“This is a great time to be starting an IR!” enthused one SPARC-DR speaker. “You get to learn from those who have gone before you.” And avoid their mistakes, understood. Believe it or not, this marks progress. Whether any given attendee had read Roach Motel was a crapshoot—many had, some clearly hadn’t—but its lessons were the floor under the entire event. I only heard “build it and they will come” in scare quotes, accompanied by polite derision.

So new repositories, those that manage to get themselves going in a budget environment where new services are a tough sell, will start on firmer ideological footing than the generation before them. That’s good. At the very least the new entrants will have a clearer sense of what they want and how they can get it than we did when we were in their shoes. Wonderful.

What are existing repositories supposed to do, then?

I asked this question at SPARC-DR in response to the speaker mentioned above. The answer I got was pretty mealy-mouthed: “Go find allies.” Um. Yes. What exactly do you think I have been trying to do for the last three years? (I lost my temper entirely with the closing keynote speaker over similar bell-the-cat rhetoric.) If I have no allies worth mentioning after three years of effort, what is the likelihood they will materialize out of the woodwork now? And how is it that a burdensome legacy of failure will help me find allies, exactly? Especially allies with political clout and money? Especially since turning a failed repository around takes more resources than have already been thrown at it?

These are the questions institutional repositories will have no choice but to confront in 2009. As library budgets shrink, money for programs of dubious efficacy disappears. Remember also that the United States does not have the nationwide research-reporting mandates that several European countries and Australia do; academic libraries have been running repositories purely out of idealism and hope. I can’t speak for you, but in my neck of the woods, these are not commodities available in generous supply.

Given that my crystal ball cracked up in a big way last year, I’m a little hesitant to make the dire forecasts my gut is telling me are warranted… but only a little. I wouldn’t be at all surprised if last year’s prediction about a repository closure was correct, just premature.

What is clear to me at this juncture is that the repository world is split in two as regards the appropriate response to faculty apathy about deposit. One chunk of the world is gamely gearing up to take on mediated deposit of the peer-reviewed literature. The other chunk of the world is using the repository for lower-hanging fruit (ETDs, undergraduate research, collapsing it with the local digital library, whatever) and doing a fan-dance around its lack of commitment to green open access. All right, there’s a third chunk, too: those that are doing both.

However, it’s the second chunk that are the problem, from the point of view of the open-access movement. Bluntly, these institutions love green open access—until it costs them resources beyond the mere provisioning of a repository. Since they are now uncomfortably aware that green open access costs more than that, they are sidling away from it in as delicate and face-maintaining a way as possible. The question for open access is how to keep its agenda alive in libraries that are no longer able or willing to dedicate specific resources to the green road.

There are options. One is to push harder on the gold road: ask libraries to dedicate funds to gold OA memberships and author fees. This is a tall order, because the subtext here is asking libraries to cancel even more journals to pay for a brighter future, but I do think gains are achievable, if perhaps mostly symbolic ones at this point.

Another is to bypass outreach to libraries and work harder on faculty, in hopes of additional Harvards and Stanfords. My own thought is that like many federal agencies with their eyes on NIH, institutions are watching Harvard and Stanford to see how their policies shake down in practice before they dive in. It’s probably too soon to push this angle too hard in the States, unless targets are chosen very carefully indeed. (Hint: The more centralized an institution’s power structure, the better a candidate it is for this approach. Sprawling campuses governed in a decentralized fashion with lots of little squabbling power centers are bad, bad choices; herding cats won’t work, as the limited impact of the “patchwork mandate” tends to demonstrate.)

A third is to ’fess up about the real costs of green OA and provide libraries and struggling repositories with a realistic roadmap to achieving it. We have these data now; the problem is that libraries are backing away from the cost of the inevitable conclusions. If the open-access movement wants green OA through institutional repositories, it will have to stiffen the collective library backbone from the top down. I don’t see this happening, but if green OA through IRs is to survive, it will have to. Yes, there were a fair few library directors at SPARC-DR, but I guarantee you the problematic ones weren’t.

We repository managers have a lot of work to do in the coming year, if we’re to make a dent in the distrust we are rightfully regarded with. We’ll just have to see how well that goes.

I used to subscribe to that view. After some things I heard at SPARC-DR, I have changed my thinking. Truthfully, your choices may be constrained in either direction; if you have money but no IT staff, you’re hiring a vendor, whereas if you have IT expertise (even at my borderline-competent level) but no budget, you have no choice but to free-ride on open source.

Let’s imagine that you have a real choice, though. You have in-house IT. You also have a budget. You have to make a choice. In that case, the question is where you want to use your people. Where is their time best spent?

My current thinking is that if you have that choice, the only defensible reason to use open-source right now is if you are seriously planning to write novel code on top of the platform you choose. If that’s what you believe in throwing your staff time at, use open-source; if you go with a vendor, obviously, there won’t be anything for those people to do.

I’m willing to give a little bit on this stance in the case of EPrints, which I believe stands up very well on its own against vendor offerings, and will continue to do so. I would also encourage repositories that use vendors to demand that those vendors support SWORD, because (so saith my crystal ball) SWORD support will open up a lot of opportunity to adopt useful SWORD-based tools under development, such as the BibApp.

Bluntly, I don’t think there’s any excuse to adopt DSpace or Fedora unless you’re planning some serious hacking—unless the value you add actually depends on that hacking. The overhead in staff time (which, let us recall, also costs money) is too high for too little return on investment otherwise.

This equation would change if there were livelier code exchanges in the open-source software communities. If an open-source solution plus solid, innovative, well-maintained, decently-supported plugins or modifications would do the job, that would be a strong competitor to a vendor solution. We don’t live in that world, not for DSpace and not for Fedora. As a minimal example, I have a metric ton of useful little DSpace/Manakin hacks—and no place to share them, or look over other people’s. You can in fact find my Manakin hacks, if you know just how to tweak Manakin URLs to look for them, and if you read a little bit of XSLT. But come on, who’s going to do that? And who (but a CavLec reader) will even know to try?

There is no plugin/mod infrastructure, technical or social, for open-source repository software the way there is for, say, WordPress or Drupal. Honestly, for code-based innovation, you have three choices: do it in-house, use EPrints, or lean on a vendor. The open-source repository communities (EPrints included, unfortunately) are pretty much useless for sharing working code.

One of the things SPARC-DR taught me is that I have been spending much too much time beating on DSpace, and my time has better uses. I’m not entirely happy with that insight, because I’m sort of fond of beating the living daylights out of DSpace, but I can’t ignore the facts.

The simple fact is that every minute I spend on DSpace, I’m not spending on content recruitment, advocacy, metadata, copyright clearance, service development and provision, and all the other aspects of repository management. Honestly, those things are where a conscientious repository manager should be spending her time.

So get her away from the software, and keep her away.

The question of design and branding raises its head here, to be sure. In my experience, the vendors in this space do as good or better a job of design and branding as those of us (myself included) who do it ourselves.

This advice is probably worth what you’re paying. Nonetheless, here it is: unless you’re adding significant value by hacking, don’t run open source; outsource.

The BibApp was a major reason I got so excited about coming to work in Madison. At the time I first heard about it, there just wasn’t anything like it, and while it shares DNA with projects such as Cornell’s VIVO and BePress’s Selected Works, it still does useful things that nothing else does.

The public face of BibApp is a set of researcher profiles (see live examples from Woods Hole Oceanographic Institution), anchored in but not limited to their publication record. Researchers or their proxies can add photographs, statements of interest, and so on. BibApp is browseable and searchable, and results can be limited by facet. Publication results have OpenURLs attached to them, so interested readers can be directed to their local link resolver.

The behind-the-scenes face of BibApp is a publication-list manager and repository-populator. BibApp vacuums up citation lists in popular bibliography formats such as RIS and RefWorks XML. (Not, alas, EndNote, because doing anything with their homegrown formats has been proven hazardous. Word to the wise: Zotero exports RIS. If I were to start a BibApp project, that’s the tool I’d use.) BibApp checks for duplicate citations, prompting to have potential duplicates resolved or removed.

It does its level best to assign authorship automatically to individuals it knows about, no matter how a name is represented in the citation. Over time, as BibApp learns more about name variants, coauthorship patterns, publication-venue patterns, and subject patterns, it becomes more accomplished at assigning authorship. BibApp performs similar authority control on publisher and journal names, and it will shortly be possible to federate this information in order to pre-populate new BibApp installations with the knowledge other adopters have built.

Once BibApp recognizes a journal or publisher in a just-imported citation, it checks with SHERPA/RoMEO for policies relating to green open access. If it recognizes that the publisher’s typeset PDF is archivable, it bundles up a package for import to a repository via the SWORD protocol, which currently works over DSpace, EPrints, and Fedora. No muss, no fuss, no bothering faculty for keystrokes!

A pilot with a fairly small engineering department at UW-Madison turned up about 1400 archivable articles. The major time investment in all this is putting together those citation lists, but in at least some cases, faculty themselves use citation-management software and can lend a hand. If not, student labor is quite adequate to compile lists. Eventually, BibApp will be able to update citation lists automatically via author-search RSS feeds from indexing and e-journal databases; that feature used to exist, had to be taken out, but will be put back!

And that’s the BibApp. The first stable release should be out in the first quarter of next year.

BibApp’s initial developer was my colleague and former SLIS classmate Eric Larson, of the Wendt Engineering Library. I also credit Wendt Library director Deborah Helman for giving Eric time and space to write code, and for tirelessly championing the software on the Madison campus. Eric shares development duties with Tim Donohue of Illinois, and Sarah Shreeves of Illinois has been instrumental in its outreach and sustainability strategies. Jointly and separately, we are pursuing several avenues for funding and additional developers, and I’m quite optimistic about its future.

That said, we are definitely looking for additional BibApp developers and early adopters, so please get in touch with us if you’re interested!

Not all repository-rats are as difficult or irascible as I am. We are, I will venture to say, a tough crowd to speak to, though. What I know about us, based on observation, is that we tend to be on the one hand ruthlessly practical, on the other remarkably attuned to our environment. I myself am more the former than the latter, but in all honesty, it’s the latter group of rats who tend to be more successful.

We are also a group of people that has had a lot of ideas that turned out to be wrongheaded shoved at us, and a lot of smack talked about and to us. Moreover, those of us who have been in this business awhile are a bruised, beaten crew, and we are cynical in direct proportion to our scars. That makes us a tough crowd.

Herewith, some suggestions for avoiding the biggest pratfalls in talking to us.

Know what we do and respect its difficulty. Breaking this dictum seems to be particularly common among career academics, and it may be part and parcel of academia’s general disdain for librarians. Nonetheless, the successful rat-whisperer knows something about the lifecycle, care and feeding, and behavior patterns of the common repository-rat.

We know when you lack clue. We do not appreciate it. We are, however, a fairly approachable species. If you don’t know what we do and why it’s hard, ask before you speak, please. Or, you know, read something like Palmer et al.

Understand our environment. There is a lot about open access in general and IRs in particular that’s just plain weird. I am constantly gobsmacked, caught utterly by surprise, by things that go on in this arena. That said, a lot of what goes on can be reasoned about from available evidence, and a number of us engage in this reasoning regularly. Peter Suber, of course, is an honorary repository-rat in this regard. I have to assume it’s the philosophy background that lets him reason so well and explain his reasoning so clearly.

When you run into an environmental phenomenon you don’t understand, such as (for example) the behavior of scholarly societies with regard to open access, you need to look for evidence (it’s there), look for analysis of that evidence (there has often been some), and do some thinking. If you don’t, you sound poorly-informed or just plain stupid.

To my mind, the key thing you must build a sense of regarding IRs and repository-rats is the power relations surrounding them. Innumerable are the speakers and writers who behave as though repository-rats were Hogwarts professors, able to wave wands and create content, awareness, administrative backing, funding, software, developer time, staff, buy-in from librarians and faculty, and copyright clearance from nothing. We’re not and we can’t, because we do not have that kind of power, not individually and not collectively either.

Do not speak to us as though we were faculty; many of us are not. Do not speak to us as though we were deans or library directors; we’re not. We are, by and large, caught in a terrible responsibility-without-authority bind. I cannot overemphasize how important it is to understand that when you speak to us. The instant you make me say “great, but who bells the cat?” you have lost.

What does work is to point out parts of our environment that we don’t entirely understand. Discipline-based insight, events outside academia that may help or hinder us, and anything happening in smoke-filled back rooms that we aren’t in are prime topics for us. Coherent visions of the river in which we are but small eddies may also serve (John Wilbanks did this beautifully yesterday), but…

Glittering generalities do not fly with us. Glittering generalities are what got us into the mess many of us are in. I am reminded of the scene in Kipling’s Stalky & Co in which some rich slob with no skin whatever in the military game slobbers on to a bunch of kids who are about to go out to India (and lose their lives as like as not) about patriotism and valour and service. “‘This man,’ said McTurk with conviction, ‘is the Gadarene swine.’”

You don’t want us to think that about you. I know you don’t. So be grounded, be realistic, be practical, be where we are.

And please, please, please don’t ever quote Lynch 2003. Trust me on this: the IR discourse has moved a long, long way on from “essential infrastructure.” We rats are so done patting ourselves on the back about how essential we are; we know differently. If any glittering generality has come back to stab us in the back, it’s that one. We loathe it. Avoid it.

We know what we’re supposed to do. We need grounded insight on how best to do it. We know we need content. We know we need increased awareness at our institutions. We know we need to speak to faculty in language they respond to. We know we need sustainable funding and staffing. We know we need allies and partners. Do not speak to us as though we do not already know these things. It is trite and patronizing. Do not speak to us as though we do not try, every day of our working lives, to make these things happen—particularly if you have never tried it yourself. It is condescending and rude.

If you have insight or experience on what we can do to get ourselves these things, though, we’re all ears. If you can actually help us get these things, we will bless you forever. I’m still waiting to hear Stuart Shieber talk, for instance. He’s as close to somebody who waved a magic wand as I can think of. I have no earthly clue how he did what he did, nor how his environment hindered or facilitated it, and I’d dearly like to know.

This is a more-in-sorrow-than-in-anger post, unfortunately about a real speech to a roomful of real repository-rats at a real conference, a speech that failed to meet its mark in a truly trainwrecky fashion. I post it in hopes of avoiding further such trainwrecks. In order to keep the focus on the larger phenomena that fed into that speech rather than the speech itself, I decline to name conference (sure, right now that’s obvious, but in a few months it won’t be) or speaker.

John Wilbanks

i.a.n.a.r.e. “I am not a repository expert.” “I almost said that I am not a repository rat.”

gets his information from Mellon and JISC reports, and links seen on blogs “especially Caveat Lector.”

“why is there a disconnect between planning to share and actual sharing?”

Disruptive processes can’t be planned in advance; planned innovation is slow.

Digital publishing is just “a bigger earhorn” because we’re still thinking that the way to communicate is through writing papers. We’ve made that better, faster, and cheaper, but the process is basically the same; boil it down to 8.5×11 pages.

Process change comes more slowly than product change.

So why is it hard to get this content? Why don’t faculty see the light?

- stable systems are resistant to change on multiple levels. No one thing will make people wake up; there are interlocking barriers to change. One such barrier is copyright, which locks up the container of the facts, not the facts… but it really locks up that container!

So we’ve moved to leasing materials, not owning them; and licensing makes it harder to unlock those facts from their containers. No indexing allowed; no adding hyperlinks.

Rights clearance is a pain! It is a block preventing process disruption. We haven’t provided enough incentive, enough “universal solvents,” to remove these blocks. Now we’re even seeing copyright applied to databases, often through confusion rather than malice (e.g. ChemSpider). What do the ideas behind CC mean, as they propagate into the scholarly realm? The rights problems are going to get MORE, not LESS complex; forcing IRs to focus exclusively on the peer-reviewed research ignores the library’s role as repository for lots of stuff. This complexifies rights issues.

- Faculty prefer carrots to sticks. So what do incentives need to do for them?

The minimum incentive needs to get faculty to go through the metadata/upload process, or to let somebody else do it.

Cartoon: “Behind one door is tenure; behind the other is flipping burgers.” If it doesn’t help them get tenure or another grant, it competes with activities that WILL help them toward tenure and grants. This is another way the system resists change! We wouldn’t NEED mandates if this weren’t so.

Let’s assume we fix the copyright and rights-clearance and incentives problems. This will create a flood of work!

Not easy to install IR software! Very powerful systems, but they need PEOPLE to run them. Can’t just go to register.com. Too frequently these people are not part of the conversation. Can be hard! (CavLec: miniature disasters post. An hour to change one link!) This, too, is another change-resister.

This is the complexity of the system we’re up against. Multiple levels of barriers, with multiple fail-safes.

reports from the front lines: building a commons is really, really hard! Takes passionate people with a clear point of view who are not willing to compromise on that view.

Everything at Science Commons is based on “running code:” legal and software code. >1000 journals now under CC license, which is pretty good! Scholar’s addendum engine, done in a single line of HTML code that can be dropped into a page. But they can’t keep data on who uses it (privacy), so it’s hard to follow up to assess.

Goal: using this to negotiate with publishers. But unless there’s a funder or institution behind this (NIH or Harvard), faculty won’t use it, because the power is on the side of the publisher (remember tenure!). “When institutions copy Harvard, we hope this will help.” (But people aren’t copying Harvard! -me)

Databases: copyright status isn’t clear, how/when to integrate isn’t clear, storage is a technical challenge. Ties lawyers in KNOTS! Default stances different in US and UK (no protection in US, some in UK). This makes Science Commons’s charge harder — but the only way to combine datasets is to eliminate the rights barriers. Solutions: CC0, and a set of “Science Commons norms” (citation, plagiarism control, etc). Dangerous to use law for that; better to use norms (yes! -me).

However, this conflicts with the protection instinct faculty have, and corporate funders even more. However, the protection instinct is frequently (? I would say “sometimes”) an instinct to protect freedom.

OA solves the legal problem, but the other problem is the “container problem” — the paper as a container for facts, the standalone database as a container for facts, are bad ways to go. Solution: Semantic Web. How do we make Google work better for science? Google finds things based on inbound links, but Google doesn’t search databases and doesn’t notice “links” to them. Goal: e pluribus unum (discovery tools that work usefully across different datastores). Get money to bribe database owners to do the right thing to make this possible.

Can put queries in URLs and then remix them by changing URLs. “Corpus of queries as links” and let people hack them. Not planning to share, but actually sharing, and throwing the result open, and it creates a commons! <$500K to make this possible! Have to do horrible screenscrapes and stuff just to create proof-of-concept to show people what’s possible when you open up.

NOTHING replaces hacking and releasing! Using trademark to protect the quality of their work, not copyright.

“Don’t plan to hack, hack! That’s the only way around the incentive problems.”

2 futures for repositories.

Note that repositories are points on a map — with no edges! No links between them! No networks! But networks create better incentives than points. Doesn’t mean “get complicated,” just the opposite: simple systems win the network game! (”OAI-PMH helps, but not enough,” with which I completely agree.) AOL and Prodigy were points, and they did cool stuff, but only “their people” could do anything to improve the system. The WWW was bloody ugly in comparison, but it was OPEN, and so huge numbers of people improved it. Three layers of openness (Benkler): physical layer, code layer, content layer. Fourth layer: knowledge layer, which means we have to deal with IP problems — so we have to engage the copyright problems.

If we do this right, we create gears instead of locks. This is the opportunity! “Open copyright, balanced incentives, and distributed workloads.”

We have to do this by solving an information problem faculty actually have. That’s the road in. (HALLELUJAH. -me) What questions can only a network of IRs solve? So that people who use IRs outcompete people who don’t.

“How does the IR keep me from flipping burgers at McD’s?”

Individual brain capacity is not scaling, but COLLECTIVE brain capacity is, so how do we make our stuff work on a collective level?

Conclusion: don’t wait. Lots of things need to happen before all this becomes real! If we wait until all the problems are solved, the commons won’t have what it needs to explode. But people aren’t watching IR space, which is the best time to create an open, disruptive system! Use existing ontologies. Work around problems rather than tackling them head-on.

Create new ways to measure things. Tenure vs. McD’s is a matter of citations; that’s the only thing we know how to measure! But what about data? Downloads? TrackBack?

We need a thousand flowers blooming, not the slow process of consensus.

Invest in your repository staff! Hard to do when facing real financial crises like the serials crisis, but “there’s nothing so expensive as cheap people.” (I am ready to cry. That’s exactly what libraries are NOT DOING.)

The latest (PDF), on repository metadata interoperability, is a classic of the genre. Smart, grounded, and despair-making. Despite its focus, there’s a lot more to this report than OAI-PMH and authority control; it asks trenchant questions about what IRs are for and whether they’re doing (or even can do) what they’re supposed to.

I keep reminding myself that the bad truths need to make themselves heard before change can happen. I reminded myself of that all the way through writing Roach Motel, and I remind myself again every time I read a JISC report.

If you’re less inclined to despair than I am, check the report out.

Imagine my surprise to find a post on sharing that nails the bureaucracy problems with institutional repositories right through the head.

We are the poster’s They. We should worry about this, in my humble opinion. Quite a lot we should worry about it.

I’ve heard many a repository-rat bellyache about how little faculty want to do to share their work in IRs. We need to stop bellyaching and start accepting that if we want pretty metadata, we’ll have to do that bit ourselves. Make it easy, make it fun, make it magic—and isn’t pretty metadata magical?—and watch our content-recruitment problems melt away.

Also worth reading is Andy Powell’s evisceration of repository success conditions and measurement. Repository rats, the question of metrics is to our address. Do you even have success conditions laid out for you? No, I don’t either. Does that make you comfortable? It doesn’t me. Idealism and look-the-other-way won’t keep IRs viable forever; eventually we’re going to have to prove our usefulness just like everybody else.

Me, I’m not looking forward to that day, not one bit. I’ve stated my case, don’t mistake me; a lot of my case amounts to “given the system it’s embedded in, the IR under its previous assumptions can’t be successful, so can we revisit those assumptions please?” But without so much as a definition of success, how far can I reasonably expect to get with that argument?

I’ve said it before. What do you want and how will you get it? Those are your success conditions, and it’s shameful that IR planning hasn’t been honest enough to answer those questions and stand behind its answers.

So this is my little cheer for Andy, for looking the hard questions in the eye without flinching.

They do have a procedure FAQ up as well as a rights and permissions FAQ. These paint a rather different picture from what I had thought initially, particularly with regard to permissions. It’s not a pretty picture, either. Some of the problems I see may have technical solutions, and I urge Harvard with all my heart to rectify them. Some don’t—and frankly, the whole picture as I now see it is one that imposes such a labor burden on faculty that I will be quite surprised if this policy lasts longer than two years.

I initially thought that the prior-commitment design of the Harvard policy, in which faculty assign Harvard a license before they so much as submit their work to a publisher, much less sign any copyright-transfer agreements, would free faculty from the hassle of worrying about rights. But what do I see?

Am I responsible for ensuring that deposit and posting of my material in the Repository complies with copyright law?

Yes, material may only be deposited in the Repository if its deposit and posting will not infringe the rights of any publisher or other party. You are responsible for determining this.

What rights must I hold in the material in order to make a deposit?

You must either own the copyright in the material, or have reserved or been granted sufficient rights under the copyright to allow Harvard to preserve and make your material publicly available without charge from the Repository. If you have a previous agreement with a publisher of your material, you should make sure that the agreement — by its own terms or by virtue of an author’s addendum you used — leaves you sufficient rights to allow deposit and posting of your material in the Repository. If you do not already have sufficient rights under the agreement, you can contact your publisher and request permission.

In hindsight, I’m not sure why I thought what I did; the NIH policy requires faculty to negotiate with publishers despite the existence of prior commitment. My naïveté aside, however, the effect of this requirement in practice will be pernicious. Harvard the institution isn’t negotiating with publishers; individual Harvard faculty are, and they are unlikely to feel they are negotiating from a position of power.

I can well understand that Harvard feels hamstrung here. Most universities are, because they don’t assert any copyright or IP license in their faculty’s publications. If they did, then it really would be a clash of the titans, which for our purposes would be preferable—but faculty wouldn’t stand for such a power grab (and what’s more, they probably shouldn’t).

As I suggested before with regard to the NIH policy, adding hassle to faculty lives is all too likely to cause backlash against open access. Negotiating copyright with recalcitrant or clueless publishers (and do not underestimate the cluelessness of publishers, particularly tiny humanities outfits with one to a dozen or so journals) is no end of hassle.

In my opinion as a repository veteran, this is disaster waiting to happen. I can see two relatively low-cost ways to mitigate it: keeping a public list of recalcitrant publishers (as I suggested with regard to the NIH), and devoting librarian time to help faculty deal with the negotiations. Neither of those strategies, unfortunately, will mitigate the anger and betrayal faculty will feel when they run into their first recalcitrant publisher.

Still, if Harvard is serious about this, there’s a better way altogether to do it: have faculty turn in their postprints and their agreements, and then let the librarians sort out the permissions. This is not cheap, because copyright is a morass and librarians aren’t gratis. It is, however, effective: faculty will perceive much less of a burden and so will comply far more readily, the institution will gather priceless data about publisher policies and faculty publication patterns, and negotiation procedures can be centralized and streamlined.

In an ideal world—pinch me, I’m dreaming—the Harvard library would leverage something like the BibApp to gather a lot of this publication information automagically, and deal with the small minority of publications that are final-PDF-archivable automagically, without so much as troubling faculty. Then they’d close the feedback loop by telling faculty which publishers are cooperative!

Ah, well. The world is far from ideal.

Now let’s turn to the procedures.

I’ve visited DASH, but I don’t see a way to submit my articles. How do I do that?

Before you can submit articles, you have to register for an account and receive permission to submit to a particular collection.

This is just critically, desperately wrong. Years of study of e-business usability have demonstrated that huge numbers of potential customers fall away at the first perceived barrier to completing a purchase. Signup requirements are therefore broken. What’s the fix? Prepopulate the DSpace database, Harvard. I know you can get an LDAP dump of faculty and their departmental affiliations. Do it. Then figure out how to keep it updated. Whatever you do, don’t make them sign up!

If you can’t do that, at least recognize harvard.edu addresses and add them automatically to the submitter group for a catchall collection. We’ve done that at MPOW; you can too. Attempted spam and undergraduate pranks will be caught in the review step that I see you are wisely implementing.

What is involved in the submission process?

Submitting articles to DASH involves a simple process. You select a collection to submit the article to from those that you have submission permission for. Then you follow a five-step process: (i) provide an appropriate license to DASH to distribute the article, (ii) describe the article (title, author, etc.), (iii) upload a PDF of the article itself, (iv) review all of the information that you provided, and (v) complete the submission.

Don’t lie to faculty; that’s just bad juju. A five-step process is not simple! You’re also, I suspect, condensing the description process; unless you’ve been messing around with DSpace (and good for you if you have), faculty members are facing seven or eight screens, not five. Don’t think they won’t notice and hate you for it.

Make the submission process simpler. I plan to, once we have the last bugs out of Manakin and I can turn my attention to the Configurable Submission widgets. Roundfile the review step; nobody pays any attention to it anyway. (That’s another thing, incidentally, that we know from years of usability testing in e-business and other contexts.) Cut the metadata down to the bare minimum, and count yourself lucky if you get that much. You’re instituting review steps; trust me, let the librarians sort it out.

When all is said and done, I suspect Harvard will have to prepare to accept scofflaws: faculty bound by the Harvard permissions mandate who simply ignore it. This distresses me terribly, because it turns a promising tool into yet another spineless, toothless public-relations-only victory. Remember: the only real win is literature hitting the Web; all else is froth.

Please, Harvard, the rest of us really need you to get this right. As is, I’m not even sure this policy will survive.

Some of these institutions are using IR software as a quick-and-dirty digital preservation mechanism, mostly for institutional records. These institutions don’t have any particular commitment to open access, nor are they under any illusions about their faculty’s commitment thereto. They’ll let faculty deposit if they’ve a mind, but that’s not what the IR is there for. A variant on this scenario is IR software considered alongside other content- or knowledge-management systems.

So far, so good; this seems decently sensible to me, though in all honesty I can’t imagine most IR software standing up well against a real CMS. I’m relieved, though; I appreciate knowing that my profession mostly has its head on straight.

Unfortunately, that’s not the whole story. At least some of these small institutions have The Shiny in their sights. Everybody has an IR these days, so why don’t we? If we build it, they will come!

In at least one such place, I am told, cooler heads higher on the totem pole are prevailing, which is all to the good. In at least one such place, the hotheads are the top of the totem pole, sadly.

To be fair, small institutions have one major advantage large ones don’t: scale. Yes, scale. Small is beautiful; at many small schools, it’s perfectly feasible for the library to have one-on-one relationships with nearly every researcher on campus, especially given that teaching-focused schools aren’t necessarily publishing all that much research! One-on-one outreach works; it’s nearly the only technique that does.

That said, I see a Sisyphean climb even for small schools. Most importantly, small schools don’t have many librarians, and those librarians don’t have much time to spare. Forget about digitization or mediated-deposit programs; they’re wasteful luxuries at a small school. Many such libraries, even many small institutions, don’t have nearly the technical savvy it takes to run even EPrints, which is the most straightforward of the Big Three open-source IR packages. (Think I’m kidding? Remember Rohan, where I interviewed during my first job hunt? Microsoft shop par excellence. If it didn’t run on Windows, they didn’t run it. Any renegade Linux boxen had to be kept deep, dark secrets.)

There tend to be cultural differences between small teaching-focused and large research-focused institutions as well, ones that do not bode well for open access. One is obvious: these places tend not to have physicists, medical researchers, computer scientists—precisely the disciplines that are early adopters of open access. What they have in plenty are humanists and social scientists, who see OA (when they see it at all, which most frankly don’t) as a commie plot aimed directly at the heart of their beloved scholarly societies.

Another is less obvious, at least to me; I didn’t know about this until I experienced it. Faculty at small schools are unbelievably suspicious of their institutions’ administrations. Anything that so much as hints at institutional oversight of research gets the major fisheye. This makes IRs and institutional bibliographies a viciously hard sell.

A third: faculty everywhere are unreasonably paranoid about plagiarism, misquotation, and idea-appropriation, and unreasonably optimistic about copyright as a guard against those ills—but in my experience, faculty at small schools are worse, much worse. Open access isn’t just a commie plot, it’s a dagger in the heart of my precious ideas!

(I’m not sympathetic to that nonsense, not at all—but that doesn’t automatically give me the tools to fight it, either. Of course I have pointed out that the datestamp in an IR is pretty good proof of priority, but galloping paranoia doesn’t hear reason.)

I can imagine a small, tight-knit institution that really truly trusts its librarians making a go of an IR, especially if the technical pieces were handled outside the institution, by a vendor or consortium. It’d be a nice thing to see. I’m not especially sanguine, however; the barriers are significant, and by no means solely technical.

Honest to $DEITY, there are days I think that librarians’ continued involvement with the open-access movement is a sign of Stockholm Syndrome.

Open-access movement? Howsabout we shut these yahoos up, please? And howsabout we stop feeding their unjustified sense of entitlement and erroneous scapegoating by recognizing librarians’ contributions to the movement? If I see another book on open access where neither “libraries” nor “librarians” is an index term, I swear I’m done. I’m just done.

But before I go, I’m going to spearhead a publicity stunt. (You didn’t think I’d go quietly, did you?) Every single library-run digital repository, institutional or disciplinary, and library-sponsored open-access journal or similar project will go dark for a day. Then we’ll see just how much of the open-access movement is left without libraries.