Self-Archiving Mandates

Laura Barnett and Hanna Hindstrom, "All research to go online," Times Higher Education Supplement, September 23, 2004

The Research Councils UK have proposed to mandate that all RCUK fundees make their articles openly accessible online by self-archiving them on the web. In a disappointingly inaccurate THES article (“ All research to go online” Sep 23), the authors get most of the important details wrong. They write:

THES: “[A] benefit of online open access publishing [italics mine] would be that academics and researchers would no longer have to rely on their institutions to provide access to articles published in subscription-only journals.”

Not only is it not open access publishing but open access self-archiving (of their own articles, published in subscription-only journals) that the RCUK is mandating for their researchers, but this does not mean that their researchers will no longer rely on their institutions to provide access to the journals they subscribe to: How could my giving away my own published articles online provide me with access to the articles in the journals my institution subscribes to? I give my articles away so other researchers webwide whose institutions cannot afford to subscribe to the journals my articles were published in can nevertheless access and use them. That is how self-archiving (1) maximises my own research impact, and, far more important, how it also (2) maximizes the return on the British public’s yearly £3.5 billion investment in research.

But the THES article misquotes me on (1)

THES ("quoting" SH): “if citations rose by 50 to 250 per cent because of online open-access publishing [sic, again: italics mine, but not the words] researchers could gain more than £2.5 million a year in potential salary increases, grants and funding renewals”

This simply leaves out altogether (2) the far more important £1.5 billion loss in potential returns on the British public’s yearly £3.5 billion pound investment in research (in the form of at least 50% more citations). Nor is this an if/then pipe-dream: The projections are based on objective, published measurements of the degree to which self-archiving increases research impact.

But by far the worst inaccuracy in the THES article – and it really does a disservice to those who pin their hopes on the RCUK policy for maximising British research impact -- is the gratuitous exaggeration of what is a real but remediable flaw in the current wording of the RCUK proposal. The current draft says

RCUK: “Deposit should take place at the earliest opportunity, wherever possible at or around the time of publication.”

But the THES article instead says:

THES: “Under the proposals from Research Councils UK, published work would not necessarily go online immediately. Academics and publishers would be allowed a grace period, which could last anywhere from a few months up to several years. The publisher would determine the exclusion period…”

This is utter nonsense, and it would make a nonsense of the RCUK policy, if this were indeed the form it took. The RCUK’s current language simply needs to be made more precise:

SH: “Deposit must take place immediately upon acceptance for publication, and access should be made open at the earliest opportunity, wherever possible at or around the time of publication.”.”

(In the meanwhile, the article has already been made visible webwide, so the authors can already email eprints of it to all those email eprint-requesters whose institutions cannot afford to access it, thereby still maximising its impact -- but with more author keystrokes per article than altogether necessary.)

The 8 co-signatories of the open letter in support of the RCUK policy, including the inventor of the web, Sir Tim Berners-Lee, are quoted correctly on this, but the THES authors don't seem to notice that what they themselves have written in THES is contradicted by the co-signatories' quote:

Berners-Lee et al.“We believe the RCUK should go ahead and implement its immediate [italics mine] self-archiving mandate, without delay.”

(More trivially, the THES authors name 4 universities, corresponding to one each of 4 of the 8 co-signatories, but omit Southampton, the university of all 4 of the remaining co-signatories, including Sir Tim!)

The last piece of nonsense is this:

THES: “Universities are not obliged to implement a repository system, which costs about £80,000 to set up and about £40,000 a year in maintenance.”

This too is based on a flaw in the current wording of the policy, which actually says that the articles

SH: “should be deposited in an appropriate e-print repository (either institutional or subject-based) wherever such a repository is available to the award-holder.”

But the cost of creating and maintaining a repository is in reality less than 10% of the arbitrary and inflated figures cited by THES.

Stevan Harnad
http://www.ecs.soton.ac.uk/~harnad/

On Sat, 24 Sep 2005, Sally Morris (SM), of the ALPSP, wrote, in the American Scientist Open Access Forum:

SM: "Not sure there is any point continuing this but, for what it's worth, increased citations do not self-evidently equate with increased return on research investment. Those who have ears to hear have, I think, already heard. I will post no more on the topic of fantasy economics."

On the topic of fantasy economics (for those with ears to hear), I quote Sally Morris when she is citing usage and citations on the subject of her own speculations about potential revenue loss for publishers:

SM: "Increasingly, librarians are making use of COUNTER-compliant (and therefore comparable) usage statistics to guide their decisions to renew or cancel journals. The Institute of Physics Publishing is therefore concerned to see that article downloads from its site are significantly lower for those journals whose content is substantially replicated in the ArXiV repository than for those which are not." [See reply.]

SM: "Citation statistics and the resultant impact factors are of enormous importance to authors and their institutions; they also influence librarians' renewal/cancellation decisions. Both the Institute of Physics and the London Mathematical Society are therefore troubled to note an increasing tendency for authors to cite only the repository version of an article, without mentioning the journal in which it was later published." [See reply.]

In other words, when usage and citations are being cited as evidence of hypothetical losses to publishers, they are not fantasy economics. But when they are cited as evidence of actual losses to research and researchers, they are fantasy economics.

As it happens, the only fantasy in all of this is Sally Morris's own fantasy "that RCUK's proposed [mandatory self-archiving] policy will inevitably lead to the destruction of journals." As already pointed out at length (for those with ears to hear), Sally adduces zero evidence in support of her fantasy. All objective evidence to date is for peaceful co-existence between journal publishing and self-archiving.

The rest is not fantasy, but facts, among them the worth of a citation to a researcher

Diamond, A., 1986. What is a citation worth? Journal of Human Resources 21, 200-215.

and, still more important, the return, in number of citations, per pound spent on research by RCUK:

Data-based estimate of 760,000 annual citations (on UK's 130,000 annual articles for RCUK's £3.5 billion pounds invested annually = 0.000217 citations per pound [Source: ISI Web of Science]

Data-based estimate of 50%-250% increase of citations for self-archived articles [Source: OST/ISI study]

Data-based estimate that only 15% of articles are being self-archived today[Source: OST/ISI study]

Therefore: 85% x 50% of 760,000 citations = 323,000 citations lost at 0.000217 citations per pound = £1.5 billion pounds worth of citations lost if not self-archived -- and gained if RCUK mandates self-archiving.

It is a real head-shaker that Sally continues to find subjective imagination-based predictions of revenue loss to publishers as a result of self-archiving to be non-fantasy, while she finds objective data-based estimates of researcher revenue loss as well as losses of the return on research investment to researchers, research and the public to be fantasy. But as Sally will safely say no more on the subject...

Stevan Harnad

---

Update Jan 1, 2010: See Gargouri, Y; C Hajjem, V Larivière, Y Gingras, L Carr,T Brody & S Harnad (2010) “Open Access, Whether Self-Selected or Mandated, Increases Citation Impact, Especially for Higher Quality Research”
Update Feb 8, 2010: See also "Open Access: Self-Selected, Mandated & Random; Answers & Questions"

---

Re: Maximising the Return on the UK's Public Investment in Research

Prior AmSci Topic Thread:
"OA advantage = EA + AA + QB + OA + UA" (Started Sep 2004)

The Open Access (OA) Impact Advantage (currently 50-250%) will shrink as we approach 100% OA. Right now we are at about 15% OA self-archiving and the advantage is in part (no one can say how large a part) a competitive advantage of the minority 15% OA self-archivers (the head-start vanguard) over the laggard 85% non-OA majority.

(Actually, 5% more is OA too, via OA journals, but as the impact advantage is harder to calculate for OA journals -- because we are not comparing within the same journal and year -- we leave it out of these calculations. The same reasoning applies, however.)

That makes it partly a race; and clearly, the race is to the swift and the battle to the strong. The competitive advantage is more reason for an individual, institution or nation (like the UK) to self-archive right now (as the RCUK will, we hope, soon be doing).

The OA impact advantage arises from at least the following 6 component factors, three of them (2,3,5) temporary, three of them permanent (1,4,6):

1. EA: EARLY ADVANTAGE, beginning already at the pre-refereeing preprint stage. Research that is reported earlier can begin being used and built upon earlier. The result turns out to be not just that it gets its quota of citations sooner, but that quota actually goes up, permanently. This is probably because earlier uptake has a greater cumulative effect on the research cycle.

2. (AA): ARXIV ADVANTAGE, the special advantage of self-archiving specifically in Arxiv for physicists, because it is a central point of call: OAI-interoperable Institutional Repositories is likely -- for many reasons -- to supersede this, so it will eventually make zero difference which OAI-compliant IR one deposits in, as access will be through OAI cross-archive harvesters, not directly through individual OAI Archives.

3. (QB): QUALITY BIAS, arsing from article/author self-selection; this does not play a causal role in increasing impact: The higher-quality (hence also higher-impact) articles/authors are somewhat more likely to be self-archived/self-archivers in these early (15%) days of self-archiving: this bias will of course vanish as self-archiving approaches 100%).

4. QA: QUALITY ADVANTAGE, allowing the high-quality articles to compete on a level playing field, freed of current handicaps and biasses arising from access affordability differences. A permanent effect.

5. (CA): COMPETITIVE ADVANTAGE, for self-archived papers over non-self-archived ones, in early (15%) days; this too will of course disappear once self-archiving nears 100%, but at this moment it is in fact a powerful extra incentive, for the low % self-archiving fields, institutions and individuals.

6. UA: USAGE ADVANTAGE: OA articles are downloaded and read three times as much. This too is a permanent effect. (There is also a sizeable correlation between early download counts and later citation counts.)

Of these six component factors contributing to the OA impact advantage, only EA, QA, and UA remain operative in the few fields that are already close to 100% OA, such as Astrophysics and High Energy Physics. Everywhere else, however, the current 15% self-archiving rates still need to do a lot of climbing to reach 100%; so for those individuals, institutions, fields and nations the CA still matters a great deal today. (The UK hence stands to gain the biggest competitive advantage by being the first country to implement a self-archiving mandate.)

Have I overestimated the UK's potential £1.5bn advantage in the longer-term, given the likelihood that other countries will follow suit, thereby cutting down on the CA component? It was partly to minimise this that I based the estimate on the most conservative end of the 50-250% OA impact advantage, underestimating it by using 50%. (It could also be 5 times as great. )

And whereas the Competitive Advantage will indeed shrink and disapper, the Early Advantage, Quality Advantage and Usage Advantage will be going strong. Michael Kurtz has shown that although articles in a 100% OA field (Astrophysics) do not have longer reference lists, hence do not cite more articles overall, they do have three times higher usage rates (UA). So authors can at last find, access, and decide which articles to cite purely on the basis of their relative merit and quality (QA), no longer biassed by the affordability (hence the accessibility) of the journal in which they happen to be published. So whereas the competitive horse-race (for who self-archives to gain the CA first) will be over at 100% OA, the cognitive horse-race (for which researcher finds what earlier: EA) will continue to favour the swift and the strong.

It is hence fair to say that although the annual £1.5 billion pounds-worth of potential impact that the UK is currently losing because it only self-archives 15% of its research output will shrink (as other nations' self-archiving policies catch up), how much it shrinks will then depend only on the true merit of British research rather than either the UK's head-start in self-archiving or the current differential affordability/accessibility of journals.

Stevan Harnad,
Moderator
American Scientist Open Access Forum

Professor of Cognitive Science
Department of Electronics and Computer Science
University of Southampton
SO17 1BJ UNITED KINGDOM

SUMMARY:The United Kingdom is not yet maximising the return on its public investment in research. Research Councils UK (RCUK) spend £3.5 billion pounds annually. The UK produces at least 130,000 research journal articles per year, but if it is worth funding and doing at all, research must be not only published, but used, applied and built upon by other researchers (‘citation impact’). The online-age practice of self-archiving has been shown to increase citation impact by a dramatic 50-250%, but so far only 15% of researchers are doing it spontaneously. Citation impact is rewarded by universities (through promotions and salary increases) and by research-funders like RCUK (through grant funding and renewal) at a conservative estimate of £46 per citation. If we multiply this by the 85% of the UK's annual journal article output that is not yet self-archived, this translates into an annual loss of £2,541,500 in revenue to UK researchers for not having done (or delegated) the few extra keystrokes per article it would have taken to self-archive their final drafts. But this impact loss translates into a far bigger one for the British public, if we reckon it as the loss of potential returns on its research investment. As a proportion of the RCUK’s yearly £3.5bn research expenditure (yielding 130,000 articles x 5.6 = 761,600 citations), our conservative estimate would be 50% x 85% x £3.5.bn = £1.5bn worth of loss in potential research impact(323,680 potential citations lost). The solution is obvious, and it is the one the RCUK is proposing: to extend the existing universal 'publish or perish' requirement to 'publish and also self-archive your final draft on your institutional website'. The time to close this 50%-250% research impact gap is already well overdue. This is the historic moment for the UK to set an example for the world, showing how to maximise the return on the public investment in research in the online era.

The United Kingdom is not yet maximising the return on its public investment in research. Research Councils UK (RCUK) spend £3.5 billion pounds annually. The UK produces at least 130,000 research journal articles per year, but it is not the number of articles published that reflects the return on the UK’s investment: A piece of research, if it is worth funding and doing at all, must be not only published, but used, applied and built upon by other researchers. This is called ‘research impact’ and a measure of it is the number of times an article is cited by other articles (‘citation impact’).

But in order to be used and built upon, an article must first be accessed. A published article is accessible only to those researchers who happen to be at institutions that can afford to subscribe to the particular journal in which it was published. There are 24,000 journals in all, and most institutions can only afford a small fraction of them. In paper days, authors used to supplement this paid access to their articles by mailing free reprints to any would-be users who wrote to request them. The online age has made it possible to provide free ‘eprints’ (electronic versions of the author’s draft) to all potential users who cannot afford the journal version by ‘self-archiving’ them on the author’s own institutional website.

The online-age practice of self-archiving has been shown to increase citation impact by a dramatic 50-250%, but so far only 15% of researchers are doing it. A recent UK international survey has found that 95% of authors would self-archive – but only if their research funders or their institutions required them to do it (just as they already require them to ‘publish or perish’). The solution is hence obvious:

After lengthy deliberations first initiated in 2003 by the UK Parliamentary Select Committee on Science and Technology, RCUK have proposed to adopt a policy requiring UK researchers to deposit, on their university's website, the final author's draft of any journal article resulting from RCUK-funded research. The purpose of the proposed policy would be to maximise the usage and impact of UK research findings by making them freely accessible on the web ("open access") for any potential users in the UK and worldwide who cannot afford paid access to the published journal version. How does this maximise the return on the UK public investment in research?

It is not possible to calculate all the ways in which research generates revenue. A good deal of it is a question of probability and depends on time: Although everyone thinks of an immediate cure for cancer or a cheap, clean source of energy as the kind of result we hope for, most research progresses gradually and indirectly, and the best estimate of the size and direction of its progress is its citation impact, for that reflects the degree of uptake of research results by other researchers, in their own subsequent research. Citation impact is accordingly rewarded by universities (through salary increases and promotion) and by research-funders like RCUK (through grant funding and renewal); it is also rewarded by libraries (through journal selection and renewal, based on a journal's average citation "impact factor"). Counting citations is a natural extension of the cruder measure of research impact: counting publications themselves ("publish or perish").

If citations are being counted, it is natural to ask how much they are worth.

The marginal dollar value of one citation was estimated by Diamond in 1986 to range from $50-$1300 (US), depending on field and number of citations. (An increase from 0 to 1 citation is worth more than an increase from 30 to 31; most articles are in the citation range 0-5.) If we convert from dollars to UK pounds sterling (£27-£710) and update by 170% for inflation from 1986-2005, this yields the range £46-$1207 as the marginal value of a UK citation today. Self-archiving, as noted, increases citations by 50-250%, but, as also noted, only 15% of the articles being published are being self-archived today.

We will now apply only the most conservative ends of these estimates (50% citation increase from self-archiving at £46 per citation) to the UK's current annual journal article output (and only for the approximately 130,000 UK articles a year indexed by the Institute for Scientific Information, which covers only the top 8000 of the world's 24,000 journals). If we multiply by the 85% of the UK's annual journal article output that is not yet self-archived (110, 500 articles), this translates into an annual loss of £2, 541, 500 in revenue to UK researchers for not having done (or delegated) the few extra keystrokes per article it would have taken to self-archive their final drafts.

But this impact loss translates into a far bigger one for the British public, if we reckon it as the loss of potential returns on its research investment. As a proportion of the RCUK’s yearly £3.5bn research expenditure (yielding 130,000 articles x 5.6 = 761,600 citations) , our conservative estimate would be a 50% x 85% x £3.5.bn = £1.5bn worth of loss in potential research impact (323,680 potential citations lost). And that is without even considering the wider loss in revenue from potential practical applications and usage of UK research findings in the UK and worldwide, nor the still more general loss to the progress of human inquiry.

The solution is obvious, and it is the one the RCUK is proposing: to extend the existing universal 'publish or perish' requirement to 'publish and also self-archive your final draft on your institutional website'. Over 90% of journals already endorse author self-archiving and the international author survey -- plus the actual experience of the two institutions that have already adopted such a requirement (CERN and University of Southampton ECS ) -- has shown that over 90% of authors will comply.

The time to close this 50%-250% research impact gap is already well overdue. This is the historic moment for the UK to set an example for the world, by showing how to maximise the return on the public investment in research in the online era.

How self-archiving increases citation impact:
http://opcit.eprints.org/oacitation-biblio.html

How much a citation is worth:
http://www.garfield.library.upenn.edu/essays/v11p354y1988.pdf

How much time and effort is involved in self-archiving
http://eprints.ecs.soton.ac.uk/10688/

RCUK self-archiving policy proposal:
http://www.rcuk.ac.uk/access/index.asp

Directory of publishers' policies on author self-archiving:
http://romeo.eprints.org/

JISC user survey on self-archiving:
http://eprints.ecs.soton.ac.uk/11006/