Andromeda Yelton has covered one of my favorite topics on her blog. In Controlled Vocabulary vs. Tagging: Three Fallacies, she raises some valid and important points. For one thing, she readily admits that when information professionals debate the taxonomy/folksonomy issue they are mostly concerned with how academic researchers will locate material.
Also, in coming up with hybrid solutions that incorporate control and freedom, librarians are often hampered by the pie in the sky dream of a single search (probably a single Google-like search). This idea of unified perfection, that if things are arranged just so, we will realized optimal, nirvana-like findability just does not work in practical application with real world systems and those pesky imperfect humans creating metadata and tags.
Finally, library and information professionals seem to like certainty. Andromeda addresses this idea of guarantees with this astute comment, “I think librarians often assume that, because they are based in rules, controlled vocabulary offers guarantees about metadata completeness or correctness, and tagging cannot.”
There are no guarantees in life, even with controlled vocabulary. In my humble opinion, we should stop fantasizing and start realizing the richness of combining the two methods, mining user generated data and continuing to employ standards for a baseline of control. See more at InfoCamp Berkeley March 5th in my session called You Say Tomato, I Say Aardvark: Taxonomy/Folksonomy Throwdown.