Problems with internalizing/socializing classification systems
Even when a company creates a well thought out classification system, things often still go wrong. People put stuff in the wrong place, add a bunch of personal folders somewhere, and at the end of the day, a lot of stuff still can't be found because it's been misclassified or the classification system has been corrupted. Old style classification in real cabinets had the same problem: companies addressed this by making someone responsible for classifying all incoming documents, even though everyone had free access to take stuff out. How do we model a system so that things don't get misclassified into an (otherwise) nice classification system?
My view: categories get internalized only by using them, or even better, creating them yourself. When you create a category chances are you'll use it more or less correctly. When someone else creates one, the probablility of correct filing drops steeply. And the mental effort it takes to understand this categorization approach is big, especially because there are no direct rewards for you.
One approach that may work is distributed metadata. But that idea is in its infancy. So what do we do?