Open Access Archivangelism

In the Society for Scholarly Publishing's Scholarly Kitchen, Rick Anderson complains of "errors and misinformation” in the ROARMAP registry of OA mandates and David Crotty calls for publishers to provide this service instead ("I think the publishing community would be better served by creating (and supporting) a resource of this kind").

ROARMAP is a registry for institutional and funder OA policies:

X-Other (Non-Mandates) (86)
Proposed Institutional Mandates (6)
Proposed Sub-Institutional Mandate (4)
Proposed Multi-Institutional Mandates (5)
Proposed Funder Mandates (12)
Institutional Mandates (202)
Sub-Institutional Mandates (43)
Multi-Institutional Mandates (9)
Funder Mandates (87)
Thesis Mandates (109)

The distinction between a mandate and a non-mandate is fuzzy, because mandates vary in strength.

For a classification of the ROARMAP policies in terms of WHERE and WHEN to deposit, and whether the deposit is REQUIRED or REQUESTED, see El CSIC’s (la Universitat de Barcelona, la Universitat de València & la Universitat Oberta de Catalunya) MELIBEA.

For analyses of mandate strength and effectiveness, see:

Gargouri, Y., Lariviere, V., Gingras, Y., Brody, T., Carr, L., & Harnad, S. (2012). Testing the finch hypothesis on green OA mandate ineffectiveness. arXiv preprint arXiv:1210.8174.

Gargouri, Yassine, Larivière, Vincent & Harnad, Stevan (2013) Ten-year Analysis of University of Minho Green OA Self-Archiving Mandate (in E Rodrigues, A Swan & AA Baptista, Eds. Uma Década de Acesso Aberto e na UMinho no Mundo).

Further analyses are underway. For those interested in analyzing the growth of OA mandate types and how much OA they generate, ROARMAP and MELIBEA, which index OA policies, can be used in conjunction with ROAR and BASE, which index repository contents.

"Leave providing the OA to us..."

If there is any party whose interests it serves to debate the necessary and sufficient conditions for calling an institutional or funder OA policy an OA “mandate,” it’s not institutions, funders or OA advocates, whose only concern is with making sure that their policies (whatever they are called) are successful in that they generate as close to 100% OA as possible, as soon as possible.

The boundary between a mandate and a non-mandate is most definitely fuzzy. A REQUEST is certainly not a mandate, nor is it effective, as the history of the NIH policy has shown. (The 2004 NIH policy was unsuccessful until REQUEST was upgraded to REQUIRE in 2007.)

But (as our analyses show), even requirements come in degrees of strength. There can be a requirement with or without the monitoring of compliance, with or without consequences for non-compliance, and with consequences of varying degrees. Also, all of these can come with or without the possibility of exceptions, waivers or opt-outs, which can be granted under conditions varying in their exactingness and specificity.

All these combinations actually occur, and, as I said, they are being analyzed in relation to their success in generating OA. It is in the interests of institutions, funders and OA itself to ascertain which mandates are optimal for generating as much OA as possible, as soon as possible.

I am not sure whose interests it serves to ponder the semantics of the word “mandate” or to portray as sources of “errors and misinformation” the databases that are indexing in good faith the actual OA policies being adopted by institutions and funders.

(It is charges of "error and misinformation" that sound a bit more like propaganda to me, especially if they come from parties whose interests are decidedly not in generating as much OA as possible, as soon as possible.)

But whatever those other interests may be, I rather doubt that they are the ones to be entrusted with indexing the actual OA policies being adopted by institutions and funders -- any more than they are to be entrusted with providing the OA.

I think it is not only appropriate but essential that services like the University of Southampton's ROAR, ROARMAP, the Universities of Barcelona, Valencia and Catalunya's MELIBEA and University of Bielefeld's BASE are hosted and provided by scholarly institutions rather than by publishers. I also think the reasons for this are obvious.

"Leave providing the OA to us..." 2

On the charitable assumption that some Schol Kitch readers are researchers, interested in maximizing OA, rather than just publishers interested in persuading the research community to “leave providing the OA [and the OA mandate information] to us [publishers],” here is some further information about OA mandates:

The objective of OA mandates is to generate as much OA as possible, as soon as possible. There are many different mandate models, varying in strength and success.

The Emory and OSU mandates are variants of the Harvard default copyright-reservation mandate model. This is not a very strong mandate model, because in principle it allows opt-out on an individual case-by-case basis. The opt-out rate at Harvard is only about 5%, however.

The version of the Harvard model adopted by Harvard FAS was upgraded so the opt-out applies only to the copyright-reservation clause, not to the immediate-deposit clause, but it still has no provisions for monitoring compliance, and no consequences for non-compliance. The Harvard FAS deposit rate is hence not much higher than the opt-out rate — despite the fact that about 60% of Harvard’s annual research article output is being made freely accessible somewhere on the web within a year of publication.

The UK has stronger mandates, such as the Southampton ECS mandate, but the strongest and most effective mandate model is the Liege/FNRS/HEFCE model in which immediate repository deposit (though not necessarily immediate OA) is officially designated as the submission mechanism for both research performance evaluation and research funding. Its annual deposit rate is over 80% and still rising. As a consequence, other institutions are now upgrading to this mandate model.

All these parametric variations of OA mandates are currently being systematically tested statistically for their success in generating OA in ongoing studies. ROARMAP is just a database for registering and linking the policies themselves, not for classifying or testing them. MELIBEA classifies the policy details in more detail, but there is no point trying to catalogue every possible mandate nuance in an index. These are not publisher copyright-restriction details: they are pragmatic institutional and funder policies, intended to generate as much OA as possible, as soon as possible. And their current full text is always linked and available together with each mandate’s specific details.

I have already posted some of the references for this ongoing research above. Anyone interested in mandate strength and success (rather than just in decrying “errors and misinformation” and in calling into question the research community’s competence to adopt and index its own OA policies) can consult these and other references — or, better still, they can use ROAR, ROAR, MELIBEA and BASE to do further analyses. All findings are welcomed by all who are interested in reaching 100% OA as soon as possible.

---

David Wojick: "The question is whether this upgrade will include accurately describing the rule systems being collected, along the lines Rick presents here, beginning with not calling them mandates. It might be better if publishers did this as a service to their authors. The number of variables in these rules may or may not be large but the number of combinations is probably huge, as is the potential for confusion. Thus accurate representation is not trivial."

Reme Melero's much-improved directory MELIBEA already does this classification extremely well. The most relevant parameters are:

requirement?
opt-out (waiver)?
when?
where?
version?
embargo?
copyright reservation?

This generates a weighted estimate of mandate strength. The weighting parameters will have to be adjusted according to actual studies of the correlation between these parameters and deposit rate. This is what Dr. Swan's database will add, together with growth charts.

So, no discernible need for a version from "publishers... as a service to their authors" -- but they're welcome to do it, as I've mentioned. All these databases are OA, harvestable and remixable.

I suggest, though, that the ScholKitch chefs stop fussing about the word "mandate": A simple dictionary look-up will show that it is polysemous (as between permitting, sanctioning and requiring). So the pertinent features are whether or not OA is required (i.e., "mandatory") and whether or not the requirement can be waived.

Some further parameters that will need to be added are:

compliance monitoring (how)?
consequences for non-compliance?
deposit as condition for evaluation?

The primary need for these parameters, however, is not so much "as a service to authors" (who need only know whether, when, where, what and how to deposit [and why!]) but as a service to policy-makers (institutions and funders), in order to demonstrate quantitatively the importance and effectiveness of the various policy parameters for optimizing their mandates.

David Crotty: “Your characterization of my suggestion is also inaccurate. By using the word "instead" you suggest a zero sum game where any actions or contributions from publishers would eliminate or prevent resources from other groups.”

No, as I keep saying, anyone is free and welcome to provide a mandate registry, and can harvest whatever they want of ROARMAP, which is OA and OAI-compliant.

You suggest that your statement that “I think the publishing community would be better served by creating (and supporting) a resource of this kind” is inaccurately described as “instead.” How about “rather” or “preferably”? (There seems to be a preoccupation in the Scholarly Kitchen with terminology rather than substance, wrapped in the rhetoric of “errors and misinformation.”)

You go on to say “I would worry about any researcher who would rely on this sort of database [ROARMAP], rather than going directly to one’s funding agency or institution and carefully reading through every single detail of any relavant policy. So if these resources are inadequate for those directly affected by the policies tracked, who are they for?”

All policies are linked in ROARMAP, so anyone who wishes can read every word of them… It seems the gripe is just with the word “mandate” — and perhaps with what was or wasn’t selected in the ROARMAP policy excerpts (often furnished by the registrants rather than ROARMAP).

My own view is that authors are far better served by consulting ROARMAP rather than anything provided by publishers, who are in a profound conflict of interest with green OA mandates, and heavily lobbying against their adoption (except when they provide extra funds to pay for gold OA).

The research community’s research is already locked into publishers’ grip with subscriptions; publishers would like to keep it that way with hybrid gold OA journals; they would also like to be the ones “complying” with those green OA mandates that they don’t manage to prevent from being adopted in the first place, via their lobbying and that they are delaying as long as possible with their embargoes, if they slip past the lobbying. Far better for publishers to see to it that the unblockable mandates are at least kept to their minimum conditions, while again locked into publishers’ grip rather than left to researchers.

It fits well into this “leave it to us” pattern that publishers would also like to host the definitive OA mandate registry that researchers consult. Deprecating ROARMAP as “errors and misinformation” is a good step in this direction…

So, is it indeed “instead,” “rather,” “preferably” or just “in addition”? Take your pick. The motivation and conflict of interest are exactly the same.

David Crotty: “This ‘us against them’ mentality continues to plague some advocates who have been unable to move past what Cameron Neylon calls the ‘angry protest movement’ stage of OA to the practical implementation stage. Having a listing of policies created by publishers would in no way harm ROARMAP, just as MELIBEA does not harm ROARMAP.”

There is no conflict of interest with MELIBEA, likewise a project of the research community: We collaborate happily (as illustrated by the 2010 link I provided). ROARMAP and MELIBEA (and BASE) have the very same objective: 100% OA, as soon as possible, which means via mandatory Green OA. In contrast, there is a profound conflict of interest between the research community and the publishing community, which persists in embargoing Green OA and lobbying against Green OA mandates. No “angry protest”: just rational, practical measures to resolve this conflict of interest in favour of research, researchers, their institutions, their funders, and the tax-paying public that funds the funders and for whose benefits — not publishers’ — the research is being funded and conducted.

And I’m not angry — I’m just tired, and impatient for 100% OA after so many years needlessly lost already,

David Crotty: “I remain confused by a movement that calls itself "open" while only selectively allowing those who pass a litmus test to participate.”

Who is stopping anyone from participating? I keep telling you that ROARMAP is open, and so is the web. Publishers can do what they like.

But if publishing interest advocates can criticize ROARMAP as “errors and misinformation” and argue that a publisher-provided mandate-registry is hence needed, surely I can respond with corrections, counterarguments and further information, can I not? That’s what’s meant by “open.”