Hmm... I suppose that there IS a big difference between sampling and categorization, though I thought from the exercise we were supposed to aim for coming up with samples that represent the whole of who we're trying to reach, as opposed to simply outlining several different samples within our population. I probably read that wrong and ran with it :-/ Still, I have my concerns about the suitability of this technique for designing learning environments.
User interactions in a software application are very different (in my mind) than open-ended learning experiences. In a software system, the mechanisms are fixed in place and any flexibilty must be designed for. Similarly, when you write a book or article it's more or less fixed in time: I use this technique for my work by writing for a couple intermediate developers I know well, and it works phenomenally. But I think that's mainly because I'm packaging up a certain "lesson" rather than having a two-way conversation when I write.
In a human system, improvisation and just-in-time customization is possible, and so I don't know how well this technique applies beyond learning programs that have a standard curriculum for all their students rather than allowing their students to come up with their own projects and open ended questions.
All that having been said I think the issue is mostly that Mendicant University is an edge case. Going through the process of trying to fulfill the spirit of this task was a good learning/thinking experience for me, even if I struggled with it.