I recently had an email exchange with Lou Rosenfeld who started a whole controversy almost by accident about something called folksonomies. There are many definitions of what folksonomies are (this, this and this for example). Simply stated, a folksonomy is an ad hoc classification system for archival material developed on the fly by users as they enter it. One clear example is a web service called del.icio.us where people store personal web bookmarks online. Del.icio.us implements the idea of folksonomy by allowing users individually to tag each bookmark with a word or series of words that describe it. In aggregate, all of these user tags form a sort of folk taxonomy of the web sites stored, albeit a different taxonomy for each individual user. So, what's the point?
To figure this out, I conducted a little experiment that I wrote Lou about. I started keeping my own del.icio.us bookmarks. I started using just tags only for over three hundred bookmarks. Then, I had an awakening and started to create blog postings based on the annotations I had started making to my bookmarks to disambiguate things. These led me to think about my audience, and I am finding that the blog postings, not the tags applied to bookmarks are generating interactions with people. The tags are useful to me for quick retrieval. I have had little success searching on tags by themselves.
I have come to the conclusion that the tags divorced from some knowledge of the individual doing the tagging are too ambiguous to provide much value in discriminating what is useful. There needs to be more. One thing that might be useful is an easier way to see who has bookmarked things you have bookmarked. No matter how they tag their bookmark, you may have something in common with them, and the rest of their archive may be worth investigating. In this light, tags become a window onto the user's idiosyncratic view of the world vs. a means of sharing.
Here is the description of my evolution as I provided it to Lou interspersed with various “epiphanies” as they arose:
- I like tagging because it is easy, and it is more flexible than many category systems where you have to define your tags in advance and then use them.
- I am actually finding that tagging can also be hard. In particular, now into month three of tagging, what do some of my tags mean? There is tags drift. I started this as an effort to categorize bookmarks for future reference. Well, unclear that that is working entirely. But, I have this great Mac app called cocoal.icio.us that allows me to search my delicious metadata (i.e., data about the bookmarks I was posting), and that is useful.
- About a month ago, I started to write extended descriptions for all of my delicious bookmarks. I wanted richer metadata that helped me better remember why the particular site was important and to make search with cocoal.icio.us easier.
- Shortly after, I decided to start republishing my delicious bookmarks as a mini-blog (http://thecommunityengine.com/webcites) to make it more visible because I wanted some place where I could uniquely get credit for all of this commentary I was generating.
- Republishing my delicious bookmarks as a mini-blog with commentary has caused me to think of my audience. That turns out to be the people who I am bookmarking since frequently I am bookmarking blogposts. My weblog software automatically tracks back to them (i.e., let's them know I am commenting on them), so there is a good chance they will see my trackbacks and read the post. There is evidence that this has begun to happen. In other words, because of the mini-blog, my delicious behavior, which started as a personal archiving effort has become very social.
- Tagging by contrast has remained asocial, but I have started to try to find things by tag. Google search may be better.
- As far as I can tell, all of Liz Lawley's thought experiments on the topic of tagging lead me to the opposite of her conclusions. I offer as Exhibit A this gem. She brings up this CMU ESP game as an example of the vulgarization that comes about as a result of having to find the least common denominator to create shared tags. My reaction, “GREAT, people are finding a way to come to agreement. Maybe they should add this game to del.icio.us as a nice non-centralized way to bring some order to the tag cacophony .”
I've been using "folksonomies" i.e., misspelled words, for five years. I worked for a site selling music t-shirts. I would data mine the Customer Service emails for folksonomies of band names, e.g., Metellica or Matallica or Metalica. All of the misspellings were phonetically correct for the region of the country where the messages originated. We set the folksonomies in the <head> section; we'd get hits on those variant spellings. I have done this for all commercial sites I've work with. I will read some or most of the Customer Service emails to find appropriate folksonomic words for that particular site. I'm certain that someone will invent a program for generating folksonomies but anyone reading emails can do it. They could be a Folksonomy Technician. "Folksonomy." Nice scientific name.
My husband put together an online bookmarking site a couple of years ago (http://www.connectedy.com). I have since taken it over, since he works another job now. Essentially it lets you create an account and convert your browser bookmarks into SEO folksonomized web pages, which can be further edited by the user. The categories and links mimic folders and files, but what we have realized recently is that the categories act like tags on the links. The other features that the site already offered making tagging actually work well. That is, it's easy to do. You can put your tags/categories into a hierarchy and search on them, and I guess it wouldn't be that hard to merge all the categories together into a communal folksonomic stew like on del.icio.us. I don't know where this field is headed, but right now it is pretty exciting.