We consider that one area which is often disregarded in the field of
reuse research are those activities which are concerned with the second
quadrant of the reuse spiral; mainly that of the evaluation of the
risks associated with a development. We have indicated how important
this is for successful software development and therefore we will
illustrate this process, of reusing risk evaluations, with the aid of
an example derived from a case study of the National Health Service.
In the second quadrant we perform the following activities:
- We use information from the similar projects which were located
in the first quadrant and investigate what options are available for
developing a system, and secondly what risk assessments activities were
associated which each option. From this information we are able to
indicate some of the risk criteria which are associated with our new
development. In terms of the health service, typical risk criteria for
developing a system which informs doctors which drugs are functionally
equivalent may be: the doctors reaction to the system; the cost of
development; the accuracy of the information it generates; or the
feasibility of building the system. Typical options may be to: develop
a database application, an expert system or perhaps nothing at all i.e.
retain a manual system.
- We can now use sensitivity analysis to allow us to evaluate which
is the best option to take for our proposed development. In order to do
this then we evaluate risk criteria individually to obtain the utility
for each risk criteria. The risk criteria for each option may be the
same but their utility will be different. (Here we use utility in a
technical sense, a concept unifying benefit and risk.) In the case of
our drug substitution program the utility for cost will depend on the
expected cost of the individual options and the utility for doctors
acceptance of the system may for instance be greater for the database
option than the expert system since doctors may feel an expert system
is de-skilling their job. Weightings are then placed on the individual
criteria (again information gained from previous projects) in order
that the interaction effects between the individual risk criteria can
be evaluated.
- By folding back the decision trees the utility for each option
can then be considered and the option with the highest utility, or
greatest expected payoff, can be selected.