I'm new to OF and the software proved to be very powerful and flexible but faceted classification is very simply a must in the digital world when knowledge is to be organized properly in relation with the particular way people think about their organized information.
http://en.wikipedia.org/wiki/Faceted_classification
The reason why faceted classification is so convenient is because it aligns very strongly with how human beings think about and organize information.
As somebody already mentioned, and in a strict faceted sense, disjointed contexts are a real problem and are cleanly addressed under the canonical faceted model (but will cause problems in OF because of a lack of an implementation of the faceted model) but overlapping facets are also extremely convenient from a human standpoint.
An example would be:
walk fido at park - Dog, Fido, Park, Exercise.
Taken in isolation, those contexts (Dog, Fido, Park, Exercise) might not actually yield the required preconditions for me to be able to walk my dog at the park.
For example if I so happen to drive by the "Park" and the Park context pops up my "walk the dog at park" action item, it is completely useless if I my dog is at home. It's a reminder I can't currently accomplish.
Or if decide to go for a workout at the gym and Exercise pops "walk the dog at park", I'm absolutely not interested in walking my dog at the park. It again acts as a reminder I can't currently accomplish.
Those are all overlapping contexts/facets I might be interested in being able to tag/filter against (I don't actually have a dog to walk but the point is the same nonetheless) :-P). This added capability would absolutely not hinder the use of single contexts for single context/facet proponents but it would add significant power to the currently implemented single facet model under OF.
It creates an interesting paradigm shift though. Instead of working with hierarchies of organized tasks in relation to a single context, the hierarchical organization is somewhat flatter, making way for on the fly faceted meta data searches based on multiple contexts. I think that both can coexist in harmony :).
Situations can always be described in either strict hierarchical terms via static nested single facet/contexts (as we are more familiar with when working with folders, for example, under multiple OSes) but it is also possible, and sometimes very desirable, to use multiple facets (that may even be considered to reside on different semantic levels in more traditional knowledge hierarchies. Think of google searches) to corner just the right context*s* relevant to perform a given task.
I think that faceted classification, for serious knowledge organization and to prevent triggering against insufficient overall context, is unavoidable.
Or put differently, a single context is insufficient to describe accurately our multi contextual (multi faceted) reality.
omniinmo