Crazy Dice, Chance, and Counting Causes

Abstract We develop a simple model of the underdetermination of theory by data using “crazy dice” – standardly shaped dice with non-standard labellings – and we use the model to illustrate a number of distinct ways that scientific data (the probability distributions of the dice) can underdetermine our theories of that data (claims about the shapes, labellings, number, and identity of the dice). We conclude by drawing some tentative conclusions regarding what this model has to teach us about the underdetermination of theory by data in general.