In the 1980s and early 90s, there was a lot of interest in work done at and around Stanford on applying concepts and methods of set theory and model theory to issues of natural language semantics, automated language processing, and human reasoning. Buzzwords from the time were CSLI (Stanford's Center for the Study of Language and Information), situation theory, situation semantics, PROSIT (Programming in Situation Theory, a programming language generalizing PROLOG), and non-well-founded set theory. While this work found applications in philosophy, linguistics, engineering, education, manufacturing, large-scale planning projects, and intelligence analysis, it never gained much traction within mathematics, the field that inspired it. (One exception was Aczel's development of non-well-founded set theory.) Its strength was the injection of some mathematical precision into complex human domains, where holistic problems for the most part defy the precision of mathematical models. In my own case, application of situation theory to industrial ethnography led to post-9/11 work for the CIA, which in turn led to further projects for the US Navy and the US Army, where the high stakes mean that even very narrow, and often brittle applications can be significant.