Free publishing

This post is a manifesto for a very simple idea: the phrase “open access” should be replaced with “free publishing”. I am asking you, the reader, to make this change. The rest of this post will try to explain why.

The fundamental reason is that the phrase “open access” has already been compromised through contact with corporate interests that are not in the interest of science. The fact that extra adjectives such as diamond open access, platinum open access and fair open access are required shows that the words “open access” have lost their useful meaning. The pledge that I am a signatory of, No Free View? No Review!, does not consider gold open access and other hybrid models to count as true open access.

Open access is named by analogy with open source software. (At least I assume it is – I haven’t found any actual evidence for that claim.) Open source software was born as a wing of the older free software movement, which was strongly associated with a relatively extreme philosophy of people like Richard Stallman. Free software is codified as the four essential freedoms. The double meaning of the English word “free” is intentional, but free software is more fundamentally about free-as-in-libre than free-as-in-beer (see gratis vs libre).

Open source software was both a re-branding of free software and a philosophically different approach to it, both with the goal of making it more palatable to businesses. This was a spectacularly successful move, and open source software is now a major part of the strategy of many businesses big and small.

My proposal is that we equally deliberately take the opposite step: from open access publishing to free publishing, and for the opposite reason: to make academic publishing as unappealing as possible to businesses.

Open access is, fundamentally, about making articles free to read, and in this it has been a complete success. There are two distinct reasons for wanting to make articles free to read: reducing the spiralling cost to run an academic library, and to make the research available to independent scholars and the general public. (This second reason is particularly important in medical science, but also applies to every subject to a lesser extent.)

If research is free to read, where does an academic publisher make their money? The answer is article processing charges (APCs), or “pay to publish”. Nature Communications, for example, is an open access journal with a cost to publish of US$ 5560 plus VAT per paper. This is unusually high, with US$ 3000 per paper being more typical.

Of course academics usually don’t pay the APC themselves – their institution covers the cost. This is just moving money around, since previously the institution was paying the cost to read. And similarly for independent scholars, they now have to pay to publish instead of pay to read. In either case, I imagine that APCs have been set by publishers in order to keep their income roughly similar (or increase it, if I’m being cynical realistic), so the actual cost to institutions and independent scholars would not significantly change.

Of course I am not naive, and I know that publishing costs money. Tim Gowers, who founded the free journal Advances in Combinatorics, wrote in this great blog post (one that I come back to repeatedly) that “the total cost of running a new journal that isn’t too large is of the order of a few hundred dollars per year, as long as nobody is paid for what they do”. This money typically comes from grants and philanthropists.

A shining example of a free journal is Theory and Applications of Categories, which has been both free to read and free to publish since 1995. Compositionality, a new free journal that I’m on the editorial board of, is funded by a philanthropist, with Gowers’ estimate being the right order of magnitude for its costs. It is clear that a free journal must be very lean indeed, but there are multiple examples showing that it is very much possible.

I have not yet said clearly what the definition of “free publishing” should even be. This is because I’d rather let the community decide. The baseline minimum is that to qualify as free publishing it must be both free to read and free to publish, both for the reader/author and for their institution. So free publishing may be defined to be the same thing as diamond open access.

Perhaps free publishing should be stricter still. One possibility is to require the use of a license such as CC BY-ND, where the author grants certain permissions to others such as the right to host copies of the paper. Notably the arXiv default license would not qualify for this. Another possibility would be an “independence” condition where the journal may not share an owner with a non-free journal. The fair open access movement require community ownership of journals, strictly ruling out for-profit publishers.

Personally my view is that these more restrictive conditions would have more marginal benefits, and would risk dividing opinion and distracting attention from the most important issues. Thus while free software considers free-as-in-freedom to be more important than free-as-in-beer, free publishing would be the other way round. This is not unreasonable, since the nature of software and publications are very different. There is definitely a reasonable debate to be had here, though.

All of these are details. My central point remains: the phrase “open access” is no longer useful and should be avoided. The phrase “free publishing”, which is surely harder to corrupt into a different meaning, should be preferred and protected from a similar fate.