Ok, I will assume you don't know this so I'll cover the basics. For example, when you see those humongous MTBF numbers, you do realize that the HD is not actually going to last 80 years. But they do give information to the bean counters so they can set aside certain financial allocations for "warranty repairs". However, you may be surprised to hear that for the most part, they don't do "real tests" as the basis of these calculations. They base the predicted performance of the new drive based upon recorded performance of the old.
To borrow on your analogy approach. If I have a 2008 Model Chrysler Whoopdedoo 400 sports car, I may note that it has the same engine, the same driver train and the same, same list of many other components. The differences between the 2007 and 2008 model might be some body styling changes, different rims, interior styling. So should one not expect the 2008 to have similar wear, failure, oil change intervals whatever as the previous model ? If one car manufacturer puts out model after model after model that has a high incidence of failure, and then the new model is made from the exact same parts, are we really to expect any different performance ?
The 7200.9, the 7200.10 and the 7200.11 from Seagate for example indicate that by their very name, that the .11 is the "next version" of the 10. The only thing that might typically change between a typical "version 10" and a "version 11" would be for example a) cache size, or b) areal density. If a drive with 3 platters and the same bearings and the same spindle and the same, same, same , same other stuff gets a boost from 16 to 32 MB, will the reliability performance change ?
Now sometimes they do change something as they apparently did between the WD360/740 Raptors and the newer WD1500 (see below)
Statistics are useful only in large numbers. One can't assume say on the basis of 30 drives what you might expect out of 100,000. And there are certainly anomalies. In fact they recently upped the numbers req'd to be "reported" and when there is not a big enough statistical sample they use the drive's most immediate predecessor to provide some indicator.
Is there one model they list as "more reliable than 90%"?
Seagate Cheetah 5k3 was more reliable than 100%.
The 2003 versions of the Raptor scored a 86 % and 94%, it was the 2006 version that tanked at 12%
WD scored well in the following:
Model / Year / Rating
WD1200JB / 2001 / 67
WD1200BB / 2001 / 69
WD1200JB / 2001 / 82
WD205bx / 2000 / 95
But look at what's happened since 2005.....
WD2500ks / 2005 / 64
WD3200JX / 2005 / 58
WD1500 / 2006 / 12
WD4000KD / 2005 / 5
RE2 400 / 2005 / 25
RE2 500 / 2006 / 4
One of the things SR has done in the last few years is change the size of the req'd database size so they can produce a qualified result. As a result you have to dig into the data to get a result many times for newer drives when there is no ranking. SO even if they rely on "past performance" to get an idea of ranking, you can still go in and get the raw data.
The Seagate ES2 has 2/48 failures or 4.2%
Failed before 1 month
Failed before 1 month
The WD7500AAKS has 7/61 or 11.5%
Failed before 1 month
Failed before 1 month
Failed before 1 month
Failed before 3 months
Failed before 3 months
Failed before 3 months
Failed before 3 months
Edit: OK, I can't resist. So, would these guys judge George W. Bush based on Bill Clinton's work too???
Only if they were completely uneducated and uninformed....(in other words, just like Bush). But they're not. They are not judging a $60 bargain basement 5400 rpm drive based upon the company's $600 15k rpm SCSI drive. They are predicting the performance of version 10 drive based upon the version 9 drive from th exact same company ..... ya know the one with the exact same casing, exact same circuit board, exact same connectors, exact same bearings, exact same spindle, exact same head movement, exact same number of platters ....
exactly like the manufacturers themselves are doing when they publish MTBF's for the new drive based upon test results for the previous version..
I don't think anyone would put Clinton and Bush were in the same product family .... Let's just say on the trip to the white house Dubya "rode the short bus" ..... but if Cheney were to succeed Bush, I think most would expect more of the same.