it seems there's a fairly easy option for redundancy.

  12. A freely available online journal on computational geometry is a great idea. I would make it broader to include combinatorial geometry. Perhaps call it "Euclidean Geometry".

    I agree with Pat that papers should be published as soon as they are accepted. But I see no reason to have issue numbers and page numbers. I like the model used at the Electronic Journal of Combinatorics, where each paper is given a paper number like "R23" and is included in a specific volume, which resets at the start of each year. The website can record the exact date the paper was submitted/revised/accepted/published. The pages of each paper are numbered starting at 1 (which makes sense given that papers are never printed consecutively). Special issues can be given a separate volume.

    I am not a fan of Open Journal Systems. DMTCS previously had a beautiful clean website, but it is now a total mess under Open Journal Systems.

  13. Thanks David.

    We discussed the volume/number/pages issue and I think we agree with you. There's no need for it if you publish it as soon as the paper is accepted.

    I also agree regarding the broader scope. I'm very fond of metric embeddings and believe that should be included. Jeff was also arguing for computational topology, so I see no reason why it should only include "traditional" CG.

    The management system is a problem. We tested the Open Journal System and it works but it's far from perfect. Are there any other systems that supports, for example automatic remainders, and so on.