Folksonomy Notes: Considering the Downsides, Behavioral Trends, and Adaptation

After my exuberant post on the powers of the Del.icio.us folksonomy, I got to thinking and reading more about using del.icio.us as a tool. I found that other folks are talking about folksonomies, too, and not just the benefits.

Liz Lawley posts about the social consequences of social tagging on the Many-2-Many blog. Her concern, if I understand it right, is that folks like me (with our “technological-driven optimism”) aren’t considering the implications of the tools we’re promoting.

She points to how users can game the del.icio.us system by tagging their entries inappropriately, a disruptive behavior which she sees as inevitable. She hopes that recent comments by Rebecca Blood heighten our awareness of such ne’erdowells. In response, I put forth this challenge: show me a perfect system and I’ll show you one that can be gamed.

What Liz also talks about, which I find much more intriguing, is the idea that users begin tagging their bookmarks differently after observing how others are tagging. I’ve done this myself. For instance, I didn’t realize that there was a tag called “del.icio.us”, so I added that tag to my latest bookmark, and I’ll probably add it to the bookmark for this post.

One thing I think we have to remember through all this is that, because of the nature of the system — that it relies on behavioral trends of many people — one person cannot make a significant negative (or positive) impact. That is the beauty of del.icio.us: you can make a slight impact one way or another, but if it proves effective others will catch on, and they’ll do it to. Thus, equilibrium can once again be attained.

In addition, none of this is new. We’ve been through this before with just about every way that we communicate. Consider the local town meeting. Ever notice that people try to get their voice heard by talking louder than others? That’s gaming the system, if you think about it.

Peter Merholz wrote a piece last October called Metadata for the Masses on what he calls both “ethnoclassification” and “free tagging”. (He later discusses the correct classification on his blog, like the good IA that he is, throwing in “mob indexing” and “social tagging”). Apparently these are rough synonyms for folksonomy. In his first piece, though, he suggests that a good strategy for building IA is to use folksonomies to see the paths that people are using, and then go ahead and pave the most popular paths. Presumably, this would mean to create some sort of “final” IA — one that would serve as the production IA, the one people actually used, which covers all the bases.

Compare that approach with the similar approach suggested by Lou Rosenfield. His idea is also that folksonomies might be used to help develop “richer, more current controlled vocabularies that can evolve to best support findability”.

I think we’re in risk of a myth being perpetuated here. It’s the myth that if the Information Architect knows where something is within their own controlled vocabulary, then other people will be able to find it better. For some systems, this may be true. For others, it may not be true. I understand the motivation to have “control” over our vocabularies, but I think we may be seeing exceptions to the rule. After all, we don’t really have control over our languages, we simply use words, phrases, and concepts how we see fit. The more we observe actual implementations of folksomonies, though, the better handle we’ll have on this.

With the consideration of control in mind, let me ask you one question: Which do you use more, Google Directory or Google Search?

Ok, that might not be a fair comparison to the current situation. But it does highlight one overwhelming example of an algorithm based on human behavior trumping one performed by human “experts”. I daresay that it’s not because the experts aren’t expert enough. I think it has to do with the amazingly complex algorithm that people use when they speak, write, and communicate words: the “whatever works” algorithm.

The other day I noticed that I was saying the word “like” a lot. I said it so much, in fact, that I sounded like a valley-girl high schooler talking on the phone. I would say “like, you know what I mean?” or “I don’t know, like, how to explain it like exactly”. When I recognized what I was doing, I wanted to exorcise the word from my vocabulary completely and forever. What crude English, I thought. My wife Alana, who was the recipient of such horrible English, didn’t seem to mind: she knew what I was talking about.

The del.icio.us folksonomy has succeeded so far because it harnesses user behavior instead of dictating it. Where people refer to the same things slightly differently, in crude English, del.icio.us does, too. If I refer to this phenomenon as “folksonomy”, someone else might refer to it as “free tagging”. In del.icio.us, this is normal because it is normal for human beings to do it.

One thing that I mentioned in response to Liz’s post was that I feel we should keep in mind how adaptive we humans are. It is a fundamental talent we have. Too often, I think, we ignore this quality, pushing for consistency over everything else, when all we need is a little explanation of how things work. Once we know how they work, we’re fine.

I’m not advocating a willy-nilly approach to designing architectures for humans: I’m advocating a willy-nilly approach to designing architectures by humans, who use a willy-nilly approach when reading and writing and speaking words.

The instinct to have one “controlled” information architecture for any set of information should be killed. People don’t all refer to the same things in the same way, and neither should our applications, web sites, or tools. If you look closely, we’re being told this every day, most notably in the growing use of content aggregators (including search engines), which essentially allow people to create their own IA outside of the domain where the content lies.

The benefits of finding something “in one place” are nothing compared to the benefits of finding something “in the place where I want or happen to look”.

Clay Shirky responds to Liz Lawley’s post by pointing out that folksonomies are a forced move.

My folksonomy+downside bookmarks:
http://del.icio.us/bokardo/folksonomy+downside

Interestingly, I could not get del.icio.us to spit back html for my multiple tag request, so I could not embed my folksonomy+downside links directly into this page.

Whereas this works:
http://del.icio.us/html/bokardo/folksonomy

This doesn’t work:
http://del.icio.us/html/bokardo/folksonomy+downside