In the above arguments, twice we have applied observation as a replacement for description, first in determining the order of troubleshooting scripts, and then in determining whether a particular set of scripts is capable of being ordered in that manner. These are both `phenomenological' (or `empirical') approaches that replace abstract descriptions with hard evidence, by changing the problem to allow employing evidence instead of theory.
In traditional approaches to troubleshooting, theory guides all actions. In the context of this paper, the precedences between tasks are theoretical. Interpreting a valid theory literally improves speed of troubleshooting by avoiding redundant tests, while an invalid theory may blind one to a potential problem by keeping one from executing tests in the order required by physical conditions.