I believe in practice and drill, not creating problems that aren't there.
Thats great if your problem or emergency is one you have practised for or drilled for, but what happens when it is not? My view is that I would much rather create some IA's (immediate actions) based on statistical probability to start solving the problem and then develop the diver's ability to think by "oppotunistically facilitated harassment" (good one Dave, I am going to steal that one) to develop a thinking diver who will insert their brain into the cause and effect loop. Drills are great as long as they are part of this process. It is when they are done for the act of doing them with no integration with reality that they become detrimental and encourage rote responses to dynamic problems.