Backup gyros - which do you trust?

Richard Kaplan wrote:

A vacuum faliure with a gradual spooldown of the AI and DG at vary rates is
a subtle process, very different from having an instructor suddenly cover up
a gyro. You are correct that there is enough redundancy in a typical GA
plane to FLY partial panel; the problem is that there is not enough
information for most pilots to IDENTIFY a partial panel situation before the
situation has become critical.

Richard,

There seem to be a number of instances where the pilot was aware
of the problem -- had described it to ATC and possibly requested
some form of assistance -- had been flying the plane for some time
more or less under control, and then lost it. So the pilot did
identify a partial panel situation before it became critical,
he simply couldn't FLY partial panel.

The accident Julian posted the link for more-or-less seems to
fall into this case, as did a couple of local accidents.

Cheers,
Sydney