Models of clutter

In his seminal ‘Models of Data,’ Patrick Suppes [1962] proposes a ‘hierarchy of models’ to define a correspondence between abstract theories and the complex activities of conducting experiment and measurement. Although he nicely distinguishes a lowest level of ‘ceteris paribus conditions’, that is, a level of ‘noises, lighting, odors, phases of the moon,’ he does not provide a model for this level, and therefore is not able to connect this level to the upper levels. The level of ceteris paribus conditions aims at reducing clutter: to mute loud noises, to fresh the air from bad “odors”, or to re-organize the schedule for observations. These attempts to reduce clutter, that is, these cleaning activities are often the most time-consuming activities in scientific practice and require a lot of creativity and intuition. Because philosophy of science is, in my view, philosophy of science in practice, these activities deserve more attention. This article, therefore, proposes an attempt to complete Suppes’s hierarchy of models by suggesting a methodology for designing and testing ‘models of clutter’ that account for the level of ceteris paribus conditions.