5 ms·

My feeling is that the misery is inhomogeneously distributed when it comes to scientific disciplines. In physics the situation does not seem too terrible. The

My feeling is that the misery is inhomogeneously distributed when it comes to scientific disciplines. In physics the situation does not seem too terrible. The American Physical Society is a nonprofit which runs a family of go-to subscription-type journals. Preprints on arxiv are the standard. Publishing innovations are popping up from time to time. A prominent example is scipost.org, which is funded by national funding agencies and is free to both publish and read. Also, their peer-review is public. Full disclosure: I have published in various APS journals.

jabl 9y ago
> My feeling is that the misery is inhomogeneously distributed when it comes to scientific disciplines. >In physics the situation does not seem too terrible. Yes, but I'd add that also in physics the misery is quite inhomogeneous. High Energy Physics seems to lean heavily towards arxiv and open access, whereas in some other subfields, such as my own in my previous life, some Elsevier journal is #1. Or lets say, on my own case, I guess one could (mostly, at least) publish in non-Elsevier journals with no serious career repercussions, but we're still very far from being able give Elsevier the finger and get by without an Elsevier subscription since one needs to read what others are doing. I'm somewhat hopeful for the longer term, but progress is painfully slow.
- jogundas 9y ago
  Just curious: which physics subfield is that? Perhaps my previous comment was overly optimistic, as both Journal of High Energy Physics and the Nature family are published by Springer nowadays.
  - jabl 9y ago
    I worked in, broadly speaking, computational studies of surfaces and reactions on them (adsorption, heterogeneous catalysis, AFM, STEM, etc.). In this area, the #1 journal is (the aptly named) Surface Science, published by Elsevier. Although us computational guys do also publish a lot in PRB/PRL and a smattering of other journals as well of course, but particularly for experimentalists Surface Science is clearly #1. So you still have to read a lot of SS articles.
mjn 9y ago
Parts of computer science are not too bad either. In my field, AI, the top-tier journals and conferences are all open-access and run by nonprofits. I mention conferences because in computer science they are important as well; the top peer-reviewed conference proceedings are as important as journals. * The top two journals are probably JMLR (machine learning) and JAIR (AI in general), both open access and run by nonprofits. No publication charges either. * The top two general AI conferences are AAAI and IJCAI; the top two machine learning conference are ICML and NIPS. All four of these publish open-access proceedings (including archives of past proceedings) and are run by nonprofits. Of course this didn't happen entirely by default, it's taken some effort over the years. AAAI has always been nonprofit, but its digital library has not always been open access. And in machine learning, the top journals used to be run by the Elsevier/Springer duopoly, which was solved in the early 2000s by 40 editors of Machine Learning journal, which included a large proportion of top researchers in the field, resigning en masse to start an open-access alternative: http://www.jmlr.org/statement.html http://www.jmlr.org/statement.html
- soulbadguy 9y ago
  The resignation letter is actually a great read. I think the author summarized the situation quite well : "We see little benefit accruing to our community from a mechanism that ensures revenue for a third party by restricting the communication channel between authors and readers."
- FabHK 9y ago
  Similarly, the editorial board of the Journal of Algorithms stepped down in unison and started the ACM Transactions on Algorithms, in protest against Elsevier, back in 2003. Don Knuth, founder of the JoA, had a hand in it. Wish this happened even more often - without the scientists (and the prestige/impact factor endowed by the scientific community), the publishers have nothing. Don Knuth's letter (typeset in TeX, of course) is well worth a read [1], and since then the situation has only gotten worse, I think. Ceterum censeo Elsevier esse delendum. [1] http://www-cs-faculty.stanford.edu/~knuth/joalet.pdf http://www-cs-faculty.stanford.edu/~knuth/joalet.pdf https://en.wikipedia.org/wiki/ACM_Transactions_on_Algorithms https://en.wikipedia.org/wiki/ACM_Transactions_on_Algorithms http://www.cs.colorado.edu/~hal/jalg.html http://www.cs.colorado.edu/~hal/jalg.html

jabl 9y ago

> My feeling is that the misery is inhomogeneously distributed when it comes to scientific disciplines. >In physics the situation does not seem too terrible. Yes, but I'd add that also in physics the misery is quite inhomogeneous. High Energy Physics seems to lean heavily towards arxiv and open access, whereas in some other subfields, such as my own in my previous life, some Elsevier journal is #1. Or lets say, on my own case, I guess one could (mostly, at least) publish in non-Elsevier journals with no serious career repercussions, but we're still very far from being able give Elsevier the finger and get by without an Elsevier subscription since one needs to read what others are doing. I'm somewhat hopeful for the longer term, but progress is painfully slow.

jogundas 9y ago
Just curious: which physics subfield is that? Perhaps my previous comment was overly optimistic, as both Journal of High Energy Physics and the Nature family are published by Springer nowadays.
- jabl 9y ago
  I worked in, broadly speaking, computational studies of surfaces and reactions on them (adsorption, heterogeneous catalysis, AFM, STEM, etc.). In this area, the #1 journal is (the aptly named) Surface Science, published by Elsevier. Although us computational guys do also publish a lot in PRB/PRL and a smattering of other journals as well of course, but particularly for experimentalists Surface Science is clearly #1. So you still have to read a lot of SS articles.

mjn 9y ago

Parts of computer science are not too bad either. In my field, AI, the top-tier journals and conferences are all open-access and run by nonprofits. I mention conferences because in computer science they are important as well; the top peer-reviewed conference proceedings are as important as journals. * The top two journals are probably JMLR (machine learning) and JAIR (AI in general), both open access and run by nonprofits. No publication charges either. * The top two general AI conferences are AAAI and IJCAI; the top two machine learning conference are ICML and NIPS. All four of these publish open-access proceedings (including archives of past proceedings) and are run by nonprofits. Of course this didn't happen entirely by default, it's taken some effort over the years. AAAI has always been nonprofit, but its digital library has not always been open access. And in machine learning, the top journals used to be run by the Elsevier/Springer duopoly, which was solved in the early 2000s by 40 editors of Machine Learning journal, which included a large proportion of top researchers in the field, resigning en masse to start an open-access alternative: http://www.jmlr.org/statement.html http://www.jmlr.org/statement.html

soulbadguy 9y ago
The resignation letter is actually a great read. I think the author summarized the situation quite well : "We see little benefit accruing to our community from a mechanism that ensures revenue for a third party by restricting the communication channel between authors and readers."
FabHK 9y ago
Similarly, the editorial board of the Journal of Algorithms stepped down in unison and started the ACM Transactions on Algorithms, in protest against Elsevier, back in 2003. Don Knuth, founder of the JoA, had a hand in it. Wish this happened even more often - without the scientists (and the prestige/impact factor endowed by the scientific community), the publishers have nothing. Don Knuth's letter (typeset in TeX, of course) is well worth a read [1], and since then the situation has only gotten worse, I think. Ceterum censeo Elsevier esse delendum. [1] http://www-cs-faculty.stanford.edu/~knuth/joalet.pdf http://www-cs-faculty.stanford.edu/~knuth/joalet.pdf https://en.wikipedia.org/wiki/ACM_Transactions_on_Algorithms https://en.wikipedia.org/wiki/ACM_Transactions_on_Algorithms http://www.cs.colorado.edu/~hal/jalg.html http://www.cs.colorado.edu/~hal/jalg.html