BTW, it's horrific how poor some of the Seagate failure rates are according to Backblaze.. The 3TB ST3000DM001 in particular, with a 26.65% annual failure rate... yikes.
It's been a while, but if I remember right, even they explained it was a statistical thing, not really an indication of bad drives.
If I recall, the WD's either failed right away, or lasted several years before drastic drops in reliability.
Seagates could fail at any time, but rarely failed at first. Over the course of 4-5 years it balanced out. If the rates really had been that low, they would have ditched them, and they haven't. Also, many of those drives were made during or right after the Thai floods. Which was a bad time for hard drives.
Personally, me and my customers have not had good luck with normal WD drives, I replace them a lot, and rarely see old ones working that don't whine/howl or have errors.. Raptors and Blacks on the other hand, are fantastic drives that I will take over any other spinners. However, I would much rather have a nice SSD.
Also, any drive health indicator that actually scans the drive, is a bad idea. If the drive is failing, that will make it worse. Best way I have found, and is based on what Backblaze found. look at the SMART info, look for corrected blocks and uncorrected blocks. The best tool I have found for this has actually been Defraggler. The problem is, as Backblaze found, each manufacturer uses a different way of writing error codes (which is BS! as you will see). Defraggler will not only show you what the drive is saying, but also show you the raw data. Several times I have a SMART drive program read the data and say everything was fine, then go look at the raw data and see that something has clearly been written there, just that the program couldn't read it because the manufacturer used a proprietary system to indicate it, this makes the whole SMART system pointless if the motherboard can't tell when there really is a problem starting. Basically, either read the raw data, or you will need the manufacturers program to get a good reading.
Since I have been doing it this way, I've had a much easier time diagnosing troubled systems and managed to catch quite a few that were getting ready to fail.