Bulletin, February/March 2007

IA Column

Taxonomy Out of the Box

by David Weinberger

David Weinberger (self<at>evident.com) is a Fellow at Harvard's Berkman Institute for Internet & Society. His book Everything Is Miscellaneous will be published in May 2007 (Times Books).

Taxonomizing is about more than just drawing boxes and connecting them with lines. Taxonomies - at least some of them - reveal the order of things. They increase knowledge by manifesting multifaceted relationships among things. In that light, tagging and folksonomies look like the vulgarizing of knowledge, and well-bred taxonomies turn up their perky noses at the ill-manner interlopers. But the new taxonomizing does more than increase knowledge. It reveals meaning. 

If Adam and Eve had followed the command to name the animals by calling this animal “Fido” and that one “Spot” instead of calling both “dog,” we’d have no language. Taxonomies arose when we felt the need to straighten up the ambiguities and messiness of language, drawing neat boxes, connecting them with lines, and letting our nesting instinct subsume dogs under mammals, mammals under animals, and animals under physical things. And when taxonomists found themselves making such remarks as, “That’s not really a tree. It’s really a big bush,” they took the metaphysical leap from organizing thoughts to revealing reality. 

In some realms, this leap is unproblematic, or at least less so. Numbers are defined as containing integers. Germany has legal borders that as of 1989 unambiguously contain Berlin. But when we move out of the realm of the axiomatic, the stipulated, and the politically determined, the ways we nest our world become more arbitrary. Some taxonomies, we assume, represent real divisions. Darwinism, for example, lets us build taxonomic trees based on causal relationships - couplings - among historic ancestors. Linnaeus, on the other hand, had pretty well given up on figuring out what God’s own order was. Instead, he proposed a classification system that would help provide us humans with a common set of references. Freed to draw lines that would be most helpful to scientists, he relied on easily observed physical characteristics - plant sex organs, for example - that would enable anyone with eyes and a guide book to figure out what she was looking at. It may not have perfectly mirrored God’s own order, but it let scientists know they were talking about the same species.

Modern taxonomies reflect this bifurcation. Some try to express real divisions in nature: the order of species, the periodic clustering of the chemical elements, the division of the universe into spatial neighborhoods such as galaxies and solar systems, the dividing of museums into wings arranged by time or type. Other taxonomies don’t much care about reality. They would prefer to reflect how we humans think about reality. For example, Corbis has a large, hierarchical thesaurus that aims at helping customers find stock digital images. Such a thesaurus would happily list Arnold Schwarzenegger under “Germans” if that’s a mistake many customers make, even though he was born in Austria. 

We usually keep these two types of taxonomies clear in our heads. We have a utilitarian arguments about creating navigable taxonomies, and evidence-based arguments about taxonomies that try to reflect reality. Sometimes the two get mixed up, and a deeper examination of the topic would probably reveal that the line between them is not nearly as sharp as it seems. Nevertheless, we are comfortable with both.

Tags and the folkosonomies they beget do not fit perfectly into either box. As a reflection of how we - the masses - divvy up our world, they at best hint at knowledge. As a navigation technique, they are certainly useful, but that is not their greatest significance. Taken as one of the many new ways we’re connecting like with like, tags represent a phenomenon that swallows up knowledge and navigation. We may tag in order to re-find online resources, and the resulting folksonomies may enable others to find those same resources (to borrow Thomas Vander Wal’s way of putting it), but tags are part of a new infrastructure that includes every way we’re putting pages things together. This includes every hyperlink in every page and blog, every playlist of songs or textbooks, every triple serving the Semantic Web, and every set of publicly declared friends on social networking sites. Each of those connections indicates a relationship, a meaning, worth someone’s time to note. 

Put all these connections together and we have something we simply haven’t had before: a global infrastructure of meaning, consisting of, a set of relationships  - mainly implicit and potential - by which we can follow and contextualize ideas of every sort. Within this infrastructure, knowledge will grow, but it will also be understood. Within this infrastructure, we will create new means of navigation, guided by not just by recall, precision and relevance, but also by significance and interestingness. This network of relationships is a new resource, one we are making for one another, sometimes on purpose and sometimes through the implicit residue of our explicit acts. Building this infrastructure of meaning is a generational task. Learning how to navigate it and how to learn from it is a job especially for the experts in knowledge, knowing, teaching, and finding. The first challenge is to recognize that the messiness of our network of networks is not a failure of knowledge, but a triumph of meaning.