Information Technology Reference
In-Depth Information
Fig. 5.1
Simulation shares
features with
theory
(explicit,
articulated concepts and
rules),
experiment
(empirical
data and a range of outcomes)
and
demonstration
(rigorous
argument according to rules)
yet remains distinct from each
of them
THEORY
EXPERIMENT
SIMULATION
DEMONSTRATION
can be regarded as mediating models because experiments sometimes
are
models—
simplified, abstractions of more complex processes—and because experiments often
have the same cognitive
function
that models do in theorizing: they are integral to the
process of inventing, constructing, negotiating and validating beliefs about the world.
5.6
Static and Iterative Models
Our simulation aims to capture this mediating function of experiments (as sets of
devices and procedures
that produce
observational results
) in relation to theories (as
sets of
hypotheses
) and to scientific communities (as sets of
actors
). This requires a
model that can be iterated in order to draw out the consequences of the assumptions
it implements. A static model specifies the form and structure of its constituents, but
cannot show what their interaction would produce in time and in the world.
The key difference between a model and a simulation is that a simulation iter-
ates the states of a model, to produce behaviour over a period of time. Simulation
differs from mathematical process models in science, which use procedures such as
integration or differentiation to calculate end-states. These linear methods generate
outcomes that are determined by the starting parameters. By contrast, iteration en-
ables us to trace the implications of structural and other assumptions of the models
that make up the theory, by exploring the range of outcomes that occur for a given
set of starting parameters. In this respect simulations are similar to experiments and
to theories (Fig.
5.1
).
Because our simulation involves an adaptive belief system, the only way to dis-
cover, say, how a model of inference works out in a particular context is to run the
simulation many times in order to ascertain the range of behaviours and outcomes.
Variability of event-sequences and their outcomes represents unpredictability due to
the outworking of social and other contingencies. Simulation makes it possible to
evaluate such assumptions experimentally, not solely in relation to logical, seman-
tic or other criteria. Historical, ethnographic and cognitive studies provide empirical
benchmarks against which the behaviour of the simulation is evaluated. This sets sim-
ulations apart from models (as static structures of propositions arranged according to
syntactic rules) and from experiments (as material components arranged according
to physical, chemical or other constraints).
