Information Technology Reference
In-Depth Information
and should only be considered in exceptional circumstances as the cost of the-
ory revival is prohibitive and the resulting theory may be different from that
originally conceived.
In addition to the theory view of a program life cycle, he also directed criticism
at the then significant emphasis of methods for program development. He claimed
that the methods based on sequences of actions of certain kinds cannot lead to
the development of a theory of the program because the intrinsic knowledge held
by a human has no inherent division into parts nor an inherent ordering. Instead
the person possessing the knowledge is able to present multiple viewpoints as
responses to requests. Where methods were supplemented with notations or
formalizations then these were treated as secondary items as the theory of a
program is intrinsic and cannot be expressed. Thus: “...there can be no right
method”.
Having outlined the basic hypotheses of Naur's paper, the remainder of this
account continues the critique of Naur's ideas and applies them to modeling and
Enterprise Architecture.
3 Programs as Models
Naur was concerned with programs, but Enterprise Architecture is concerned
with the production of models of interconnected systems or components. Thus
we need to explore the relationship between programs and models and use that
as a basis for analysing the applicability of Naur's hypothesis in the context of
Enterprise Architecture.
A major activity in software engineering and computer science in general is
modelling and as Fetzer [1] has noted “the role of models in computer science
appears to be even more pervasive than has been generally acknowledged..”. A
key feature of modelling is the existence of an isomorphic relationship between
the parts of the model and the parts of the thing modelled at some level of
abstraction. Smith [16] whilst noting these different types of models emphasizes
the nature and importance of “representation”:
“To build a model is to conceive of the world in a certain delimited
way... Computers have a special dependence on these models: you write
an explicit description of the model down inside the computer...”.
Smith suggests this feature distinguishes computers from other machines because
they run by manipulating representations. “Thus there is no computation with-
out representation” [16, p, 360] If we pursue this analysis further: From Naur
we can state that the program is a theory; from general computation principles,
we can state: the program is a model. This leads to the notion that there is
an equivalence between program = theory = model. We might moderate this
further by noting that a program is a representation of a slice of “the” theory. In
general though, this blurring between programs, theories and models is confus-
ing and inaccurate. While models may exhibit an isomorphic relationship with
their subject matter, this relationship may not reveal the theoretical connections
