Cliff Lynch always gives highly insightful comments on technology issues. In remarks from the recent Educause conference, Lynch comments about folksonomies. You can listen to the podcast yourself in which he also touches on a number of other topics, but I’m going to paraphrase his words on folksonomy:

Lynch is puzzled about the entire debate over folksonomies, which doesn’t look like anything new to him. It’s natural for groups of people to develop their own linguistic practices to describe things for retrieval, which is why people do it. There shouldn’t be any reason that folksonomies and traditional descriptive techniques cannot co-exist. Folksonomies are problematic across very large groups. Also, over time there this is the danger that language will drift, which is why a standardized vocabulary is needed in some instances. He provides the example of medical literature, which is highly complex and wouldn’t be well suited to an ad hoc folksonomy. Most importantly, Lynch points out that when we’re able to “combine full-text access with any kind of human added vocabulary then we can do computational statistics so that we can move between the two in order to give a navigational ability that enriches both.”

So, full-text retrieval, highly structured classification, and folksonomy are all useful tools and we need to leverage the tools together. Professional bias keeps people from seeing the common perspective. Lynch believes that we are now only touching the potential of statistical correlation among retrieval tools.