Static metrics cannot as easily nor accurately address software reachability as can dynamic metrics. Although collecting data about information loss is useful for assertion placement, reachability analysis is still necessary to determine whether the assertion will be exercised.
Let 
represent the fault detectability of program P when
tested with test suite D. And let 
represent
test suite D after it has been augmented with enough test cases such
that all statements in P are exercised.
Note that 
could be empty,
in which case 
. Since reachability is the
first event in the fault/failure model, it is a necessary condition
for faults to be detected, hence
(This is similar to Weyuker's Monotonicity Axiom [14].)