In this 14-minute screencast I interview Scott Dart, who blogs here, about how tagging works in Vista’s Photo Gallery. I wanted to look over Scott’s shoulder, rather than do this myself with my own photos, because Scott’s been managing a lot of photos in this app for a long time, and he’s in a position to reflect on the evolution of his tag vocabulary.

The metadata storage strategies discussed here lately are just plumbing. What you see in this screencast is the payoff: An application that will be, for many people, the first experience of a style of personal information management that relies on tagging and search as much as, or more than, on folders and navigation.

Conventional wisdom was that people could never be bothered to invest effort in tagging their stuff. What del.icio.us and then Flickr and then a host of other web applications showed is that people will invest that effort if the activation threshold is low and the reward is immediate. On the web, the rewards are both personal (I can more easily find my photos) and broadly social (I can interact not only with friends and family but with like-minded photographers everywhere). On the desktop, the rewards will mainly be personal and more narrowly social (friends and family), though if photos can bring their tags with them when they travel to the cloud, the broader social rewards become available too.

One of the fascinating threads in this screencast is the interplay between foldering and tagging. In principle you don’t need a folder hierarchy rooted in the file system, and doing away with it entirely would reduce the concept count. In practice that’s not yet possible, if only because cameras don’t produce endless streams of uniquely-identified files. When DSCF0004.JPG rolls around again, you have to put it into a different file-system folder than the last time.

It’s too bad, really, because those file-system folders serve little other purpose. They’re conceptual clutter that obstructs your view of tagging, and of tag-oriented search and navigation, which is where all the action really wants to be.

A further complication is that, unlike most of the popular tag systems on the web, tagging in Photo Gallery is hierarchical. You don’t have to use it that way, you could keep a flat list of tags, but the system invites you to nest your tags in a way that seems folderish but that has a magical property. The same thing — not a copy of the thing — can be in two or more places at once.

It’ll be fascinating to observe what people make of this. For example, that magical same-thing-in-two-places property may seem less magical to the majority of folks who don’t know what I know about directory structures on disks. I experience cognitive dissonance when I see a “real” file-system hierarchy and a “virtual” tag hierarchy living in the same navigational tree. But somebody who never had a strong sense of the difference between those two modes might not be bothered at all.

Are people actually using tags to organize and search for their photos? According to Scott, data from the opt-in software quality metrics (SQM) feature — which relays anonymized usage data to product teams for analysis — says that they are.

How private tag vocabularies develop, and what happens when they intersect with the web, are two processes that I’d love to be able to study over time. That raises an interesting question. Can I access that SQM usage data myself? Could groups of willing participants pool their data and do independent analyses of it? It’s our data, there’s no reason why not. Does anyone know how?