Page 46 -

40 B. Edmonds

4.1 Introduction

A common view of modelling is that one builds a ‘lifelike’ reflection of some sys-
tem, which then can be relied upon to act like that system. This is a correspondence
view of modelling where the details in the model correspond in a one-one manner
with those in the modelling target—as if the model were some kind of ‘picture’ of
what it models. However, this view can be misleading since models always differ
from what they model, so that they will capture some aspects of the target system but
not others. With complex phenomena, especially social phenomena, it is inevitable
that any model is, at best, a very partial picture of what it represents—in fact I
suggest that this picture analogy is so unhelpful that it might be best to abandon it
altogether as more misleading than helpful. 1
Rather, here I will suggest a more pragmatic approach, where models are viewed
as tools designed and useful for specific purposes. Although a model designed for
one purpose may turn out to be OK for another, it is more productive to use a tool
designed for the job in hand. One may be able to use a kitchen knife for shaping
wood, but it is much better to use a chisel. In particular, I argue that even when
a model (or model component) turns out to be useful for more than one purpose,
it needs to be justified and judged with respect to each of the claimed purposes
separately (and it will probably require recoding). To extend the previous analogy,
a tool with the blade of a chisel but the handle of a kitchen knife may satisfy some
of the criteria for a tool to carve wood and some of the criteria for a tool to carve
cooked meat but fail at both. If one did come up with a new tool that is good at both,
this would be because it could be justified for each purpose separately.
2
In his paper ‘Why Model?’, Epstein (2008) lists 17 different reasons for making
a model: from the abstract, ‘discover new questions’, to the practical ‘educate
the general public’. This illustrates both the usefulness of modelling but also the
potential for confusion. As Epstein points out, the power of modelling comes from
making an informal set of ideas formal. That is, they are made precise using
unambiguous code or mathematical symbols. This lack of ambiguity has huge
benefits for the process of science, since it allows researchers to share, critique
and improve models without transmission errors (Edmonds 2010). However, in
many papers on modelling, the purpose that its model was developed for or, more
critically, the purpose under which it is being presented is often left implicit or
confused. Maybe this is due to the prevalence of the ‘correspondence picture’ of
modelling discussed above, maybe the authors conceive of their creations being
useful in many different ways, or maybe they simply developed the model without a
specific purpose in mind. However, regardless of the reason, the consequence is that
readers do not know how to judge the model when presented. This has the result that
models might avoid proper judgement—demonstrating partial success in different
ways with respect to a number of purposes, but not adequacy against any.

1
With the exception of the purpose of description where a model is intended to reflect what is
observed
2 He discusses ‘prediction’ and then lists 16 other reasons to model.