Steve Rubel makes an excellent point regarding the power of folksonomy in response to a Business Week article on tagging
The BW article also underscores the rising importance of tags for marketers. Often I am asked why I think tags are a big deal. The reason is simple - it makes consumer-generated content a lot more discoverable.
Micro Persuasion: When It Comes to Tagging, You're It
Folksonomy is not a threat to search. Rather, it enhances search by making the meaning of links more explicit. As I discussed five days ago, this is one of the chief motivators behind xFolk (Josh Porter take note).
Steve further discusses how search engines might use tagging:
As I told Wired News earlier this year, I believe that Yahoo and Google will eventually enable their users categorize and share searches under specific tags.
Micro Persuasion: When It Comes to Tagging, You're It
Not so clear that this is the way they will go. What would make much more sense is for people to tag individual links returned by a search for meaning. Google and other major search engines could then use this information to determine the relevance the person found in a particular link given the search keywords. Think this is all theoretical. Yahoo may have already started down this path by purchasing flickr. Further, it's one of the uses you can currently make of technorati tags, particularly if you factor in attention.xml. It's also one of the ideas behind feedmarker.
I'm going to ask why more often. This is great, Bud.
What I find interesting is that what search is doing and what folksonomies are doing is basically the same thing: aggregating explicit behavior. The explicit behavior in search results is the creation of a link. The explicit behavior in folksonomies is the creation of a tag. Multiply those by several orders of magnitude, and you've got yourself a recommender system.
What this suggests to me is that we're finally getting around to creating tools that take advantage of actual user behavior...and finding out that it works well (or at least as good as alternatives!). It also begs the question: what other explicit behaviors can be aggregated?