Hierarchical decomposition is often a useful tool for dealing with complexity,
but many of the theories and predictive models follow an extreme reductionist
approach, which may not always be valid.  In some situations, it is hard to
accept the low level of detail, the precise numbers that are sometimes
attached to subtasks, and the validity of simple summations of time periods.
Furthermore, models requiring numerous subjective judgments raise the question
of whether several analysts would come up with the same results.