Open access: response (I)
On 24 October Dr Klaus Graf kindly posted a comment on this blog, consisting primarily of links to three articles on the subject of open access publishing. I am grateful to Dr Graf for his attention and feel, in return, that his comment deserves a response.
Of the three articles to which Dr Graf provides links, I have not yet read the one in German – I fear I will need some linguistic help with that – and so will not respond to it at present. The other two make cases for open access publishing. Of these two, the stronger seems to me to be ‘An Approach to Open Access Author Payment’ by Donald W. King, so I will comment on that one here. I plan to return to the other, ‘Full open access to articles’, in due course.
The link to the King article, which appears in D-Lib magazine, is: http://www.dlib.org/dlib/march10/king/03king.html.
Before commenting in detail on the article, I should provide some context. My reason for examining the subject of open access in this blog is that I am considering establishing an open access publication programme. My argument in previous postings is that I could see that open access publishing can provide advantages on tactical or pragmatic grounds. However, I was critical of the idea that open access offered a superior model in toto – and especially critical of the idea that this claim could be based on ethical foundations.
I found the King article of most interest in one respect. I have argued that replacing a traditional model with an open access model should be thought of primarily as a transfer of costs and that it was not clear, therefore, how it could be seen to reduce costs at the same time. King’s article, however, gives some attention to this question. It argues that there would be savings in, for example, scientists’ time and in avoiding inter-library loans. I do not think my previous posts have given due consideration to such points.
However, there is much in the article that I take issue with. King’s argument is that the national government should ‘make author payment for all peer-reviewed science and medical articles’. Government should cover author payments 100%.
I note that King’s article refers (a) only to journal articles and (b) only to science and medicine. This raises two questions: (a) what about other kinds of publications – books, for example, or grey literature pieces; and (b) why only science and medicine? And implicit in these questions are questions of definition: what does or doesn’t count as a journal? what does or doesn’t count as science or medicine? These might seem merely academic questions. If, however, government policy – and hence access to government funding – comes to rest on these points, they will no longer be merely academic: they will become operational questions.
If the government were to offer to underwrite the cost of publication, I do not suppose, as a publisher, that I should mind. Money is money. That, however, is to look at the issue from a purely local, self-interested, point of view.
From a more general, societal, perspective I have to say I felt enormously disappointed when I read, in King’s abstract, the following: ‘it [i.e. open access] requires extensive government support, which may or may not be feasible, but the approach is presented here nevertheless’. If King wishes to exclude a consideration of feasibility from his assessment, then of course he is at liberty to do so. But what then is the value of the assessment? Feasibility is hardly a negligible concern.
More fundamentally, the cause of my disappointment is that the argument seems both banal and out-dated. Banal, because I think most industries could argue that life would be better for them if only the government would kindly pay all the bills. And out-dated, because in these fiscally challenging times governments are not exactly begging for suggestions on ways to spend more money. We can’t hold these debates in a vacuum.
At the very least, we need to consider where this additional government funding that King argues for is going to come from. The argument that the additional funds amount to only a fraction of the total budget hardly cuts the mustard. By the same logic, one could argue that, since it’s such a small sum, it could instead be deducted from the current R&D budget, on the grounds that nobody would miss it (though of course they would). Coming from a statistician, as King apparently is, this argument from fractionality makes disappointing reading.
King is aware that additional government funding has an opportunity cost: that is why, to his credit, he is careful only to use the word ‘free’ in inverted commas. Yet despite this awareness, he seems reluctant to confront the implications. The significant absence is the word ‘tax-payer’.
There are further weaknesses in the argument with regard to the role of government. King works in North Carolina, a state I have never been to. It may be that North Carolina is an exceptional place where such phenomena as government waste, bureaucracy, and inefficiency are absent. If so, it is unlike any place that I have been to.
The idea of founding some huge government agency to fund science publishing feels me with dread. I can imagine the plate-glass building, the shiny new office furniture, the hefty account signed off for a branding company to invent a logo, and so on. Maybe I am being unfair here – my imagination stoked by the recent behaviour of some agencies in the UK – but the fact is that I can’t actually find any administration cost factored in. (To give one example from above: what will be the cost of having to adjudicate on whether some article or journal qualifies as ‘scientific’?) So what bureaucratic cost is assumed in this model?
There are other problems with the proposal for government funding. In science publishing, there is an increasing premium on rapid decision-making. How well do ‘rapid decision making’ and ‘government’ fit in the same sentence? And governments, when they spend a lot of money on something, tend to start to want to take control. A plea for ‘joined-up thinking’ leads to the argument that budget of hundreds of millions should be used to pursue other policy objectives at the same time. The idea that the government would passively sit back and just sign the cheques does not ring true.
These arguments bring feasibility back into the frame. In the process, they confirm that an assessment that omits questions of feasibility has little value. Moreover they reveal what is, in effect, a bias in the analysis. King quite properly identifies the weaknesses in the traditional system and counts them against it (without, it should be noted, assuming away questions of feasibility on that side!) What we don’t see is a treatment of the weaknesses of the proposed governmental system. They seem to me to have been, as it were, air-brushed out of the debate. Now, if you count deficiencies on one side of the argument but not on the other, it is not surprising that one then comes down in favour of the latter. Not surprising, but not persuasive either.
One final point – one that I would have thought a statistician like King might have anticipated – concerns Goodhart’s Law. According to this law, any statistical correlation breaks down when one starts to use it as the basis for policy. Note that King argues that the number of articles produced under a government-funded system would not grow significantly. In particular, he argues that peer-review would act as a brake on the number of articles accepted.
This is a very weak argument. Peer review does more than one thing. It analyses strengths and weaknesses in draft article and it arrives at a holistic judgment on whether the article deserves to be published. The latter is implicitly informed by a sense of opportunity cost – that, with limited resources, recommending one article for publication implicitly amounts to recommending that some other isn’t published. But if we remove that constraint, and government benevolently picks up the cheques for any article deemed publishable, the number of published articles will assuredly grow. After all, many articles are rejected, not because they’re unpublishable, but simply because they’re not quite as strong as some others. The number of articles published, and the budget required to finance them, will grow. Great for scientists, fine for publishers: but we had better continue not to mention the tax-payer.
The argument for open access originates from reasonable, indeed honorable, aims – to make research available to people who can’t afford or access it at present; and to provide researchers with a wider audience. But to move from such aims to the conviction that science journals should be centrally funded – well, that requires what we might politely call a spacious argument.
I return to the original context. My argument has been that, though open access is not preferable to traditional publishing as a total solution, it does have its place as an alternative within a mixed economy. That mixed economy is already developing – some journals have retained the traditional model, some are purely open access, and others use a hybrid model. A mixed economy doesn’t offer us an ideal world, but it does provide real benefits – not least, a degree a choice. And it is feasible.