On Roles of Models in Information Systems - Intentional Perspectives on Information Systems Engineering - page 21

3.1 Engineering Practice
Engineering projects follow a well tested and successful approach, which is the same
for all kinds of engineering. Work is done in phases, each phase addressing different
issues at different levels of detail. Each phase must produce well defined and well
documented results, which are subjected to evaluation prior to being accepted as
basis for further work. The first project phases are concerned with setting the objec-
tives of the projects, e.g., what is the final outcome of the project, what is the budget
(time, money, competences). That is, the first phases are concerned with develop-
ing the requirements to be satisfied by the project. The next phases are concerned
with developing engineering designs that satisfy the stated requirements. Each new
phase adds new detail to the design proposals from previous phases. The end-of-
phase evaluations concern technical issues, as well as project economy and plans
for the remaining work in the project.
Information systems engineering differs from classical engineering in the cost
profile of the projects. In classical engineering, e.g., the building of roads and
bridges, the lion's share of the costs are invoked during the last phase of the projects,
when the actual realization of the engineering design is done. During this last
phase many hands and machines are put to work, and components of consider-
able size and cost are assembled according to the plans developed in the previous
phases.
In information systems engineering the project is essentially over when the soft-
ware have been written and tested. There is no building phase which is comparable
to the building phase of classical engineering. The lion's share of information sys-
tems engineering and software engineering is associated with the requirements
development, the design, the programming and testing. These phases in classical
engineering carry costs that are small compared to the total project expenditures.
The costs of making mistakes during the design are huge in most classical engi-
neering projects because of the large investment of time and money in the last phase
of the construction work, e.g., a bridge that falls down is a calamity. Therefore all
project phases are subjected to strict control and management practices. These prac-
tices have been taken over in information systems engineering projects where the
costs are large of making mistakes during requirement development and system
design.
The classical engineering approach has been much criticized in the software engi-
neering community and is seen by many as being both old-fashioned and traditional.
The classical engineering approach is known by different terms in the software engi-
neering community, the most common terms are top-down design and the waterfall
method when the classical approach is used for information systems engineering.
Several other approaches have been proposed. A common feature among many of
the competing proposals is the early development of software prototypes, and a
gradual expansion and continuous testing of a prototype, until the prototype has
been developed into a final product.
The two major approaches to information systems engineering are the top-down
“requirements-first” approach and the bottom-up “agile” approach.