Sunday, May 21, 2006

The financial folly of pay-per-view, for the funder


Some of the best arguments for open access, can be found by simply looking at the alternatives, such as pay-per-view. It is hypothesized that the costs to a funding agency of reviewing the previous work of a grant applicant, following the pay-per-view model, could quite easily average $12,000 US per grant application, or more. This makes the payment of a very modest fee, such as the PLoS $1,500 US per article, for top-notch open access publishing for access to everyone, everywhere - not only a good idea, but an incredible bargain, too. Reviewing the works of the self-archiving author provides significant benefits in terms of both costs and time for the funding agency, that perhaps it makes sense for the funder to prioritize or expedite such applications. It is hypothesized that even without conscious intent, expediting of grant applications by self-archiving authors may be a natural phenomenon, due to the time savings and simplicity of review of previous work.


Let's look at pay-per-views, or purchase of article on a one-at-at-time basis, from the point of view of the research funding agency.

An application for a research grant is received. The application is reviewed by a committee. (As an aside - publishers sometimes think the funding agencies don't properly appreciate peer review. Nothing could be farther from the truth - the scrutiny the final article receives is nothing compared to the careful review before a funding decision is made).

For the sake of simplicity, let's assume that the researcher has previously published 20 articles, and there are 5 people on the review committee. Note that a great many researchers will have many, many more articles than 20, and many a review committee includes more than 5 people.

For the price of pay-per-view, let's take the $30 US that Elsevier charges for an article that I published in 1997, "Information literacy skills: An exploratory focus group study of student perceptions", Research Strategies, Volume 15, Issue 1, Winter 1997, Pages 4-17. (If anyone would like to read this article, please let me know - that would give me needed incentive to dig up an author's version for self-archiving).

5 reviewers each retrieving 20 articles on a pay-per-view basis at a rate of $30 US per article results in direct costs of $3,000 US to review the researcher's previous work. Factor in the indirect costs - 20 invoices each X 5 researchers - and this amount could easily double ($6,000 US per application).

Funding agencies need to review all applications, of course, not just the successful ones. If the rate of acceptance is 50%, the cost per funded applications doubles again ($12,000 US per application).

All this, just to pay-per-view for the application, and not a soul has seen the results of the funded research! Doesn't this scenario make the idea of paying one modest fee - such as the PLoS $1,500 US for top-notch publishing services, for access to everyone, everywhere - seem like not only a wonderful idea, but an incredible bargain, too?

Compare this pay-per-view scenario with the time and cost-efficient review of the works of the self-archiving author. See, for example, my own works in the SFU Institutional Repository, or E-LIS. Not only are there NO costs - either direct or indirect - but this approach takes an absolute minimum time for the busy reviewer of applications, too.

Given the vast difference in time and costs for full review of grant applications - perhaps funding agencies should prioritize, or expedite, review of applications from self-archiving authors? Or, is the expediting of such applications merely a natural phenomenon - after all, if the author's work is openly accessible, accessing previous work takes no time at all, right?

This post reflects my personal opinion only and does not represent the opinions or policy of the BC Electronic Library Network or the Simon Fraser University Library.


  1. If they were published in PLoS, the same 20 references would be free to each of the 5 referees, but each one would have cost $1500 to publish -- thus the cost, though borne by previous funders, is 20x1500x2= $60,000.

    Agh, I'm missing something there, but I can't put my finger on it.

  2. Thanks for your comment - here is what I think you are missing, Bill:

    With open access publishing, the total cost for world-wide dissemination for all time, which you have calculated at about $60,000 - is roughly equivalent to the cost of grant reviewers' access to the previously published works of one grant applicant under a pay-per-view model.

    Same cost - but the magnitude of the difference in access is hard to contemplate!

    Plus, with open access, there is time savings for the grant reviewers - no need to request articles, deal with invoicing, etc.

    To update the cost projections: PLoS has increased their fees, however the PLoS fees do not need to be doubled (this is a factor to account for inefficiencies of pay-per-view) - so the PLoS total would now be closer to $50,000, not $60,000, but close enough for ballpark purposes.

  3. Ah, the 2X was to get a figure "per approved grant" -- so yes, it doesn't apply to the OA cost, which is for all time (you're right, that's what I was missing).

    So if we take your updated figure of $50K, that is $2500/paper. This seems to fall roughly in line with the cost-per-published-article figures I've seen lately (not counting journals who claim $30,000/paper!). To compare with the PPV model, we'd need to know how many times an "average" article is needed, and paid for, by reviewers.

    Using your figures, $6000 for 20 papers is $300/paper -- so for these OA and PPV back-of-the-envelope "models" to come out even an average paper would have to be required by ~8 different grant review panels.

    Most PIs will submit at least one grant per year over a, say, 25 year career, so a conservative guess would have each paper being needed 20 times.

    But if we assume the panel can co-ordinate so as to only buy one copy of each article and make further copies for each other, the cost per application reduced by 4/5ths, so that puts us in the neighborhood of $1200 for 20 papers, or ~$60/paper. At that price, even-up is ~40 review panels. Further, if funding agencies keep digital copies of each paper they buy, they don't need to buy it more than once.

    But co-ordination might not be (legally) possible.

    But would a review panel really read an applicant's last 20 papers?

    But the review panel usually has to review a project as well as an applicant, and the project description will have references, some of which the panel will want to read.

    But indirect costs run around 50% on average for grants; surely they are not so high for review panels, whose overheads do not include equipment, consumables and so on?

    But, but, but. Even if we could find sources for all the assumptions, perhaps such modeling is only an opportunity for the anti-OA lobby to tie up the debate in endless nitpicking over numbers. In good faith or bad, one can always dispute an assumption, or find a contrary example, or whatever.

    I'm a researcher: I want OA even if it represents no saving at all. The added convenience would be of enormous value to me, and the increased access for researchers at less well-funded institutions would have a knock-on benefit throughout the scientific community. There are other knock-on benefits, too, such as those that go with the data mining opportunities that OA provides. It's virtually impossible to put a price on those benefits, at least until we can look back over some years of near-100% OA.


Thank you for your comment. Comments on IJPE are moderated.