In a different thread here, someone made an apt allusion to a
high-performance driving course. Anyone may wish to take such a course in
the belief it will make them a more complete or more well-rounded driver,
and this uncompromising attitude toward one's self is deserving of
respect. It will not, however, make them less likely to have an accident
- probably the inverse. The secretary who gets nervous when she has to
parallel park is far less likely to have a serious accident. And if large
numbers of people started subscribing to such courses, it is likely the
roads would become more dangerous, because in *some* people this would
favor an overconfident attitude.

Compare this to single-engine recovery in light twins. All pilots have to
demonstrate proficiency when they know an engine failure is likely to
occur, and they have been practicing for it. In real-world incidents
however, we know that a large majority fail to apply their training
successfully in light twins, with marginal excess horsepower. This still
leaves some who do succeed, which makes the training pertinent - combine
this with the fact that avoidance is not practical (despite the best
measures, engine-out incidents will still occur without warning) and the
training clearly becomes indispensable.

What about partial-panel in IMC? I share your impatience with marginal
instructors when it comes to those who instruct IR, yet who are afraid in
IMC. They should be like a fish in water in hard IMC, even partial panel.
Here again, avoidance is not a satisfactory answer, as vacuum failures
will occur without warning, and only partial-panel training will bail you
out if you can't see out the window when it happens. Let's see though what
happens to IR training requirements as the vacuum systems become obsolete,
and the classic AI/DG failure becomes an anachronism.