> The alternative point of view has been put forward out of the
> AI and DAI tradition.
> The argument there is that the simulation is the theory.
In my opinion, the two types of theorizing have a different status.
Discursive theories construct geometrical metaphors (in language) that
enable us to communicate about "a picture" or "a trajectory". They
require a perspective or a set of assumptions in the case of the social
sciences (because of the otherwise overwhelming complexity; cf. Weber).
In the simulation one abstracts from the specific content of the theory
and tries to formalize it into an equation or a set of equations. Each
equation in the simulation should represent a substantive theory.
However, the simulation "recombines" the various theories. It therefore
enables us to "weight" the perspectives, for example, as subroutines.
This provides us with a view on the phase space of possible events and
trajectories of the system(s) under study. This perspective, however, is
algorithmic, i.e., one has formalized the substantive variation (e.g.,
by declaring a random generator).
The consequent theory is a formal theory that has lost its direct access
to the substances that went into the simulation. The substantive
researcher may be able to use this feedback for improving his/her
theorizing (or not). In the AI tradition one can use the formal theory
for exploring the virtual reality (and then perhaps base another
construction, e.g., a robot, on it). But the feedback to sociological
theorizing requires the loop to substantive theorizing. Both objectives,
of course, are perfectly legitimate, but they aim in different
directions (that is, at different "realities").
As was noted in previous contributions to this thread, the elaboration
of the substantive theory by formalization for the purpose of the
simulation makes it first clear where "gaps" are left, and secondly, it
enlightens us where inconsistencies between different elaborations
emerge. The substantive theories are expected to compete with one
another in explaining the phenomena, but potentially along different
axes definable in the simulation.
For example, neo-classical economics is interested in market equilibria
at each moment in time. This dynamic perspective can be elaborated into
comparative statics. The perspective of evolutionary economics, however,
is interested along the time axis in how markets can be upset by
innovation (e.g., Schumpeter). The two theories compete for the
explanation, but along (nearly) orthogonal axis. The simulation enables
us to recombine the two almost incommensurable perspectives when
studying how technological innovations upset market equilibria. This may
lead to or resonate with a new field of substantive studies that study
innovations as interaction effects between market forces and knowledge
With kind regards,