One issue here is that there are far more code contributions (and snippets - which I think can also benefit from categorization) than there are tutorials, faqlets and categorized questions:

Nodetype	count
sitefaqlets	145
tutorials	229
categorized questions	808
sourcecodes	1635
snippets	1834

I also think this is a useful way of aiding research, but I wonder how efficient it would be to maintain a categorization of such a large pool of nodes, both in terms of database load (is nodegroup the right construct for this?) and for the users that would have to maintain the categories.

curators++ (I work with real curators - they are a delightful bunch!)