Creating a Folksonomy .:. grubbykid.com

Creating a Folksonomy

For anyone who has struggled to create or use a system to classify objects in a logical fashion, one often finds that once you add one other person to the system, the system breaks. The more people you add to the system, the greater the system breaks. Your beautiful classification system has been rendered neither easy nor beautiful, and consistency breaks down. I want to take a bit and talk about a newer way of thinking about classification.

All is not lost! Enter the Folksonomy. Webservices such as del.icio.us and Flickr, including a project yours truly is working on, are working toward a solution. The key are tags. Say you have a photo of the Chrysler Building, how do you classify it? It is in New York, is a building, is tall, the photo is black & white, and from a vintage camera. In a hierarchical system, what would be your top-level classification be? Physical location, or time, or type of media? Well, actually is should be all of the above. Instead of a hierarchal system like in your browser bookmarks file, you have a series of tags which describe the object. The key, is that the user is the one doing the describing, and thus the categorizing. The magic comes about when you combine tags together - you can create an algorithm of connections which records how many times different tags are combined to create connections. Back to our example, if "New York" and "Building" are combined on multiple photos, then there becomes a relational link between the two based on frequency of the two (or multiple) tags occurring together.

On a more philosophical level, this is the manifestation of talking about decentralized information systems. How long has the literary world talked about monads, Leibniz, Deleuze and his Rhizome? Tags are the real world implementation of the Rhizome. That is, a rhizome is an "acentered, nonhierarchical, nonsignifying system without a General and without an organizing memory or central automation." The hive mind which we have seen operate in the animal world is beginning to order our webservices. Instead of top-down heirarchies, we have small, simple instructions operating on small bits of information, which create massive complexity. This is exactly what Stephen Wolfram discussed in A New Kind of Science (if you get passed his BS) with his Cellular Automata.

This might seem like a small switch, but it has massive consequences. Information Architects spend time and money trying to come up with schemes, when they could just let users tag data. I think Gene Smith slightly misses the point with tags. Gene says:

I think folksonomies can work well for certain kinds of information because they offer a small reward for using one of the popular categories (such as your photo appearing on a popular page).

The point isn't about getting my photos or del.icio.us links onto the front page. I could care less. It is about creating a system of classification that is flexible enough for the user to use, and for the tags to become aggregated with each other.

Another interesting discussion at PlasticBag is entitled, Towards tag-based bookmark management in web browsers. It is a lengthy article, but seems that Tom is reinventing the wheel a bit.

To summarize the problems with current bookmarking systems then, we could say that (1) the process is slow and annoying (2) that it requires us to continually refine and redevelop our taxonomies if we're going to keep track of everything, (3) that URLs can belong in a number of bins and that (4) we can be left with unmanageably large lists. An ideal system would therefore speed the process up of both bookmarking a site and retrieving it later.

He is correct and his ideas are admirable - hierarchical bookmarking is broken - and something needs to happen. However, dependence on any local system is also part of the problem - portability and locality hamper free transfer of the bookmarks. Users move between multiple computers daily, and computers break down. If your new fangled local tagging systems works great, but then your computer dies along with your data, you are back to square one. del.icio.us solves most of Tom's problems in terms of simple tag-based structure, and ease of use, and solves the portability and locality problem.

I can't stress how much del.icio.us has changed the way I "bookmark" things. I no longer bookmark items locally - I just popup the del.icio.us tagger, and bam there it is. I have been experimenting with pulling the del.icio.us bookmarks to a local "backup" every time I login to my powerbook in order to seed QuickSilver's bookmarks plugin for use. I also have been experimenting with tagging sites which I visit daily using book:per.pol, book:nyc, and book:blog which I then pull into a "links" page via MagPie RSS (with full caching on to spare Joshua the load). I chose those tags on purpose because those are "special" tags which I wouldn't use on a daily tagging basis. Additionally, I pull in my inbox links via MagPie to populate the bottom of the links page. You could very well create a web interface which is exactly like what Tom advocates in his mockup, but the del.icio.us interface is constantly improving, so I would start from where Joshua has left off.

Of course, if Joshua takes his ball home, all 1300 or so of my bookmarks are toast. Perhaps a solution is some sort of backup, both locally and to another webserver would be a solution, but then you get into a whole 'nother problem of synchronizing the backup.

20041130 - Edited for clarification

Trackback Pings

TrackBack URL for this entry, Creating a Folksonomy:

January 17, 2005 · Grubbykid.com :: Words - Software Cruft and Hacks
January 19, 2005 · Grubbykid.com :: Links - MetaFilter gets tags

Comments

Tom Coates says:

I didn't really go into as much detail as I might have liked, which is always the problem with my longer pieces. Certainly I think pretty much any app that contains data that you could sync online should sync online, and I'm also very keen on leaving plugs and sockets exposed so that people using one system can plug other ones in to it, creating as a consequence some kind of decentralised or centralised social system for using them. However, we do have a concept of locally held bookmarks in our applications (we in fact still also have local applications, which may change but hasn't yet) and I think that must be considered likely to continue and worthy of improvement. I'd a bit nervy about leaving all my links and content under someone else's control which is precisely why I moved from Blogger to MT and for this reason at the moment I'm uncomfortable moving towards del.icio.us at the moment. But even if my concerns were totally resolved and I'm not convinced that I would use del.icio.us to the exclusion of my own local browser. More likely they'd be used in parallel - it's quite conceivable to me that the local / public binary might be a useful distinction to keep in deciding how to keep track of personal -> publically appropriate links (where del.icio.us would be a subset of 'all links I want to store').

Posted by: Tom Coates at October 6, 2004 1:13 AM #

Tanya says:

"Instead of a hierarchal system like in your browser bookmarks file, you have a series of tags which describe the object." Ah, you're also tapping into the power of a faceted classification system. Way more powerful than the strictly hierarchical. Doesn't have anything to do with whether the system is distributed (folksonomy) or centralized -- can be either. Have a look at A Primer on Faceted Navigation and Guided Navigation and perhaps some links from here.

That strictly hierarchical stuff is so 19th century. Let's stamp it out.

Posted by: Tanya at December 1, 2004 3:58 AM #

striatic says:

"Of course, if Joshua takes his ball home, all 1300 or so of my bookmarks are toast. Perhaps a solution is some sort of backup, both locally and to another webserver would be a solution, but then you get into a whole 'nother problem of synchronizing the backup."

like import/export OPML with feed readers?