In this paper we analyze the efficacy of basic path loss models at predicting median path loss in urban environments. We attempt to bound the practical error of these models and look at how they may hinder practical wireless applications, and in particular dynamic spectrum access networks. This analysis is made using a large set of measurements from production networks in two US cities. We are able to show quantitatively what many experienced radio engineers understand: these models perform poorly at predicting path loss in even relatively simple outdoor environments and are of little practical use aside from making crude estimates of coverage in the least demanding applications. As a solution, we advocate a renewed focus on measurement-based, adaptive path loss models built on appropriate statistical methods.