I’ve been working on a (secret-ish!) side project recently that hopes to index a lot of profiles, and allow people to browse through and compare those profiles as convenient.

One thing I’m struggling with at this early stage is the taxonomy architecture. For example, one taxonomy will probably end up being based on geography. My first impulse was to structure the taxonomy like this: USA -Northeast –New England —Massachusetts —-Greater Boston Area —–Somerville -Southeast –Etc, etc, etc.

I have since been informed that this is insane, or at least a moderately bad idea, particularly in Drupal. So my next idea might be equally insane: Just allow everything to be free tagged with vague prompts and a catch-all parent taxonomy, or perhaps assign some root taxonomies (location, genre, medium) with free tagging. Intuitively, it feels like another moderately bad idea to me, but take heart from the highly apocryphal story I was told as a youth about MIT: School administrators, tired of students ignoring paths and treading right over the grass, decided to leave new quads unpaved entirely. After a year, the university planners went back and paved over the areas where no grass had grown due to foot traffic.

Peter Merholz has a great blog post about a similar phenomenon at UC Berkeley, except administrators took the opposite tack, erecting barriers to prevent such “cow paths.” Merholz goes into why this was a bad idea, but to quote the late Michael Chrichton, “Life will find a way.” Since the project’s initial users will be a relatively select bunch, I’ll take the (apparently urban legend) MIT approach.

P.S.: I couldn’t find any actual sources for this particular myth, but perhaps it was related to The Oregon Experiment, a try at communal design at Oregon University.