As for the 100% redundancy claim, we both know that is rubbish, particularly if used as a portable device (though getting a bit large to be portable)
It's not that large J, and while we both know that 100% isn't possible in real life no matter what,......what do you think the chances are of 4 independant hard drives failing all at once?
Seen 2 simultaneously fail in a 5 disk Raid 5 when a 3rd disk was being replaced. They don't believe it was electrical issue during hot withdrawal, but flex in the drive bay as the old drive was stiff.
Believe it or not, the server stayed up, and due to Netware caching, we copied off all virtually all of the data

.
Not had too much luck with hot plug - I once changed a 500Mb disk in a Proliant, and HOT plug was apt - bloody thing started smoking, and the NetWare server promptly abended. Couldn't recover diddly squat off that one

Went through a patch in the early part of millenium when temps were more critical that any glitch on environment control would wipe out 30 odd disks a day, sometimes on same array.
So, yes, it does happen

. But I agree, very rare

But in Martin Imber's case, I was under impression he wanted a long term archive, and in that case, I simply wouldn't use disks - if you leave on, there is significant risk of corruption, if you switch off, you run high risk of drive sticking.
As an aside, it is most definately possible for 2 disks in a 2 disk mirror to break
