Thousands in the machine learning community say they will boycott Nature’s paywalled Machine Intelligence Journal, which is set for a January 2019 release. So far, 2482 researchers have signed a petition pledging to not submit any work to the journal or review or edit any of its papers.
The key objection to the new publication is its subscription-based revenue model, wherein the journal will charge a paper submission fee and individual and institutional subscription fees.
This, of course, touches a nerve in the machine learning research community, which prides itself on an open sharing culture. It has been easy for the public to access ML papers, codes, and in some cases even datasets. Up-to-date papers from major conferences such as NIPS and CVPR can also be easily shared. ArXiv is sometimes criticized for its relatively loose review mechanism, but its huge volume of papers can nevertheless be accessed by a global readership free of cost.
Leading the boycott call are deep learning pioneers Geoffrey Hinton, Yann LeCun and Yoshua Bengio; research leads at Google, Facebook, Amazon and IBM; and academics from MIT, Stanford, CMU, and Oxford. OSU Professor Emeritus Thomas G. Dietterich, who initiated the petition, was Executive Director of the journal Machine Learning from 1992 – 1998, before the publication became the open-access Journal of Machine Learning Research (JMLR) in 2001.
Nature announced plans for Machine Intelligence last November, as a new online-only publication that would cover the “best research across the field of artificial intelligence” with relevant reviews and commentaries. All editors are internal staff working under Chief Editor Liesbeth Venema.
Open access in academic publications is an ongoing issue. Research access in areas such as biology, neuroscience, psychology, and social sciences can be extremely costly, even for wealthy institutions like Harvard.
The academic community has been battling publishers for open access for a decade. Sci-Hub, created by Alexandra Elbakyan in 2011, has an archive of over 64.5 million papers available for direct download. Reddit founder Aaron Swartz’s attempts embroiled him in a legal fight with Jstor and MIT that ended in tragedy.
Outside the machine learning community, research papers in other AI subfields such as robotics and hardware chips are difficult for non-specialists to obtain. There are also labs like Boston Dynamics that can live outside the “publish-or-perish” curse.
The machine learning community is rebelling against a publishing orthodoxy it regards as an impediment to progress. In a Reddit discussion thread, one researcher posted that he would judge a paper based on its ability to spread and instruct further research, rather than focusing on the publisher’s prestige.
Guide2Research has compiled a list of top journals in the cross-disciplinary field of machine intelligence, ranked according to their JCR impact factor, SJR, and Scopus H-index. Interested readers can check here for more information: http://www.guide2research.com/journals/
Journalist: Meghan Han| Editor: Michael Sarazen