My previous post on locus of deposit for scholarly self-archiving provoked a few reactions, as I thought it might. Stevan Harnad’s is the most thorough and notable. I think we each missed a few points.
Let me make a few statements, just to get them out of the way:
- Every research funder should mandate OA to the research it funds.
- Every institution that produces research (college, universities, research institutions) should mandate OA to the research it produces.
- Every researcher should ensure her research is OA.
These obligations are independent and co-equal. This particular discussion is at the level of policy — the first and second points — since practice has shown that policy is significantly more effective than unenforceable exhortations to authors in general. So our question is not whether to mandate OA, but how.
Where he’s wrong: Dr. Harnad repeats a claim that I dismiss verbatim in my original post: “Not All Research Is Funded, But All Research Is Institutional.” In the body of his post, he weakens this statement to “virtually all”. But my point — that not all academic journal literature is written by authors at colleges, universities, or research institutions — remains. I don’t know what percentage of authors are independent scholars or employees of non-profits, government, or corporations, but I’d hazard a guess that it’s at least 5% across all fields, and probably as high as 10% or 15% in some fields. (Harnad doesn’t offer a number either, but claims it’s “such a minute fraction” it’s not worth worrying about.) Institutional mandates won’t touch these authors (by definition). Funder mandates may touch some, though I’d suspect that most such research is unfunded. Regardless of where the author works, we need OA to academic journal literature. Where non-institutional authors are funded, we need funder mandates to ensure OA to their research. Since these authors don’t have an institution, they don’t have an institutional repository, so they need the flexibility to deposit in a funder or thematic/disciplinary repository instead. This is, to me, the minimum sane policy for funder mandates — not just for non-institutional authors, but for authors whose institution doesn’t offer an IR. (Harnad notes that some funders do offer “back-up” repositories for grantees without IRs, but without endorsing this flexibility.)
Harnad also deftly re-defines the traditional meaning of “institution” in “IR”, to mean “any organization whose employees might ever produce research” rather than the commonly-understood sense of “an institution of higher education or research”, i.e. colleges, universities, and research labs. As a result, he reduces further the number of “unaffiliated” researchers. While some non-HEI/research organizations that produce research may establish repositories, as Médecins Sans Frontières did, many won’t, especially as the research output declines. (If one employee per 1,000 publishes an article per year, does it make sense to host and manage your own IR? Doesn’t it make more sense to tell your employees to deposit in a higher-visibility thematic repository?)
My point about authors with multiple institutional affiliations also goes unheeded. Harnad suggests that this situation “only” requires one deposit, “followed by export to the IRs of the rest of his institutions”. Well, now we’re talking about export, which is exactly what I call for between IRs, thematic, and funder repositories. There’s no difference between exporting from one IR to another and between exporting from a thematic repository to an IR.
Harnad also understates the resource requirements for adequately maintaining an IR. I’ll let Dorothea Salo answer this point. Since I spent much of today waiting for my email server to resume functionality, I know that servers don’t just maintain themselves, even if you didn’t dedicate any resources to customer service or development.
Where I was wrong: Since more research comes out of institutions (as a whole) than is funded (as a whole), institutional mandates offer broader coverage. (But a given funder may sponsor significantly more research than a given institution produces, and there are fewer funders than institutions, so they give more “bang for the buck” as initial targets.) To the extent that funders are ahead of institutions (and they generally are), it is useful to use any leverage that exists to encourage institutions to deploy IRs and mandate deposit in them. Why must institutions deploy IRs? For the same reason that funders must offer their own repositories: you can’t require authors to deposit somewhere that doesn’t exist. Just as a funder shouldn’t mandate deposit without offering a “repository of last resort”, neither should an institution. Having established that both institutions and funders should offer repositories (since they should both require self-archiving, and you can’t require deposit without offering a “repository of last resort”), we return to the earlier question: do funders have any leverage to encourage institutions to deploy IRs and mandate deposit? I am less sanguine about this than Harnad, but flooding IRs with deposits certainly could have the effect of building institutional constituencies for IRs and thus for institutional mandates. (However, this assumes the IRs are well-managed — otherwise, you’re breeding dissatisfaction with them — and while funders can have measure of control over the management of their own repositories, they can’t do so with IRs. You have to hope that institutions will decide the requisite resources to maintenance. You entrust your infrastructure to a third party. Enforcement now includes not just deposit by the author, which is easy to measure, but adequate maintenance by the institution, which is less easy.)
Therefore: Is there some promise in using funder mandates to prod universities toward institutional mandates? Yes, there’s some. Is it a sure thing? No. Do funders still need to provide a “repository of last resort” to deal with exceptions? Yes. Is it still easier for funders to monitor compliance and ensure adequate levels of service (including preservation) by requiring grantees to get their work into the funder’s repository, one way or another? Yes.
So I’m willing to shift a bit from my prior position of the symmetry of funders and institutions vis-à-vis repositories. But I still don’t see the IR warranting the level of primacy that Harnad would assign to it.
As for policy recommendations for funders, then, I’m agnostic between these two options — but admit either is preferable to my earlier suggestion:
- Require grantees to deposit in their IR, if they have one. Require also that grantees ensure their publication ends up in your repository, wherever they initially deposited it. If they are accustomed to depositing in an institutional repository, they’re responsible for working with your repository manager to ensure their work gets harvested. Better still if the repository allows users to set up automatic harvesting themselves, either for a single publication or for all the author’s future publications, if they don’t want to talk to the repository manager.
- Encourage grantees to deposit in their IR, if they have one. Require also that grantees ensure their publication ends up in your repository, wherever they initially deposited it. If they are accustomed to depositing in an institutional repository, they’re responsible for working with your repository manager to ensure their work gets harvested. Better still if the repository allows users to set up automatic harvesting themselves, either for a single publication or for all the author’s future publications, if they don’t want to talk to the repository manager.
There are costs and benefits to either, in my view. Requiring offers the potential benefit of building constituencies for IRs and institutional mandates. But it also comes with the cost of introducing another step to the process, and runs the risk of backlash if IRs are poorly managed, a factor over which the funder can have little control.
(Aside: Funder repositories come with their own cost: the increased potential for government interference in science that comes with greater centralization, since the largest funders are public agencies. As I’ve written before, this is a minor but non-trivial consideration: minor because published research, by definition, is available elsewhere, and because funding decisions are more of an “interference” than tampering with a non-exclusive database of results could be; but non-trivial because dissemination ought not to be restricted or tampered with, particularly as repositories become more important as a dissemination source. Luckily, we can erect firewalls for academic freedom, as I’ve proposed. In addition, we have friends like the Scientific Integrity program at the Union of Concerned Scientists as watchdogs for such policies when they are adopted.)
Update: Things I forgot to say:
- LOCKSS: lots of copies keeps stuff safe. As long as we can have export/cross-deposit down to a few clicks (or dropping an email to the repository manager with a link to the item to harvest), there’s little harm in having multiple identical copies, and some benefit in terms of preservation.
- Heather Morrison’s comments succinctly point to some of the benefits of both thematic/disciplinary and institutional repositories. (Some qualities of each apply to funder repositories.) This is one of the (previously) unstated assumptions in my argument: if authors are depositing somewhere, it’s because they see a benefit to doing so, so let them keep on doing it, and make it simple for them to ensure it ends up in your repository, too.
- Peter Suber’s comments in his February newsletter (#3 and #13) summarize well the gist of what I concluded in this post and the previous. Not surprisingly, that’s where this whole thread started: Bernard Rentier’s post, to which my previous post was responding, was in turn a response to Suber’s comments. But even if we ended up in the same place, it’s (hopefully) with a more thorough understanding than we started with.
Well said, Gavin. Agreed on all points.
Better interoperability and LOCKSS-like sharing agreements among repositories might defuse a lot of this tension. OAI-PMH is flatly insufficient; OAI-ORE might turn out to be helpful. I’m still anticipating the day when a given article needs to appear in three or more repositories because of overlapping mandates — I think IRs need to get ahead of that problem ASAP.
Worth looking at is Andy Powell’s post, particularly on the topic of multiple copies affecting Google PageRank. There’s a research project lurking there, with implications for repository-interop design.
Hyperlinked version: http://openaccess.eprints.org/index.php?/archives/522-guid.html
Gavin Baker comments:
GB:
“Every research funder should mandate OA to the research it funds.
Every institution that produces research (college, universities, research institutions) should mandate OA to the research it produces.
Every researcher should ensure her research is OA.”
Noble wishes. But alas not being fulfilled — and it’s been a long wait. That’s why specifying institutional locus-of-deposit for existing funder-mandates matters: to help make these wishes come true instead of just continuing to hope they will come true of their own accord.
GB:
“[P]ractice has shown that policy is significantly more effective than unenforceable exhortations to authors in general. So our question is not whether to mandate OA, but how”
Indeed. But exhortations to mandate are not effective either! So this is an exhortation to make a specific modification in existing (funder) mandates to make them more effective in generating further (institutional) mandates.
GB:
“Where he’s wrong: Dr. Harnad repeats…“Not All Research Is Funded, But All Research Is Institutional”… he weakens this statement to ‘virtually all’…”
Not all researchers have fingers — only “virtually all” do: That does not mean it cannot be mandated that the deposit keystrokes need to be done.
GB:”[N]ot all academic journal literature is written by authors at colleges, universities, or research institutions”
Fine. This changes nothing regarding the potential benefits of funder mandates stipulating IRs as the default locus of deposit. As noted, where the fundee has no IR, they can and should deposit in a back-up CR such as DEPOT.
GB:”Institutional mandates won’t touch these [unaffiliated] authors (by definition). Funder mandates may touch some, though I’d suspect that most such research is unfunded”
That makes this case even more irrelevant. This is about funder-mandates inducing fundees’ institutions to adopt institution-mandates. No IR? Or no Instiutution? Deposit in the DEPOT — and let’s not let this minuscule number of exceptions obscure the potential benefits of funder mandates stipulating institutional deposit as a rule!
GB:”(Harnad notes that some funders do offer “back-up” repositories for grantees without IRs, but without endorsing this flexibility.)”
I did indeed endorse it, many, many times: And, in case of doubt, I endorse it here too.
GB:”Harnad also deftly re-defines the traditional meaning of “institution” in “IR”, to mean “any organization whose employees might ever produce research”… As a result, he reduces further the number of “unaffiliated” researchers”
Yes. And your point is…?
(“Every institution is just (a) a piece of free software http://www.eprints.org/software/, (b) a $1000 linux server plus (c) a few days of sysad set-up time away from having its own IR. In the meanwhile, there are interim alternatives like the DEPOT…”)
GB:
“Doesn’t it make more sense to tell your employees to deposit in a higher-visibility thematic repository?”
Most definitely not. It makes immeasurably more sense to mandate that both fundees and institutional employees deposit in their own IR (or DEPOT) and let high-visibility CRs harvest the deposits thereafter, because that will generate more institutional mandates. That is in fact the very essence of the point I am making.
GB:
“My point about authors with multiple institutional affiliations also goes unheeded. Harnad suggests that this situation “only” requires one deposit, “followed by export to the IRs of the rest of his institutions”. Well, now we’re talking about export, which is exactly what I call for between IRs, thematic, and funder repositories. There’s no difference between exporting from one IR to another and between exporting from a thematic repository to an IR.”
That’s the symmetry-fallacy again, which is exactly what I was refuting: There’s no point “calling for” (“exhorting”) the export of nonexistent (because undeposited) content. The point here is about funder-mandates designating institutional deposit in order to generate institutional-mandates (waking the “slumbering giant”, “more bang for the research buck,” or whatever you want to call it) so as to generate all that missing content!.
GB:
“Harnad also understates the resource requirements for adequately maintaining an IR. I’ll let Dorothea Salo answer this point.”
Umm — I’d rather this point was answered by the managers of the few successful, well-stocked IRs that are actually doing what OA IRs are for — rather than doing anything but (and at high cost)…
GB:
“Where I was wrong: Since more research comes out of institutions (as a whole) than is funded (as a whole), institutional mandates offer broader coverage.”
No, institutional mandates would offer broader coverage (indeed, universal coverage, if universally adopted). But only 30 have been adopted so far; and so what we are talking about here is how funder-mandates stipulating institutional deposit could induce more institutions to mandate deposit for the rest of their total output, funded and unfunded.
GB:
“(But a given funder may sponsor significantly more research than a given institution produces, and there are fewer funders than institutions, so they give more “bang for the buck” as initial targets.)”
Again, the point is not that a funder generates more research than a single institution (though most do), but that a funder generates research at many institutions. That’s what will give them “more bang for the buck” if they mandate IR rather than CR deposit. (And we are talking about making this small but important change in existing funder-mandates, not about just wishing for more funder-mandates.)
GB:
“[D]o funders have any leverage to encourage institutions to deploy IRs and mandate deposit? I am less sanguine about this than Harnad, but flooding IRs with deposits certainly could have the effect of building institutional constituencies for IRs and thus for institutional mandates.”
Sounds like you are and are not sanguine, Gavin!
And funders do have enormous “leverage” with their fundee institutions. But we are not talking about direct leverage to require fundee institututions to create IRs or adopt deposit mandates — just the indirect leverage that would result from specifying IRs instead of CRs as the funder-mandate’s default locus of deposit.
GB:
“Therefore: Is there some promise in using funder mandates to prod universities toward institutional mandates? Yes, there’s some. Is it a sure thing? No. Do funders still need to provide a “repository of last resort” to deal with exceptions? Yes.”
Agreed! So what are we arguing about?
(Yes, “repositories of last resort” do need to provided — though it’s not clear why they have to be provided by the funder; nor how many of them are really needed. There’s plenty of room, for example, in DEPOT, which currently has only 66 deposits after almost two years of existence, simply because funders are mandating direct deposit in their own CRs instead of the fundee’s IR — for which DEPOT was specifically created to serve as the back-up.)
GB:
“Is it still easier for funders to monitor compliance and ensure adequate levels of service (including preservation) by requiring grantees to get their work into the funder’s repository, one way or another? Yes.”
Here’s the way: import/export/harvest them from the IRs (and DEPOTs). And the fundee institutions will be more than happy to collaborate in monitoring and ensuring compliance with the funder’s grant fulfillment conditions.
GB:
“Require grantees to deposit in their IR, if they have one. Require also that grantees ensure their publication ends up in your repository, wherever they initially deposited it. If they are accustomed to depositing in an institutional repository, they’re responsible for working with your repository manager to ensure their work gets harvested. Better still if the repository allows users to set up automatic harvesting themselves, either for a single publication or for all the author’s future publications, if they don’t want to talk to the repository manager.”
Yup, that’s it! So what are we arguing about, Gavin?
GB:
“[Or] Encourage grantees to deposit in their IR, if they have one.”
Now that, in contrast, would be useless, just as encouraging (“exhorting,” “calling for,” “wishing”) authors to deposit — or funders or institutions to mandate — is useless. What we are talking about is a tiny but concrete and specific change in the implementational details of actual, adopted funder mandates, so as to require institutional (or, as last resort, DEPOT) deposit, in order to generate more institutional mandates. Merely “encouraging” it will generate nothing.
GB:
“[T]he risk of backlash if IRs are poorly managed, a factor over which the funder can have little control…”
Why are we speculating about the possibility of full, ill-managed IRs when what we are actually faced with is virtually all IRs (whether well-managed or ill) empty! Necessity is the mother of invention: Fill those IRs with their valuable intended content, and institutions will nurture them as the invaluable assets they will prove to be. Keep fussing instead about the current or future management of idle or nonexistent IRs and you will have years more of idle or nonexistent IRs.
GB:
“Funder repositories come with their own cost: the increased potential for government interference in science that comes with greater centralization”
Let’s not get carried away! We’re talking about published journal articles, made openly accessible, free for all. If they are deposited in each researcher’s distributed IRs, they can thence be harvested into multiple CRs. If there is government “interference” (whatever that means) in governmental CRs, there will be plenty of other harvested CRs (e.g. google scholar!) where those came from.
Stevan Harnad
American Scientist Open Access Forum