It was interesting hearing Bran Selic talking about UML on se-radio. For me there were three interesting points that I remember from the interview. He covered the history of UML. He made the great observation that people often focus on the static parts of UML such as class diagrams, where as often the most useful diagrams are those that show the dynamic interactions of the objects. I remember from studying CORBA services in the past that the interfaces often told one very little until one looked at the interaction diagrams to see the way the methods were used. Third, he pointed out that work was being down on defining the semantics of UML. This should be an interesting read.