Abstract

It is an underappreciated fact within the philosophical literature on laws of nature that many scientific laws require the aid of a supporting cast of additional modelling ingredients (such as boundary conditions, material parameters, interfacial stipulations, rigidity constraints, and so on) in order to perform their traditional role in scientific inquiry. In this paper, I suggest that this underappreciated fact spells trouble for some recent reformulations of David Lewis's Best Systems Account (BSA) of laws of nature. Under the auspices of 'pragmatic Humeanism,' several philosophers have recently argued that the criteria of strength and simplicity that lay at the heart of Lewis's original formulation should be replaced with alternatives that are more sensitive to the role that laws play in scientific practice. Although the criteria that these philosophers put forward differ in a variety of ways, they are primarily concerned with the ability of laws to furnish us with predictions and encode information. This, I suggest, is a problem. If it is true that many scientific laws do not on their own perform some of the roles with which they are traditionally associated, then they are unlikely in isolation to make meaningful contributions to the predictive strength of a system or encode information about particular systems. Such laws are thus unlikely to end up in the best system, and so these accounts will have trouble conferring lawhood upon them.