Folksonomies and Relationships

Another interesting comment on my Folksonomies in the Enterprise post came from Sebastian who said about folksonomies versus taxonomies:

"Yes, they are similar, but not same. A taxonomy usually contains formal relationships between the elements of the taxonomy. You don't have such explicit relations in a tag cloud."

This reminded me of an article I read a while back on taxonomies in zoology.  For hundreds of years zoologists have been fitting species into a well defined taxonomy.  If it has a hooked nose, partially webbed feet, and big floppy ears, then it must be related to species X.  Taxonomy at its greatest, right?

The article I was reading was about zoologists that are now using DNA for determining the taxonomy classification.  Lo and behold, what they found was that the old style taxonomy had many errors.  It had relationships between animals that weren't actually related.  And, interestingly, there are camps of zoologists that don't want to change to an accurate (DNA-based) taxonomy.

The problem with taxonomies is that many relationships are in the eyes of the beholder (the people that created the taxonomy).  People strive to create order out of chaos - whether that order makes sense or not.  They assume relationships where none actually exists.  But one person's rationalization of chaos does not necessarily match another's.  And, when this happens, there will always be camps that view their rationalization is superior to any other. Luckily, I could never imaging an enterprise architect falling into this trap when defining their service taxonomy in UDDI, could you?

So, I guess my question is whether a taxonomy built with assumed relationships (which, by their nature have a significant degree of "false positives") is really any different from a folksonomy which tries not to impose assumed relationships?  Or is a typical taxonomy just a potentially inaccurate folksonomy (a folksonomy with the addition of unreliable relationships)?

Sebastian's comment was interesting in that he pointed to a project that tries to determine real relationships out of tag clouds.  This strikes me as very much like the use of DNA to determine real relationships among species.  It's also like my analogy from the previous post on Merriam-Webster, whose role is to document not to assume.  "Semwiki, meet Merriam.  Merriam, meet Semwiki.  I think you two will have a lot to talk about."