Self-Archiving Mandates

Many thanks to Robert Kiley of the Wellcome Trust (WT) for responding to my recommendations on optimising the Trust's Open Access Mandate, but unfortunately Robert only repeats points with which I am already very familiar, while passing in silence over the actual substantive points I have raised, repeatedly, ever since the Wellcome Trust mandate was adopted 5 years ago (and even earlier than that).

Let me summarise the (many) positive aspects of the Wellcome Trust Mandate before specifying, once again, the negative aspects that can so easily be fixed.

POSITIVE ASPECTS OF THE WELLCOME TRUST (WT) OPEN ACCESS (OA) MANDATE:

(1) The WT OA mandate five years ago (2004) was the world's first funder mandate and helped to inspire many others.

(2) The WT OA Mandate not only came earlier than the NIH policy, but it was a mandate (requirement) from the very outset, whereas the NIH policy lost 2 years by being initially formulated as a request rather than a requirement.

(3) The WT in general (and Robert Kiley and Robert Terry in particular) have worked valiantly and tirelessly to promote OA and OA mandates during the ensuing 5 years.

NEGATIVE ASPECTS:

(1) The WT OA Mandate stipulates direct deposit in PubMedCentral (PMC) instead of institutional deposit and central harvesting; this counterproductive constraint has since been imitated by other funders following WT's example. Institutions are the universal providers of all OA output, funded and unfunded, across all disciplines. If funders mandate institutional deposit, they encourage and reinforce universal adoption of institutional OA mandates (and gain a powerful ally in monitoring and ensuring compliance); if funders instead mandate central deposit, they discourage and compete with universalizing the adoption and implementation of institutional mandates.

(2) The WT OA Mandate permits a delay (embargo) of deposit for up to a year after publication. If funders instead mandate immediate institutional deposit, with no exceptions, the institutional repository's "author email eprint request" Button can provide 'Almost-OA' to would-be users while access to the deposit itself is embargoed; otherwise researcher access, usage and impact are needlessly lost during the embargo.

(3) The WT OA Mandate allows the option of publishers doing the PMC deposits in place of WT's fundees. This not only makes fundee compliance vaguer and compliance-monitoring more difficult, but it further locks in publisher embargoes (with no possibility of authors providing Almost-OA to tide over user needs during the embargo period) and further discourages convergent institutional mandates.

All three of these dysfunctional implementational details can be easily and fully remedied by simply specifying that deposit should be in the fundee's IR (or, if the fundee's institution does not yet have an IR, in an interim IR such as DEPOT) immediately upon acceptance for publication. That's all. The negatives are thereby immediately nullfied and the WT funder mandate becomes the optimal model for adoption by other funders, as well as a strong impetus to the adoption of complementary deposit mandates by institutions.

Now I reply to Robert's responses:

On 9-Dec-09, at 12:20 PM, Robert Kiley (Wellcome Trust) wrote:

RK: "Stevan

"Your last post made a number of comments about PMC, UKPMC and the Wellcome Trust, which I'd like to respond to:

"1. Empty repositories?

"[1a] You indicate that "PMC (or its emulators)" are bereft of content. This is not the case. PMC currently has around 1.9 million full-text articles. [1b] Looking at the compliance levels for funder mandate, we see that around 43% of Trust funded research is made available through UKPMC in line with our mandate. We are actively working to increase this figure. [1c] It is also worth noting that, because of our support for gold OA, a significant proportion of these articles are freely available at the time of publication."

[1a] Let me define an "empty repository": it's a repository that captures 0-15% of its total annual target content.

Why? Because 15% is the default baseline for spontaneous, unmandated deposit. You are not doing better than the default baseline if you are not capturing significantly more than 15% of your repository's total annual target content.

What percentage of the global annual output of peer-reviewed bio-medical journal articles -- per year -- do you think that PMC's grand total of 1.9 million articles represents?

It's only that (annual) figure (minus 15%) that tells you how non-empty a repository is, not the grand total -- and certainly not the grand total for a central repository whose denominator (the annual amount by which you must divide the annual deposits to calculate the percentage) consists of all the annual biomedical research output on the planet (or even all annual biomedical research originating from the US).

This is what I called the "denominator fallacy" in my prior posting.

[1b] In contrast, PMC is capturing 43% of WT's target content in 2009. That's certainly better than 15% (or NIH's meagre 5% before they upgraded their deposit request to a requirement). But that's mandated content.

And 5 years after the adoption of the WT mandate, 43% isn't really that good either. Indeed it was only last year that WT itself was about its low compliance rate:

In contrast, institutions that have adopted and implemented deposit mandates are doing a good deal better than that: Over 60% and well on the road to 100% about 2 years after adoption.

And the reason for the successful institutional mandates' success is quite evident: Institutions are the quotidial employers of their researchers, not just their occasional funders. Institutions have annual performance reviews for salary, promotion and tenure. They are in a position to mandate -- as the University of Liege has notably done -- that the procedure for submitting one's annual publications for performance review is henceforth to deposit them in the in the institution's IR; otherwise the publications will be invisible. This is a simple internal bureaucratic requirement, rather like the ubiquitous transition from submitting on paper to submitting online.

Institutions, as we all know, are also very eager that their researchers should receive research funding. Hence institutions are eager to be involved in helping researchers prepare grant applications as well as to ensure that they fulfill all grant requirements if funded. Fundees' institutions are hence funders' natural allies in ensuring and monitoring compliance with the funder's deposit mandate -- as long as the designated deposit locus is institutional. Moreover, funders mandating institutional deposit of the articles resulting from the research they fund, and institutions' involvement in ensuring compliance, also encourages institutions to go on and mandate deposit of the rest of their research output too.

But if it is instead stipulated by the funder -- and (I have to repeat this each time) stipulated for no good reason at all, since it confers no advantages whatsoever, either functional or practical, over institutional deposit, only disadvantages -- that the deposit must be central, then the fundee's institution is in no better position than the funder to ensure and monitor compliance. In addition, the institution then has the opposition of its researchers to contend with, if ever the institution contemplates adopting a deposit mandate of its own: Researchers (quite understandably, and justifiably) do not want to have to deposit willy-nilly in divergent multiple loci, institution-internal as well as institution-external. Add to that the further confusion added by the fact that fundee "compliance" can be fulfilled by publishers depositing in PMC instead of fundees, and after a one-year embargo, and you have both grant fulfillment conditions and mandate incentive conditions as ill-served and as hard to monitor as they could possible be.

And, again, for no good positive reason whatsoever.

[1c] Yes, the WT money that could have been spent on supporting more research, when it is instead redirected to paying for Gold OA publication, does increase the uptake of Gold OA somewhat. But is that the objective? Or is the objective rather to increase OA as much as possible -- which is what the Green OA deposit mandate itself would do, if compliance were indeed insured and monitored.

As to the best way to contend with the 1-year embargo at this point -- that's up to WT to decide. 63% of journals already endorse making institutional deposits OA immediately upon publication. If WT finds it a better use of its research money to pay for immediate Gold OA for the remaining 37% (rather than relying on the Institutional Repositories' "email eprint request" Button to allow the author to provide almost-immediate, "Almost OA" during the embargo), that's a judgment call. But it's not an argument for insisting on central deposit rather than institutional.

Note, though, that WT is on the side of the angels in having mandated OA already, rather than just offering to subsidize Gold OA. The trouble is that the "mandated Green OA deposit plus subsidized Gold OA option" policy is far less adoptable, for example, by poorer funders, or funders more anxious to use their scarce funds to fund more research rather than to subsidize Gold OA publishing. This is especially today, when OA can be had without cost, by mandating Green OA and just letting subscriptions continue to pay for publishing. And this remains true until/unless Green OA ever makes subscriptions no longer sustainable. Then (and only then) a transition to Gold OA will be payable out of institutions' windfall subscription cancellation savings -- and for a lot less than today's Gold OA's pre-emptive asking price, since the only thing left to pay for then will be peer review -- without the need to syphon away any additional research money.

Moreover, the example of pre-emptive payment for Gold OA has inspired another nonstarter, from funders and institutions that are not yet on the side of the angels: They are redirecting scarce research or institutional funds today, needlessly, to pay for Gold OA today without even first mandating OA, as WT has done. That's the worst of all possible worlds (and encouraged by the example of needless and ineffectual profligacy on the part of others, even when they do couple it with a Green OA mandate too...)

RK: "2. "National PMC's are a joke"

"The idea that UKPMC is a "joke" is not shared by the Wellcome Trust, or indeed the other 7 biomedical research funders in the UK who have established (and funded) this repository.

"You suggest that UKPMC simply holds UK content. This is not the case. It holds all the PMC content, but each "locality" (UK, Canada etc) can build services in top of that to ensure that it meets the needs of the research community we are trying to serve."

That's all splendid, and not the joke at all.

The joke is the notion that all these countries need a national PMC as the place to mandate deposit!

Of course all manner of harvesting services can be superadded to any number of harvested collections -- national, disciplinary or what have you. That's not the joke. The joke is that national funders are slavishly adopting the wrong-headed notion that they, like NIH/PMC, need their own national, central place to deposit their mandated contents -- instead of doing what NIH/PMC should have done in the first place (and should convert as soon as possible to doing now), which is to mandate institutional deposit, and harvest/import from there to any central collections or services they may wish to provide.

RK: "In January 2010 we will be launching a new UKPMC site which will offer users:

"A) A single access point to around 20 million bibliographic records - drawn from PubMed, Biological Patents, Agricola and Clinical Guidelines databases - as well as the 1.5m+ full text articles in UKPMC"

Splendid. But now please explain to me why the worthy and welcome goal of offering users a single access point for all these worthwhile contents needs to be reached by requiring UK-funded authors to deposit in UKPMC to fulfill their deposit mandates, rather than in their own IRs?

(And I do hope you won't reply that it's in order to accommodate the publishers, who need to deposit in UKPMC! Those articles are by UK fundees too! Let those fundees simply, and uniformly, deposit all their (mandated) articles in their own IRs, regardless of whether they are published in paid Gold OA journals, free Gold OA journals, subscription journals with OA embargoes, or subscription journals without OA embargoes: One size fits all, funders and institutions alike, across nations as well as disciplines, for both funded and unfunded research: Deposit institutionally.

RK: "B) Additional, local content. This includes guidelines from NICE and other NHS bodies, plus relevant (i.e. biomedical) theses derived from EthOS. So, by way of example, when you search the new site for say "management of stroke" you will be presented with relevant PubMed citations, full-text articles, UK clinical guidelines etc in one search."

All very valuable stuff -- but nothing in this is contingent on mandating central deposit. Harvesting of distributed content is the name of the game, in the online era. (We don't deposit directly in Google either. Google harvests distributed locally hosted content.)

RK: "C) New citation services. For every article (be it full text or just the bibliographic citation) you will be able to see all the papers which that paper cited, as well as all the other articles which cite that paper."

Lovely, stuff but nothing to do with the only point at issue here, which is whether or not mandating funders have any good reason to require divergent central deposit instead of convergent institutional deposit. (The latter might even help accelerate the institutional mandates you'll need to turn those bibliographic citations into full- texts -- at least for living authors...)

RK: "D) New text-mining services. Our colleagues at EBI and NaCTeM have build tools to textmine the content in UKPMC. In the first release (January 2010) users will be able to see in a "summary box" which will provide details of what genes/proteins, organisms etc are discussed in the paper they are viewing. Over the next 18 months this textmining functionality will be developed further in include chemical compounds, disease names etc etc."

Again very valuable, and again completely orthogonal to the question of locus of deposit -- which, to repeat, is the only one I keep banging on about.

(If funders wish to mandate deposit in specific formats, such as XML, they can do that equally well regardless of locus of deposit -- though I would not myself recommend over-constraining format requirements at this early stage, when it is the articles that are missing and sorely needed, rather than the documents already being accessible, and only the right format being sorely needed. And if, in contrast, the deposit tagging and format are being enriched by some other central service, rather than the author, that too can be done irrespective of locus of deposit, again through central importation or harvesting.)

RK: "The "franchise" model that PMC uses is akin to that developed for the human genome project inasmuch as content is mirrored to a number of sites (e.g. NCBI, Sanger, and DDBJ) but each centre develops their own interface to this content. So, the core content at PMC, UKPMC and PMC Canada is identical -- but each centre will develop their own valued-added services."

The "franchise" model is equally compatible with central deposit and with distributed institutional deposit and central harvesting...

RK: "The UKPMC Funders Group - led by the Wellcome Trust - are, with the support of European partners, exploring the possibility of creating a single, Europe-wide OA repository for peer reviewed biomedical research papers -- a Europe PMC. A workshop to discuss this is taking place on the 2nd December at the Berlin 7 meeting."

All these collections and re-collections of biomedical research papers and services are welcome, but have nothing to do with mandated deposit locus.

RK: "3. Why the Trust favours the author-pays model

"The Wellcome Trust has always argued that:

"A) dissemination costs are simply research costs
"B) publishers add-value to the research article"

Fine. Call the costs what one may: those publication costs and values are being paid for in full by subscriptions today. What is missing is access to those publications (for those whose institutions can't afford the subscription costs). Green OA provides that access, in full. And mandates provide Green OA, with no extra cost. It's up to WT if they want to spend more research money on reforming publishing, rather than just mandating that their fundees provide the access that is missing. But let paying for Gold OA not be mistaken or misrepresented as the fastest, cheapest or surest way to provide the missing access. It is simply using research money to try to reform publishing.

Nor can WT represent favouring the payment for Gold OA with scarce research funds over providing Green OA at no extra cost as something that favours OA: It does not. It simply diverts research money to pay pre-emptively for Gold OA, when it is not even needed; it disfavours the cost-free Green way of providing OA; and it sets an unfortunate example for other funders contemplating what they can do to increase OA.

RK: "It follows, therefore, that these costs have to be met -- and that is what the Wellcome Trust (and others) do."

It only follows that those costs have to be met if there is also a reason why research money has to be spent on reforming publishing today, when what is really needed today, urgently, is more research access, not less research money, nor publishing reform.

(Publishing reform will be needed, and will happen, if and when -- and only if and when -- universal Green OA makes subscription journal publishing unsustainable. But if and when universal Green OA ever does so, it will, by the very same token, also release the subscription cancellation funds to pay for Gold OA without the need to redirect scarce research funds. Indeed, universal Green-OA-driven subscription collapse will also force journal publishing to cut obsolete products and services (such as the paper edition, the online edition, access- provision and archiving) and their associated costs, downsizing to just the service of peer review. The distributed network of institutional repositories (and any harvester services thereover) will do the access-provision and archiving. So instead of receiving less research funding, researchers' institutions will enjoy a surplus from their annual windfall subscription cancellation savings.

RK: "It is also worth pointing out that when an APC fee is met, the Trust requires the publisher to provide a number of services:

"A) Deposit the final version - marked up to an XML standard - directly into PMC, where it must be made freely available at the time of publication. [So, no embargo, no "email request buttons", and no work on behalf of the author.]"

If you offered your fundees the choice (without fear or favour) of spending the WT research money on research or spending it to spare themselves the few keystrokes it takes to deposit their postprints (63%) and fulfill email eprint requests (37%), do you have any doubt as to what choice they would make? Especially if the designated locus of deposit were institutional, and hence they were already depositing their unfunded research that way...

RK: "B) Attach a licence to these articles, thus ensuring that anyone who want to re-use the work (e.g. text-mining, creating translations, re-using for different audiences etc) can do so. Whether such rights extend to author manuscripts is, at best, unclear."

More important, those rights and re-uses are completely superfluous. What's urgently needed (and prominently missing) today is online access to the articles, free for all. What comes with that territory is the capability of any user to search, link, read online, download, print-off, store and data-crunch a personal copy. In addition, harvesters like google can and will harvest and invert it. "Different audiences" can use the same URL. Translations (for the lucky few where it's wanted) can, as always, be handled on a case by case basis.

Let's talk again about any "text-mining" beyond this when there's enough OA text to make it worth talking about.

RK: "C) These articles can also be included in the OA subset, thus allowing institutions (and others) to harvest, via OAI, relevant full-text content."

That sentiment is not unworthy of Marie Antoinette! "Let the institutions harvest back their very own content, because we have elected to mandate that it must be deposited institution-externally." (Harvesting, for the record, is something central harvesters do over distributed providers of the content, not the reverse, i.e., not distributed providers of the content, harvesting back their own content from an institution-external central deposit locus where their own content providers have been required to deposit it, instead of depositing it institution-internally in the first -- and only -- place. (That's like saying: let everyone deposit their content in google, and then harvest it back if they want to host it locally.)

SUMMARY: Not one substantive reason has been given for WT's continuing insistence on central deposit rather than institutional deposit (plus central harvesting). Nor has a compelling reason been given for favouring paid Gold OA over free Green OA. (But if WT were, as I hope, to go on to mandate institutional deposit, paid Gold would become a minor matter, because as more institutions added their institutional mandates to WT's and other funders' mandates, the absurdity (and non-scaleability) of paying pre-emptively for Gold OA today, rather than just depositing for Green OA at this time -- whilst the potential funds to pay for Gold OA are still locked into subscriptions that are paying for subscription publication in full -- would become more and more obvious. The confusion and uncertainty about this today are simply a result of the extreme sparseness of OA content -- whether Green or Gold -- today [c. 15%], as well as the extreme rarity of OA mandates [c. 100/10,000].)

Stevan Harnad
American Scientist Open Access Forum

Today (Dec 10 2009) begins the comment period for President Obama's OSTP Public Forum on How Best to Make Federally Funded Research Results Available For Free. Comments will be in three phases:

Implementation (Dec. 10 to 20): Which Federal agencies are good candidates to adopt Public Access policies? What variables (field of science, proportion of research funded by public or private entities, etc.) should affect how public access is implemented at various agencies, including the maximum length of time between publication and public release?

Features and Technology (Dec. 21 to Dec 31): In what format should the data be submitted in order to make it easy to search and retrieve information, and to make it easy for others to link to it? Are there existing digital standards for archiving and interoperability to maximize public benefit? How are these anticipated to change.

Management (Jan. 1 to Jan. 7): What are the best mechanisms to ensure compliance? What would be the best metrics of success? What are the best examples of usability in the private sector (both domestic and international)? Should those who access papers be given the opportunity to comment or provide feedback?

Please do comment at the OSTP site (you'll need to register first).

My own comments follow:

It would be a great benefit to research progress in the US as well as worldwide if the US were to require not only NIH-funded research journal articles to be made freely accessible to all users online, but all federally funded research journal articles.

BENEFITS: The benefits of making all US publicly funded research publicly accessible online would not only be in the fact that all tax-payers (and not just those who can afford to subscribe to the journal in which it was published) will be able to read and use the research their taxes paid for, but, even more important, it will allow all researchers (and not just those whose institutions can afford to subscribe to the journal in which it was published) to read, use, apply and build upon all those research findings, again to the benefit of the public that funded them, and for the future research advances for the sake of which research is funded, conducted and published.

WHICH RESEARCH? Which federally funded research should be made publicly accessible online? Start with all research that is fully funded federally, in all scientific, technical and scholarly fields, and then work out agreements in the case of joint private funding. Most private funders will likewise want to ensure maximal usage and impact for the research they have funded. If they want it published at all, they will also want access to it to be maximized.

TIMING OF DEPOSIT: Allowable embargo time should be minimal, but, far more important, the requirement should be to deposit the final, peer-reviewed draft, immediately upon acceptance for publication, in the author's institutional repository, without exception. 63% of journals already endorse making the deposit Open Access immediately. For the remaining 37%, the deposit can be made Closed Access, with only its metadata (authors, date, title, journal, abstract) accessible publicly during the allowable embargo. That way researchers can send the author a semi-automatic email eprint request for an individual copy to be used for research purposes. This will tide over research needs during any embargo.

LOCUS OF DEPOSIT: It is extremely important to require institutional instead of central deposit (which is what several funders require now, e.g., NIH requires central deposit in PubMedCentral, PMC). Institutional deposits can be easily and automatically harvested or imported into central collections and services like PMC (or Scirus or OAIster or Citeseer, or, for that matter, Google Scholar and Google).

The NIH requirement to deposit in PubMedCentral (PMC) is an extremely counterproductive handicap, needlessly slowing down the growth of public access for no good reason at all. Institutions (universities and research institutes) are the universal providers of all research output, funded and unfunded, across all fields. If funders mandate institutional deposit, they encourage and reinforce universalizing the adoption of institutional public access mandates across all their fundees' institutions (and they gain a powerful ally in monitoring and ensuring compliance with the funder mandates).

But if funders instead require central deposit, they discourage and compete with universalizing the adoption and implementation of institutional public-access requirements. Nor is there any advantage whatsoever -- functional, technical or practical -- to requiring central rather than institutional deposit; it only creates needless obstacles to the universal adoption of public access and public access mandates for all research output. (It's rather like web hosts depositing their web pages directly in google, instead of hosting them locally and just letting google harvest them.)

WHO DEPOSITS? The current NIH public access policy allows the option of publishers doing the PMC deposits in place of NIH's fundees. This not only makes fundee compliance vaguer and compliance-monitoring more difficult, but it further locks in publisher embargoes (with less scope for authors providing individual access to researchers during the embargo) and it further discourages convergent institutional mandates (with the prospect of researchers having to do multiple deposit for the same paper, institution-internal and institution-external). The ones responsible for ensuring that the deposit is made, immediately upon acceptance for publication, are the fundee and the fundee's institution, by monitoring the deposits in their own institutional repository. Publishers should be out of the loop.

DEPOSIT WHAT? There is no need at all to be draconian about the format of the deposit. The important thing is that the full, peer-reviewed final draft should be deposited in the fundee's (OAI-compliant) institutional repository immediately upon acceptance for publication. A preference can be expressed for XML format, but any format will do for now, until the practice of immediate Open Access deposit approaches global universality (at which time it will all converge on XML as a natural matter of course anyway).

It would be a needless handicap and deterrent to insist on any particular format today. (Doc or Docx will do, so will HTML or PDF or any of the open formats.) Don't complicate or discourage compliance by gratuitously insisting on more than necessary at the outset, and trust that as the practice of public access provision and usage grows, researchers will converge quite naturally on the optimal format. And remember that in the meanwhile the official published version will continue to be generated by publishers, purchased and stored by subscribing institutions, and preserved in deposit library archives. The public-access drafts are just supplements for the time being, not substitutes, deposited so that it is not only paying subscribers who can access and use federally funded research.)

MONITORING COMPLIANCE: What are the best mechanisms to ensure compliance? To require deposit in the fundee's institutional repository immediately upon acceptance for publication. Fundees' institutions are already co-responsible for compliance with funders' application and fulfillment conditions, and already only too eager to help. They should be made responsible for ensuring timely compliance with the funder's deposit requirement. It can also be made part of the grant requirement that the funder must be notified immediately upon deposit by being sent the deposit's URL, so it can be linked or imported for the funder's records and/or harvested by the funder's designated central repository (e.g. PMC).

METRICS OF SUCCESS: Institutions already have an interest in monitoring the usage and impact of their research output, and their institutional repositories already have means for generating usage metrics and statistics (e.g., IRStats). In addition there are now central means of measuring usage and impact (free services such as Citeseer, Citebase, Publish-or-Perish, Google Scholar and Google Books, as well as fee-based ones such as SCOPUS and Thompson-Reuters Web of Science). These and other rich new metrics will be available to measure success once the deposit requirements are adopted, growing, and supplying the content from which these rich new online metrics are extracted. Which of the new metrics proves to be the "best" remains to be tested by systematically assessing their predictive power and their correlation with peer evaluations.

COMMENT AND FEEDBACK: Once the research content is openly accessible online, many rich new tagging, commenting and feedback mechanisms will grow quite naturally on top of them (and can also be provided by central harvesters and services commissioned by the funders themselves, if they wish, or the metrics can simply be harvested from other services for the funder's subset of their content).

PRIVATE SECTOR USABILITY: Metrics will not only make it possible for deposit rates, downloads, citations, and newer metrics and their growth to be measured and monitored, but it will also be possible to sort uptake metrics into those based on public access and usage, researcher access and usage, and industrial R&D and applications access and usage. But the urgent priority is first to provide the publicly accessible research content on which all these uptake measures will be based. The measures will evolve quite naturally once the content is globally available.

---

For more detailed guidelines on optimizing OA mandates (what to mandate depositing, where and when to mandate deposit, and how to integrate institutional and funder mandates), see 1, 2 & 3).

All federal agencies that fund scientific, technical and scholarly research should require fundees to make the resulting peer-reviewed articles to be made freely accessible online (”Open Access”).

There is no objective reason why any publicly funded research that is published in peer-revewed journals should be accessible only to subscribers.

The costs of providing free online access are minimal. Fundees should be required to make their articles publicly accessible by depositing them in their own institutions’ OAI-compliant Open Access Repository.

The deposit should be made immediately upon acceptance for publication. For the minority of publishers who do not yet endorse making the deposit Open Access immediately, an embargo of 6 months can be allowed during which only the deposit’s metadata are openly accessible but the author can provide individual eprints for researcher purposes in response to individual user email requests mediated by the institutional repository software.

All empirical data indicate that the optimal embargo is no embargo. If there is any access embargo at all, the result is needlessly lost research usage and impact. There is no field of science or scholarship, fast or slow, that benefits — or fails to lose — from denying access to peer-reviewed, published results once they have been accepted for publication.

The only version of a paper that needs to be made freely accessible to all users online is the author’s peer-reviewed final draft, immediately upon acceptance for publication. There are no advantages at all to later versions of the paper, only disadvantages, because fewer publishers endorse making their proprietary PDF freely accessible. Eventually the author’s final draft can be in XML format, but for now any format (doc, docx, html, pdf, etc.) will do.

Only mandatory deposit is successful and effective. All other alternatives fail to generate deposits above the spontaneous unmandated level of about 15%. See Arthur Sale’s studies.

The only relevant structural characteristics of a public access policy are the ones already mentioned: mandate deposit of the fundee’s peer-reviewed final draft in the fundee’s institutional repository immediately upon acceptance for publication. (Preferable format, but not obligatory: XML.) Compliance should be monitored by the fundee’s institution, as part of the grant’s fulfillment condition. Maximum permissible embargo before making the deposit Open Access: 6 months (but preferably no embargo). Deposits can be harvested to central collections and services like PubMed Central, but deposit should be institutional and not central, in order to reinforce and facilitate complementary institutional mandates to deposit unfunded research.

Stevan Harnad
American Scientist Open Access Forum