This lecture and supporting document is based on two talks given by Professor Clay Shirky in the spring of 2005 -- one at the O'Reilly ETech conference in March, titled "Ontology Is Overrated", and another at the IMCExpo in April, titled "Folksonomies & Tags: The rise of user-developed classification." The written version is a heavily edited concatenation of those two presentations.
Shirky discusses categorization, aiming to persuade the audience that much of the conventional wisdom surrounding categorization is incorrect. Specifically, he argues that many of the methods currently being applied to categorization in the digital realm are ill-suited, largely because they carry over outdated mindsets from previous approaches.
Furthermore, he demonstrates that what is observed on the Web represents a significant departure from past categorization strategies rather than a continuation. The latter part of his discussion ventures into more speculative territory, acknowledging that it is common for old systems to break down before new replacements are fully recognized. This phenomenon is observable in the music industry today, serving as a prime example of the ongoing shift in categorization.
Shirky believes that what will emerge are far more natural methods of organizing information than what existing categorization schemes permit. This new approach is predicated on two elements: the link, which can direct to any piece of content, and the tag, which serves as a means of attaching labels to links. The approach of tagging -- applying labels freely without the limitations of predefined categories -- might appear chaotic, yet as demonstrated by the Web, it is possible to derive considerable value from large, unstructured datasets.