Cloud Hosting

When a scholarly paper is published, someone has to pay.

Publishers like Institute of Electrical and Electronic Engineers (IEEE), my professional society, and Springer charge big bucks to read their papers. The fees are billed to individual subscribers and, more commonly, to companies and universities who want to give their employees access to the papers. My own university, Baylor, like most research universities, has a considerable library budgets on account of these publisher fees.

There is growing pressure to kill these fees in favor of open access to scholarly papers. Thus, anyone can read a scholarly paper at any time for free. Free access takes the money from the pockets of publishers so they push back. Someone has to pay, eventually, so open access, as it is called, means that the author bears the cost.

The author doesnt always bear the cost personally. Sometimes, the authors institution foots the bill. Cash cow organizations like Google readily pony up. In the U.S., if a public university coughs up the money for their employee author, the fees are paid indirectly either by the states taxpayers. When the fees are paid by government granting agencies like the NSF, NIH or NIST, they are paid by the US taxpayer.

MDPI, an open access publisher, charges authors between 1000 to 2000 Swiss Francs (over US$1000 to $2000) to publish in their open access journals.

Where do these fees go? At IEEE, paper authors, reviewers, and associate editors are not paid a dime. All is volunteer. When I was Editor-in-Chief of IEEE Transactions on Neural Networks, I was voted money to support one part-time clerical position but I personally received no compensation. A prevailing view is that the prestige of being an editor or author is payment enough. In academia in general, a major component in a professors promotion and pay raises is paper count. As the saying goes, the Dean cant read, but the Dean can count.

Professionally sponsored conferences also feed the cash-hungry publisher. I have been the Organizational Chair for numerous conferences, big and small. Volunteers from universities and companies run the conference but are paid nothing. The publications from these conferences are handed over to a publisher who then sells them.

There is an old joke that graduate students are machines that turn coffee into papers. Publishers take these papers and turn them into cash.

I am currently the Editor-in-Chief of the open-access journal BIO-Complexity and, yes, we have a publication fee. After thorough peer review, we have waived the fee for every paper thus far published in BIO-Complexity. Because the journal is small, we have been able to run it with all volunteers.

There are ways to avoid the publication cash grab described above. A popular tool is arXiv.org (pronounced archive where the X is pronounced like a k, as in the Greek chi. The site, managed by Cornell University, will post almost any scholarly work for free without peer review, at the request of an author. Someone, of course, reads it in advance, to be sure it is not whacko. But there is no peer review as such. Paper topics are physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, electrical engineering and systems science, and economics. I am aware of no bias in the acceptance of papers for arXiv, but in these days of blatant censorship by places like Amazon, YouTube, Twitter, and even Paypal, one never knows.

A second way to get hold of a paper for free is Google Scholar. Enter a topic like artificial intelligence ethics and you get a search of papers addressing the topic. To the right of many of these listings, you will find a link to a free copy of the paper. Many authors will also post copies of their papers on their own web sites. Google scrapes the web and then nicely links them here.

At one time, publication count was limited by paper supply. Each paper was typeset and printed, and a hard copy was sent to subscribers via snail mail. Printing runs were fixed. There was no print-on-demand. These physical constraints are all gone. Many conferences and journals no longer even generate hard copies. Everything is on line. Even older hard copy has been digitized and made available. For a price.

When web access became ubiquitous, publication count exploded. The graph of the IEEE publication count is revealing.1

Here are my speculations about the story this graph tells. Publication count increased linearly about 2500 papers per year until about 2000 when there is a sharp uptick. From 2001 to 2010, the increase was, on average, about 19,000 papers per year or roughly four times larger than before. The break point occurs when papers became available electronically. Without the requirement of hard copy publication, making a paper available became cheaper. When I came to Baylor from the University of Washington in 2003, the Baylor library had yet to subscribe to IEEE Xplore.

There is a dip in the curve for a few years, starting in 2010. I cant think of a reason why. But the upward trajectory starts again in 2014, breaking through 300,000 in 2019. Thats over a thousand papers every weekday. There was also a COVID dip in 2020 that is not shown on the graph.

Supply and demand pushes publication counts up and up. Academia rewards professors who publish and publishers are more than happy to accommodate the demand. For a price.

[1] Heres how I got the data. I went to IEEE Xplore and did a search year by year. IEEE, the Institute of Electrical and Electronic Engineers, subsumes electrical and electronic engineering and computer science. At over 400,000 members, IEEE is the worlds largest professional society. In Xplore, I entered the year 2000 and Xplore responded with 69,762 papers. This was repeated from 1980 to 2019. The search is not perfect but gives reasonable numbers.

You may also wish to read: Why its so hard to reform to reform peer review. Reformers are battling numerical laws that govern how incentives work. Know your enemy! (Robert J. Marks)

See the rest here:

Inside the Economics of Science Papers - Walter Bradley Center for Natural and Artificial Intelligence