Unquiet

The big to-do among geeks these days is **tagging**, the categorization of everything by words that describe the items in question. At [del.icio.us][del], people tag web pages; on [Flickr!][flickr], pictures; and at [technorati][tech], they tag just about everything. So what’s the big deal? Why is tagging so popular these days, and what benefits does it provide?

[del]: http://del.icio.us/
[flickr]: http://flickr.com/
[tech]: http://technorati.com/

Most people interact with computers in a manner that directly correlates to a filing cabinet — lots of folders, everything existing in only one specific folder. That’s great, if you’re incredibly organized and have a system that remembers where you put everything (like the card catalog of libraries yore), or have the mind to remember the most likely spot for everything. Still, you have _only one_ way to find something stored in your cabinets.

Most modern systems offer search capabilities, tracking of recent documents and other assorted ways to overcome the limitations of such a linear storage system. What if you needed to find a document on somebody else’s computer, but you didn’t know its name or where the person saved it? The limitations of linearly organized systems become apparent when the system must be shared with others; either the group must agree upon a common system of organization (for example, the layout of the system directories in Windows, Linux or Mac OS X), or the group suffers from differences in individual organization styles.

Here’s an example. Joe stores his 2004 state taxes in this way:

Taxes / 2004 / New York

Ann, on the other hand, prefers to organize this way:

2004 / New York / Taxes

Obviously, a good search system would make short work of this example, but it illustrates how different people choose to organize their lives differently. Joe organizes his life by subject; Ann organizes her life by date. So can we find a way to organize that becomes independent of organizational quirks?

Imagine if, in addition to saving a file in a “location” (or folder, or disk, etc., as in current computer systems), the user added tags describing the file (”metadata” in technobabble):

2004, taxes, tax, New York, NY, state, income, W-2, 1099

Tagging relies on free-association as well as the usual organizational styles we see above; all of the same information is present. Even if Scott saves his file in `04/tx/stt/nyrk`, searching for relevant tags makes finding the right file very easy. In essence, tagging allows us to store a single item in many different ‘folders’, thus reducing the likelihood that individual styles of organization slow us down.

Even in tagging systems, individual quirks affect organization, but in shared systems, certain words become more popular than others in describing content. For example, I might tag my site a ‘weblog’ but see that most people prefer the term ‘blog’, so I add that as well. The community defines a certain standard through evolving tag usage (a “folksonomy”, and as users grow used to tagging, they appear to be more liberal with the number of tags they apply. As a result, the number of potential ways to find a thing increases significantly.

###So what am I getting at, anyway?

I got to thinking about tagging as applied to an address book. How could tagging enhance the functionality of my address book, and _more importantly_, how does it enhance my interaction with the people in my life?

At least as applied to people, tags seem to imply _relationships_. Some imply group membership, for example “hiker”. Others, such as “Father” or “Mom”, carry more complex meanings: both imply “family.” So in the context of people, some tags function very much the same as applied to documents, whereas others describe levels of interaction, function, and trust. In addition, descriptive tags can be self-applied, which offers very important benefits.

In the first case, I might like to inform all of my friends who enjoy hiking about a new photo album of my Appalachian Trail thru-hike. Instead of having to remember off the top of my head who likes to hike, I can simply search my contacts for tags such as “hike”, “hiking” and “hiker” and retrieve a list of everyone who might enjoy my photos. This becomes even more powerful if my friends each maintain their own list of tags about themselves in addition to the ones I might have applied to them. Suddenly, I no longer have to rely on my memory at all; my friends opt themselves in to their own interests! Even if they don’t take responsibility for their own profile, I can maintain a list of tags that I apply to my contacts.

Moreover, when I define that certain tags (”Dad”) possess other meanings — for example, familial, trust-related, or genealogical — I layer on functional levels of meaning that I can also use to my advantage. Granting certain functional meanings to a select group of tags (in essence, tagging the tags) creates a method of describing relationships that adapts to input.

If I create a profile and tag **John** with ‘father’, and John (my father) tags **Richard** with ‘brother’, a system that understands familial relationships could a) understand that Richard is kin, and b) Richard is my uncle. Similarly, Richard would automatically see that I am his nephew. If I assign trust meanigns to my extended family, Richard automatically inherits that level of trust. This same concept can be applied to friends, coworkers, and other relationships, both complex and simple.

The [XHTML Friends Network][xfn] describes just some examples of information we could apply to people in our lives. The same benefits of that system apply here: think Friendsterl.icio.us!

[xfn]: http://gmpg.org/xfn/background

###Putting code where my mouth is…

Having spent so much time explaining the idea, I want to explore it more fully in code by developing an address book that implements free-form tagging and shared self-tagging; once a working proof-of-concept exists, then I’d like to approach the second and more complex level - a method to define the layered meanings of a tag and the relationships it implies, and the ability of the system to interpret those definitions into useful results.

Simply, I want to be able to have all my contacts in an address book that allows them to keep their own information up-to-date, to describe themselves in a free-form way with tags, allows me to describe them in a free-form way with tags, to search on those tags and to receive meaningful results (i.e., a list of hikers who are friends). And then, I want my address book to be able to extrapolate information I have not directly entered — all kin who have a profile in the system, regardless of whether I know my direct relationship to them or not. (Privacy concerns, of course, should and will be addressed)

Anybody wanna help?