Sullivan et al. proposed the application of net option value (NOV)
analysis to measure software modularity. The idea is that a module creates value in
the form of options: one has the right but not the obligation to replace a module with
a new/better version. The more likely a module is subject to change and the more
independent it is, the higher option value can be generated. Their analysis is based on
design structure matrix (DSM) models where both design and environmental conditions,
such as requirements, are uniformly modeled as design variables. Cai et al.
proposed modularity vector so that design evolution can be simulated and the impact
of changes can be predicted based on the variations of NOV values and other
measurements. As a simplified variation of NOV analysis, Sethi et al. proposed new modularity measurements, such as design volatility, concern scopes and independence
level, based on DSM models there both concerns and designs are uniformly
modeled. These measurements were used to quantitatively assess which programming
paradigm, AO, versus OO, is more stable under given changes and which one
can generate higher option values from design level.