I don't that I passed on the information with any raised eyebrows... I just found it interesting that Google and CM would report such similar studies (but somewhat dissimilar results) in such a short time frame. MTBF is an acronym for 'mean time between failures'. The 'mean' refers to an average of time to failure.
At work I have some servers that have bun running 24 x 7 for over 6 years. That's a bit over 52,500 hours. We're all afraid to hit the power button on these boxes until we have replacements ready to go into service.
I personally think that they could run for another 6 years no problems. But I'd also personally give them somewhere around Buckley's chance of spinning back up from a power loss.
At anyrate it is all a meaningless argument in most server environments as the OS / application / hardware platform will go end-of-life before most HDD's ever reach their MBTF values. I can count on one hand the times that we've had a catrosophic failure issue due to HDD's, and have four fingers left over.
HDD's are reliable, for the most part. Starting them and stopping them will shorten their life.
My 2p