The concept of tagging provokes two responses in librarians, often simultaneously: excited joy and abject fear. Joy at the concept of user participation encouraging the semantic relevance of data and fear that allowing users to control the process of subject analysis via uncontrolled keywords is the quickest route to information anarchy in history. I gave a paper on this very topic at the Cataloguing and Indexing group's annual conference 2008.
The conclusion to which I came was that there is a place for both types of information organisation: controlled vocabularies to promote sturctured subject analysis and keyword tagging. The problems with tags are numerous and have already been discussed at length: they use natural language with no cross references, homonyms lead to misguided searching (try searching for Turkey - bird or country?) and differences in spelling within a language across countries is not cross referenced with each variant. No one tells people whether to use china or China when talking about the country, even though in the physical world the initial letter should be capitalised. This leads to vast amounts of relevant information potentially not being retrieved. The worst problem of all concerns the motivation behind tagging: a great majority of web users tag primarily for their own information organisational needs and not to provide subject analysis for others. A cursory glance at a book on Librarything.com tells us that tags are used for the reader/owner's benefit as well as to analyse the subject of the book.
I personally use 'read 2009' and 'tbr' as tags to track my own reading habits. Other users can ignore this but it does provide another dimension to tagging. People also rely on the cconsistency of taggers; often this doesn't happen and the same concept ends up being represented in multiple ways.
Does this matter? Isn't there an interest there for researchers to look at how different people semantically translate a book's concepts? One woman's comedy might be another's tragedy. It is interesting to search on tags across the whole librarything catalogue. Which books are mostly 'unread'? Which books are in everyone's 'to be read' pile? Isn't it fascinating to see the size of the tag words change depending on how many people have used them? The wisdom of the crowd is a phenomena that can and does often work. The trick is to get as many people as possible tagging. Although this might sound like I am encouraging the data anarchy, somehow it seems to be that from the uncontrolled comes forth a consensus. Not a controlled vocabulary per se, but some form of organisation. If a book or a blog mostly uses certain tags, others are often encouraged to use tags that are already there. Whether or not this actually destroys the concept of tagging as an uncontrolled force is another debate!
I am interested in the use of tagging within the library catalogue. Don't recoil in horror! I have been a cataloguer and am very aware of the need for a controlled backbone of structured vocabulary with inbuilt semantic meaning and cross references. LCSH should not be replaced. I am proposing another type of subject analysis running parallel with the catalogue records. This library's catalogue is an inspired model whereby user's tags run parallel to catalogue records. The tags do not come into contact at all with the actual MARC21 coding. The tags are stored separately and are dynamically loaded together. It is a radical step to take, to be sure, but the possibilities are endless. The name of the paper could be added as a tag to books on current reading lists. Finding information would become that bit more flexible for our users without losing any of the important work done behind the scenes by LCSH. It is certainly something to ponder. Anyway, back to Librarians who Librarything...