About a third of the Newberry conference was a “look! see how cool we are?” laundry list of finished or ongoing projects. Not my thing—I’d rather do cool things myself than look at other people’s cool things, except when their cool things inform my cool things somehow. So since I’m not actually doing anything especially cool at the moment, those presentations left me cold.
Except for the late-Monday one about this digitization-training program, which I am considering investing in myself. Texts I know how to digitize. Images, not so much. This looks like a solid, affordable bootstrap course. Wisconsin librarians: there’s apparently a session just for y’all in October. From what I’ve seen, it’s worth signing up.
Me, I won’t have time for this until 2005. That’s how it goes, I guess.
The rest of the conference was good chewy stuff about lessons learned, challenges faced, and problems still to be overcome. I was remarkably unsurprised to hear from Tim Cole of UIUC that OAI has run into problems with EAD data, all the way to invalid and incorrectly character-encoded XML. EAD is a bit of a rats’ nest to begin with, and frankly, it’ll be a long time before Webbish metadata (pick your schema; I don’t care) is as solid and reliable as a good MARC record.
The problem is of course larger than EAD or OAI. Standards organizations do a stunningly awful job of communicating and enforcing best practices. The EAD samples on the EAD’s own site aren’t XML-valid. The W3C, to mention a larger example, didn’t get anywhere evangelizing good HTML and CSS until the Web Standards Project stepped in. I don’t know what the answer to this problem is; I just know it’s a nasty problem.
I asked Cole what kind of feedback loop there might be from an OAI harvester/archive to an institution whose metadata isn’t up to snuff. Answer: shrug, “we’ve built some automated input filters.” It was pointed out that much metadata is generated as part of a project grant and then never touched again—once the grant money is gone, there’s no incentive and perhaps no ability to fix stuff that’s wrong or outdated.
But the lack of a feedback loop means lots of junk in OAI archives, and junk pouring in at an increasing pace because nobody knows any better. Sigh.
(The grant-money-causes-problems motif recurred throughout the conference. More on that later; not something I’d thought about before, but it’s definitely important, so I want to communicate it. For now, hold your horses.)
The John Unsworth keynote on the value of digitization was well-written and well-presented, but didn’t offer much novelty to this proto-librarian. Sighted people do click over, though, just to mouseover the introductory image and see what FTP errors can do to the best-intended digitization efforts. Too cute.
John Price-Wilkin of the University of Michigan and Kris Brancolini of Indiana University gave smart, balanced talks on in-house work versus outsourcing that had me nodding recognition and approval. The short summary: some things you keep in-house if you have a lick of sense, some things you outsource.
I am terribly glad that Price-Wilkin pointed out that relying on “unique” staff causes extra costs and perhaps meltdowns. If somebody on your staff can do something important that nobody else can do, you are just begging for trouble.
Both my previous employers got themselves in dutch on this, and the sad thing was how avoidable the whole problem was—the employees with the specialized knowledge wanted to teach it to other employees who wanted to learn. Relying on unique knowledge is a Stupid Management Trick that ought to be stamped out.
I also agree with Price-Wilkin that outsourcing design and specification construction is a bad, bad idea. You must understand the final product you want from your vendor, and you must give your vendor the most specific and detailed specifications you can come up with, to guard against said vendor handing you crap. Seen it happen, again and again and again.
In fact, Brancolini said point-blank not to outsource any kind of job you don’t have some idea how to do yourself, and I agree with her. Some vendors will take you for a wild ride. Some will inflate costs. Some will simply disrespect you (I have been guilty of this myself, though not with any of TAG’s clients—thus far, I have simply refused to take on as clients people I can’t respect). Some will return you garbage with the best will in the world. (I have a story on this point which I will share later.)
If you must do this—and Brancolini didn’t say this; it’s purely mine—hire a second vendor to check what the first one is doing. I’m serious. Do it. Mulberry Technologies calls this “independent validation and verification,” and it may seem expensive and unnecessary until you think about the costs incurred by cleaning up shoddy work later. (I’d hire Mulberry in a hot minute to do IV&V, by the way.)
Otherwise (back to Price-Wilkin and Brancolini now), do pilot projects, do part of the job yourself and outsource the rest of it, experiment, do something—yes, it’ll be expensive, but the risk reduction is worth it.
The obvious corollary to all this vendor distrust is that proofing and review must be done in-house too. (Putting on my occasional-vendor hat, I tell my clients to look hard and with jaundiced eyes at what I did. I’m far happier to correct a mistake fast and early than end up with an unhappy client later. Thus far, I am happy to say, no unhappy clients.) Price-Wilkin said bluntly that his outfit rejects vendor stuff until they get it right. Bravo.
Price-Wilkin and Brancolini also mentioned how important it is for digitization-knowledgeable people to get in on the earliest stages of project planning. The bugaboos here will sound awfully familiar to CavLec readers: wildly optimistic time and money budgets, utterly impossible-to-implement ideas, vague and vendor-exploitable RFPs, and so forth.
All that said, outsourcing has its place. Vendors can exploit economies of scale that small domestic outfits can’t. They may be able to afford better equipment. Their staff does this for a living (as opposed to, say, an academic library that hires a lot of transient student staff). They may run more checks on the results than you can (though, Brancolini was careful to say, you can’t rely on this). And the reality is that a lot of digitization work simply cannot be done domestically on the funds allotted for it. Whine about the job loss all you like (and apparently in Indiana they are whining quite loudly), but it can’t be helped unless libraries, archives, and museums are better-funded.
Hm, I think this will do for one post. The session I just outlined was, I think, the most valuable of the conference. If the PowerPoints land online, I will certainly link to them.
