A case study in model management: the US Coast Guard, GAO, TEFA-2 and structured modeling with embedded languages

Abstract

We present something of a case or field study of model management. We begin by listing and discussing a series of commonly-voiced reasons not to use modeling and model management systems. We then summarize the findings from a GAO (Government Accounting Office) report on a proposal made by the US Coast Guard. The GAO report was highly critical of the analytic work behind the Coast Guard proposal and this led to its rejection. We then revisit the list of reasons not to use models and model management systems and we find them largely implausible. Following this, we describe the elements of a prototype model management system, TEFA-2, which is able to store information about models that is sufficiently rich to generate structured modeling reports as well as documented Mathematica files, which can be used to execute the models and obtain graphical output from them. TEFA-2 is a simple system (although it draws upon the sophisticated theory underlying structured modeling) and only a prototype. Yet, we believe, it and systems like it can go far to ameliorate the kinds of problems that provoked the GAO report.