Babies, bathwater, serendipity, bad vibes and mass confusion…

Writing a weblog posting right now is the last thing I should be doing, but I’ve wanted to jump in on this topic for a while now… My time is extremely limited, so I’m just going to blast this out — if I’m more incoherent than normal I’ll beg the reader’s indulgence, and hopefully I can revise this over the next few days.

A couple weeks ago, Scott Leslie wrote a characteristically thoughtful post on the joy of browsing, in which he described the concerns of some library patrons at the prospect of an automated stacks retrieval system here at UBC. (Incidentally, this is a concern I share — I think preventing access to the stacks is a terrible idea. When I was grad student, some of the best works I encountered were discovered in the course of free-form aisle wandering. While writing my thesis I actually took a lower-paying job at the McGill Library shelving books because I enjoyed the serendipity effect so much — that, and I had the chance to suck up to people with the power to reduce my perpetually massive overdue fines.)

Scott uses this anecdote to make a few observations on the nature of metadata approaches:

… for some time I have wanted to respond to the seemingly unmitigated glee of some of my fellow ed tech bloggers over the death knell being sounded for metadata, in particular human generated multi-level taxonomies (see this, this, this, and many others to boot).

To be fair, many of them are not taking on taxonomies per se, just the onerous means that many systems (especially learning object repositories) have placed on users to create metadata. Really, I have no argument with this.

But I do have an issue with the effects that so-called ‘flattened keyword browsing’ will have on the above noted serendipitous ‘pleasure of the unexpected.’ When we throw out multi-level taxonomies, we also through out the side-effect of teaching people about the shape of the collection (and thus the shape of the discipline) as they look for things.

If you follow the links, they point to me and to people I frequently collaborate with… I have tremendous respect for Scott — I’ve been privileged to work with him on a few projects, and I am keenly aware of how knowledgeable he is not only about repositories and metadata, but emergent online community models as well. When he levels a criticism like that, it merits a response.

First off, I want to clarify the extent of my ‘unmitigated glee’ at the prospect of metadata’s demise. While anyone who reads this weblog can tell that I am blissed-out by how well tools like Furl, del.icio.us and Flickr use stripped down, user-friendly metadata entry to create personal and networked collections (as opposed to managed collections) of content, I hope I’m not giving the impression that I am condemning more structured metadata approaches. I’m simply not qualified to make such an assessment.

I have thought for a long time that central, structured, organizational approaches have their place, and that they must be enhanced by the creative application of simple, distributed user-level tools to fill in the gaps. If you have spent thousands of dollars creating a carefully-designed learning resource, in my view you’d be insane not to store it in a standards compliant system, and have it catalogued according to a fairly rigorous metadata standard (by a cataloguing professional). And though this approach is tedious and difficult to implement, it can bear very sweet fruit. Over the past few months I have been giving regular workshops for faculty that are nothing more than pointers to the remarkably rich collections of resources that have developed out of this approach. I really hope I don’t seem as if I am diminishing the achievements of projects such as the NSDL, the RDN, or MERLOT for that matter. When I present these collections to instructors at UBC, the response has been wholly and unreservedly positive. These and many other collections represent a tremendous achievement.

(And I should note that the other two anarchistic malcontents that Scott cites are themselves the developers of repository systems, CAREO/APOLLO, and the Maricopa Learning eXchange. I myself have struggled to support an instance of CAREO here at UBC. So if we seem a little too enthusiastic about alternative models, I hope it’s obvious that we come by our frustration with the existing approach honestly.)

What has been apparent to me for a while now, however, is that these larger scale approaches do not presently meet the needs of smaller, less defined digital works. It makes perfect sense to carefully catalogue and store an expensive piece of educational multimedia, but what about smaller, more ephemeral pieces of digital media such as lesson plans, link collections, reflections on practice, or quick and dirty image objects? If it costs more to metatag a resource than it does to create it, obviously the structured model is going to break down. I personally don’t claim that the “small pieces” approach should replace all the efforts to collect learning content — and I don’t think Alan and D’arcy do either — but I do think it provides a necessary alternate channel for content and communication that presently is slipping through the cracks. It also provides a lower-threshold loosely-structured infrastructure for smaller, low-budget implementations. It provides a framework for individuals to get into the game, and to tailor their online outputs to their own needs.

If I were to offer an analogy, I would suggest that the stripped-down, distributed approach to resource collection and distribution might co-exist with more formal approaches in much the same way that weblogs already supplement traditional news publications. The relationships between big and small players are still being negotiated (which is why us little guys sometime sound so strident — we’re essentially in a bargaining process: “you give us RSS feeds, and we’ll give your resources way more usage”…), but ultimately I think both sides need the other to flourish.

Another reason some of us sound exuberant about the emergent model is that this stuff is fun. Having been in the trenches flogging a top-heavy and incomprehensible learning object model to skeptical users for nearly five years now, it’s a relief to see ways of performing similar tasks that actually work in the wider web world. It’s hard to tell if the major players are taking note of what’s happening — my experience at EDUCAUSE last week tells me that these approaches are still WAY out of the mainstream — and the natural instinct for some of us with tiny soapboxes is to compensate by raising our voices: “HEY! LOOK OVER HERE! YOU’RE MISSING SOMETHING!” Scott and D’Arcy have both recently posted links to work that suggests some middle ground between structured and emergent metadata schemes is indeed possible. When I reflect on how quickly the terms of this debate have shifted over the past couple of years, I think there’s reason for optimism that pedagogical peace may soon be at hand.

Then again, by the time that happens, I’ll be bitching and moaning about something else.

I do have some thoughts on “the shape of the collection (and thus the shape of the discipline)” in the age of digital reproduction. But I am not without shame, nor without mercy, and therefore will defer my sparkling insights on this issue for another time — it will no doubt be a post every bit as illuminating as this one…