In beginning of 99 when I began to populate the psyberspace web engine I only coded in support for 2 levels of categories (bad dbdesign ). This has had an unintended advantage of limiting the complexity of my ontological evolution. Thinking about this I decided to investigate how my category assignment had evolved over time. Without using lexical similarity measures to perhaps look at changes in categorical decisions, I was able to use SQL aggregation and some middle tier logic ( and a few single pixel gifs) to study the temporal course of my category assignment.