Limitations of Models

Can lead to oversimplifications.

“There is no denying that much of the work in designing communication models

illustrates the oft-repeated charge that anything in human affairs which can be
modeled is by definition too superficial to be given serious consideration.”

Some, like Duhem’s (1954), believe there is no value in models at all:

We can guard against the risks of oversimplification by recognizing the

fundamental distinction between simplification and oversimplification.
By definition, and of necessity, models simplify. So do all comparisons.
As Kaplan (1964) noted, “Science always simplifies; its aim is not to
reproduce the reality in all its complexity, but only to formulate what is
essential for understanding, prediction, or control. That a model is
simpler than the subject-matter being inquired into is as much a virtue as
a fault, and is, in any case, inevitable [p. 280].” So the real question is
what gets simplified. Insofar as a model ignores crucial variables and
recurrent relationships, it is open to the charge of oversimplification. If
the essential attributes or particulars of the event are included, the model
is to be credited with the virtue of parsimony, which insists-where
everything is equal-that the simplest of two interpretations is superior.
Simplification, after all, is inherent in the act of abstracting. For
example, an ordinary orange has a vast number of potential attributes; it
is necessary to consider only a few when one decides to eat an orange, but
many more must be taken into account when one wants to capture the
essence of an orange in a prize-winning photograph. abstracting. For
example, an ordinary orange has a vast number of potential attributes; it
is necessary to consider only a few when one decides to eat an orange, but
many more must be taken into account when one wants to capture the
essence of an orange in a prize-winning photograph.

Models can miss important points of comparison. Chapanis (1961), “A model

can tolerate a considerable amount of slop [p. 118].”

Can lead of a confusion of the model between the behavior it portrays

Mortensen: “Critics also charge that models are readily confused with reality.
The problem typically begins with an initial exploration of some unknown
territory. . . .Then the model begins to function as a substitute for the event: in
short, the map is taken literally. And what is worse, another form of ambiguity is
substituted for the uncertainty the map was designed to minimize. What has
 happened is a sophisticated version of the general semanticist’s admonition that
“the map is not the territory.” Spain is not pink because it appears that way on
the map, and Minnesota is not up because it is located near the top of a United
States map.

“The proper antidote lies in acquiring skill in the art of map reading.”

Premature Closure

The model designer may escape the risks of oversimplification and map reading
and still fall prey to dangers inherent in abstraction. To press for closure is to
strive for a sense of completion in a system.

Kaplan (1964):

The danger is that the model limits our awareness of unexplored

possibilities of conceptualization. We tinker with the model when we
might be better occupied with the subject-matter itself. In many areas of
human behavior, our knowledge is on the level of folk wisdom ...
incorporating it in a model does not automatically give such knowledge
scientific status. The majority of our ideas is usually a matter of slow
growth, which cannot be forced.... Closure is premature if it lays down
the lines for our thinking to follow when we do not know enough to say
even whether one direction or another is the more promising. Building a
model, in short, may crystallize our thoughts at a stage when they are
better left in solution, to allow new compounds to precipitate [p. 279].

One can reduce the hazards only by recognizing that physical reality can be
represented in any number of ways.