Patrick Lambe sent me a copy of his new book, Organising Knowledge -- Taxonomies, Knowledge and Organisational Effectiveness, after I mentioned that I would love a chance to dive into the book. Given the title, how could I not be curious.
And I have certainly relished the book. Ostensibly, it is geared toward librarians and information scientists, but I see a much wider audience than that - at least into the knowledge management crowd if not further. I particularly enjoyed getting an education on what taxonomies are (and are not) and how taxonomy work relates to knowledge and to organizational effectiveness, which encompasses the first half (or more) of the book.
So, what is a taxonomy? My default assumption of a tree structure is only one example of a taxonomy. It turns out there is much more flexibility and purpose in taxonomy work than I knew. But what is it? Lambe gives three aspects to taxonomy (p. 5):
- A taxonomy is a classification scheme
- A taxonomy is semantic
- A taxonomy is knowledge map
The classification scheme is the one where most of us probably stop. The semantic element embodies the idea that the terms in the taxonomy mean something to the community for which the taxonomy has been developed. And those terms have some relation to one another, as described in the taxonomy. This leads to the knowledge map aspect: the taxonomy can be looked upon as a way to describe how the community thinks about the content. It is a map of what is known or what is covered by the community in question. One needn't even have content to get an understanding of what the community is about. Taking these last two elements together implies that different communities can very easily have different-looking taxonomies, even if they cover similar materials.
So, that was the "a-ha!" at the start of the book. But at the end of the book, Lambe gave me another one. Taxonomies are not intended to be molded in concrete, never to change. They bring together the desire to order things with the fact that people keep wanting to rearrange the "things" as they get used. He even suggests that when things stop changing, the taxonomy dries up and gets boring too. One of his final comments is that the taxonomy helps make sense out of the changing landscape of human behavior.
An effective taxonomy sits between Chaos and Order and mediates the two; it does not, as it's so often assumed, represent the domain of Order unequivocally. (p. 259)
Another take-away for me had to do with how users interact with taxonomies, rather than how taxonomies need to be formalized and constructed. People don't begin thinking about a topic at the top level of a taxonomy, nor do they begin at the bottom. They start in the middle and traverse up or down or sideways depending on whether the concepts at that level of the taxonomy fit the situation in which they find themselves. This concept is intriguing because when we are outside a problem, looking in, we want to see the whole tree or hierarchy of information. But that is not how it works for individuals in the trenches. With a geographic map, you don't present someone with a map of the entire state of Ohio when they are looking for a street address in Mentor. Similarly, don't show them Ambrose Drive when they want to get from Mentor to Bay Village.
Counter to Lambe's suggestion that people start traversing at some mid-point in a map, David Weinberger has an interesting statement that "knowledge starts at the miscellaneous." There is a default assumption in our culture - particularly in light of discussions of taxonomy - that the miscellaneous is where knowledge ends. Weinberger has said this many times, including in his new book, Everything is Miscellaneous. (I picked up his talk for the Library of Congress Series on Digital Futures, where he discussed the idea of the miscellaneous in 2004, on Audible as a free download.)
I took a bunch of other notes with the thought that I might dump them into this review, but this has gotten long enough as it is. I include a few interesting tidbits below the fold, if you are interested in reading some more.
I note that James Robertson has also reviewed the book. Feel free to go read the book yourself.
Other highlights from Organizing Knowledge
Taxonomies can be represented as anything from a list to a tree to a concept map to a matrix to faceted trees to hierarchies. The final structure of a specific taxonomy depends on what best fits for the community and the content being taxonomized.
Taxonomies are designed to keep us together and to keep you out. This is another example of Lambe turning common ideas on their heads with his discussion. In building a semantically relevant depiction of the knowledge of the community, you are guaranteed to keep people outside of the community in the dark. To encourage cross-functional communications, you need to engage in boundary-spanning activities. Maybe even creating a boundary-spanning taxonomy, but this will not represent all of the knowledge of both domains. Rather, it will provide the translation between the domains and give people a means to start a communication with each other.
All categorization is subjective and fluid. It will be different tomorrow. Taxonomies aren't carved in stone, even if we think of them that way.
Since taxonomies are never "perfect," Lambe notes that human connections can make up for the holes in the taxonomy. However, in strong social settings, the holes can easily be hidden from view. In a later section, Lambe refers to societies that are high context vs. low context in the same way. High context societies fill in a lot of holes, based on contextual queues that are non-obvious to outsiders.
In fact, there can be "bad taxonomies" that reinforce old or incorrect ideas. Since taxonomies act as a map of knowledge, if they ignore or obscure an important aspect of what can be known, then it is very difficult to overcome this obfuscation. If the concept isn't visible, then it is possible to never address it. This has particularly bad ramifications in the social milieu, but it can also limit innovation and vision within organizations.
"Before a taxonomy is implemented, it is simply a theory about how people think (p. 202)." I like the connection between this idea and the idea in project management that a project plan is merely the map of how we think the project will go. An active project, just like an implemented taxonomy, is always going to go different ways than we think. Taxonomies are fluid, after all.
Why do corporate websites die? Why are (most) academic websites so boring? They are overly-homogeneous and structured. They don't allow for diversity and flexibility and change. As soon as they are rolled out, they are outdated.
Taxonomy work must integrate with the other work happening in the organization. If it doesn't, there are sure to be conflicts. Lambe talks about information architecture, whose primary role is arranging content for use by the community. The primary role of taxonomy is in creating a means to organize the warehouse. Lambe also addresses the needs of records management, which is much more interested in ensuring the reliability and accessibility of business records than of designing a community-specific taxonomy. But these two groups (and many others) must be considered when dealing with taxonomy, as they are sure to impact or be impacted by taxonomy work.
Lambe also addressed the future of taxonomy work in his final chapter, providing some useful thoughts about how social tagging (folksonomies) and ontologies relate to taxonomy work. They all operate along a continuum of precision/ambiguity (and several other factors). And they can work in conjunction.