Folksonomies - Majority Rules & Tagging Abuse

I was flipping through the Shifted Librarian's notes on Liz Lawley's keynote at this year's Internet Librarian. Lots of good stuff there, but the part that struck me was on folksonomies and tagging. First of all, let me recap from the notes:

"tagging is NOT letting you see the long tail, because good folksonomies depend on critical mass and you’ll lose the smaller pieces

nobody cares enough about that obscure government document to tag it

when building tools, can look at what people call things on del.icio.us and then use that in the tools

what happens when people start relying on these tagging tools

do I really want a majority rules approach to information retrieval?"

Now I do recognize that Tagging is not going away any time soon, and will be an option for 'rough' collection classification for the foreseeable future, but Liz Lawley's comments are simply spot on! It definitely is a 'majority rules' approach, and the gaps left in a purely user classified system could be substantial (perhaps depending on the audience?). For Librarians, our expertise revolves around 'the long tail' and our ability to locate information within the smallest niches. As such, Librarians shouldn't be too quick to buy into the use of tagging for important content collections.

For me, this is more than interesting - it's an important factor in considering Folksonomy classifications. Debatable? Perhaps, but a necessary consideration.

I haven't commented on the use of folksonomies a whole lot. While the concept isn't new to me, I do have a few concerns, and like many others, am still considering the 'when & where' of practical application.

My biggest issues surrounding Folksonomies are the lack of synonym and homonym control, and the huge potential for abuse & tag spamming. I 'might' be able to get beyond the whole lack of a 'controlled vocabulary' thing, but 'tagging abuse' is a tough one. The value of tagging (beyond offering a 'personalized classification' system - which I think is a good thing) seems to occur when the number of users hit critical mass (as noted in the quote). If we need many users to make the system work, but can't trust the users not to spam the system, there's a problem.

The long and short of it is, I can't help but be a fence sitter on this one. Folksonomies look useful: for personalized classification, for identifying preferred terms for Controlled Vocabularies (perhaps a preliminary classification stepping stone?), and for smaller 'trusted groups' of users - the potential does seem to be there...

Let's just say, I'm 'getting to yes' slowly.