"You can't fly this plane
partial panel, so just couple up the autopilot to the GPS and have it
fly the approach" - which is, no ****, what glass-panel Cirrus pilots
are told.
In that case it's no wonder the insurance rates are so high.

In one sense, no. I don't believe there is a single accident
attributed to lack of partial panel proficiency in a Cirrus, with the
possible exception of the guy who launched into 800 ft ceilings on the
first flight after major panel work and wound up pulling the chute.
The airplane is quite reliable. People are not crashing because they
are depending on systems that fail.

But in another sense you are right. People are crasing because their
skill level is not up to the airplane, and this basic problem is not
being addressed. Limited panel flying has intrinsic value over and
above coping with the particular failure being simulated - it forces
the pilot to become sharper, to get more out of the instruments, and to
become smoother. It improves all aspects of his flying.

As long as all the automation works, the Skyhawk IFR pilot can be a
Cirrus IFR pilot with his existing skill set
This is no different than telling a VFR pilot to set the auto pilot
and let it fly if he/she runs into bad weather, or poor visibility.

Let's say it's a difference in degree only, and not in kind. On my
first flight in the glass panel Cirrus I asked the owner why there was
no CDI other than in the PFD. He didn't see my point. I explained
that if the PFD went out, the only approach we could shoot would be a
GPS or GPS overlay (using the Garmin display) since there were 2
GNS-430's but no external CDI for either. I felt this was acceptable
(what are the odds of PFD failure in conditions where GPS approach
conditions are not in range?) but suboptimal. He then explained that
the factory recommends not shooting a manual approach with a failed PFD
at all - just couple up the autopilot and let it do the job. This
despite the fact that altimeter, ASI, compass, and AI are all
available.

But at least part of the problem must be laid squarely at the feet of
the people doing the teaching and testing. This pilot took his IFR
checkride in his Cirrus, and the DE insisted he do a manual LOC
approach with the PFD off. Of course the GPS is NOT as accurate as a
LOC close in, but the DE didn't want to hear it. Thus I am reluctant
to blame the peope who are not being properly trained - what chance do
they have if even the DE's have no clue?

You have not created a better pilot, you have given him/her a crutch
to make up for lack of skill which is a very poor teaching method and
dangerous practice.

Like I said,
Personally, I think that's a ****-poor way to do things.

The real hazard, though, is not that the system that the pilot is
depending on will fail. These are fairly new airplanes, and those
systems are reliable. They're not failing a lot. The real problem is
that the system only does what it's built to do. Training makes a
pilot better overall. Substitute systems for training, and you better
hope you have systems to do EVERYTHING the pilot does, because without
the training, you will have an inferior pilot.

Michael