Bottoms-Up!

Flipping through the 2nd edition of the Polar Bear book, I found an interesting section that portends the use of bottoms-up, user generated, emergent navigation systems, my latest muse. I did not expect to find such a passage, given Peter Morville’s recent talk on folksonomies at the IASummit, but nonetheless both he and Rosenfeld we able to portend the coming change in navigation systems even a couple years ago.

The passage begins on page 355. Here it is:

The practice of information architecture has come a long way since the early 1990s. We began with highly centralized, top-down approaches, attempting to leverage careful planning into stable solutions. We did some good work, but learned the hard way that change is a constant and surprises should be expected. More recently, we’ve been exploring bottom-up approaches that tap the distributed intelligence within our organizations to nuture emergent, adaptive solutions. The following table compares classic or “top-down” IA to modern or “bottom-up” IA:

Classic IA	Modern IA
Prescriptive	Descriptive
Top-Down	Bottom-Up
Planned	Emergent
Stable	Adaptive
Centralized	Distributed

As we struggle with these ideas, an interesting question arises. Do we create information architectures or reveal them?

So, given that the 2nd edition of the Polar Bear book was published almost 3 years ago, the authors have certainly had time to get used to things like folksonomies coming along and stirring things up. After all, the table displayed above is no less than a great overview of the differences between taxonomies of the classic IA and folksonomies of the modern IA. So, my dear reader, why is Morville still bearish on folksonomies…? Even the wonderful folksonomy video post by Peter Van Dijck doesn’t shed light on this.

Dan Bricklin (co-creator of Visicalc, the first spreadsheet) on the February 4th edition of the Gillmor Gang, quickly talks about how information architecture has changed. He mentions that Google was the big surprise, cutting through the riff-raff in a time when much research was being done on automated taxonomies and hierarchical systems.

Bricklin makes a distinction that I think is hugely important: automation. Automation means “automatically controlled”: to be automated means to have controlling decisions made without conscious or voluntary intervention.

But what about Google, you ask? Isn’t that automated? Surely, people aren’t sitting there deciding which links to give a higher pagerank to, right?

Actually, that’s exactly what’s going on here, but the people in question don’t work for Google. The people in question are the ones making the decisions about what is valuable to them by publicly linking to another URI. (In other words, the people actually making the web sites.) This is the important part of Google, and it is NOT automated. What is automated is the aggregation and display of this behavior. It’s a subtle, but vital, difference.

Also, at the E-Tech summit del.icio.us’ creator Joshua Schacter brings up another important point related to this topic (folksonomies in particular). As quoted in Cory Doctorow’s impressionistic transcript:

“We have different axes of why you’re tagging, what you’re tagging and how it happens. Flickr you mostly tag your stuff for your own purposes, in Technorati, it’s your stuff for others’ purposes, and in Delicious, it’s others’ stuff for your purposes. They’re different things, they don’t necessarily flow together.”

One of the things that this tidbit shows is how different folksonomies can be, even though they’ve been lumped together by those talking about them (including me). Given that, and the growing interest in these things, it seems as though we’re only just beginning to scratch the surface in terms of bottoms-up architectures.