Originally Posted by jim2100
Like ajhieb, you are confusing certainty and probability. As I already explained a few posts back. But I am not going to give a probability and statistics lesson here, so you will have to correct your misunderstandings in that area yourself.
So if you don't intend on explaining yourself, do you expect to be taken seriously when you just keep repeating "Wrong again."
Had Backblaze made an effort to get a representative sample of the drives available from the manufacturers, then you could say with a fair amount of confidence, that Manufacturer A is better than Manufacturer B. But they didn't. They didn't do a random sampling. They didn't do a representative sampling. They bought a bunch of cheap drives. Period. That isn't a good sampling to base claims about the brand, no matter how many of those drives they had.
If you could show some sort of correlation between the models represented, and the ones not
represented, then you certainly could
state that based on the results that all the models from Man A are generally more reliable than all the models from Man B. But no one, not Backblaze, not you, not anyone has made that correlation. The untested drives aren't connected statistically in any way to the ones tested without that correlation. This has nothing to do with absolutes, other than there is absolutely no connection between the drives studied and the drives not included.
It's not a matter of Brand A is more reliable than Brand B with x% of certainty. I'm well aware of what certainty is in that context. The thing that you keep ignoring is that the study wasn't actually looking at brands, it was looking at models. Brand is wholly irrelevant until you show some statistical connection . And you haven't. In the context of the study, "brand" is a wholly abstract concept. You might as well have named some of the drives "Jack" and the others "Jill" and based the study on that. Then we could determine that people named Jack are more reliable than people named Jill. It sounds silly, but it's no more ridiculous than the leap you're trying to make, without first showing the connection between what was studied and what wasn't.
You can keep telling me I'm wrong until you're blue in the face, but until you bring something to the discussion besides that, you may as well be going "blah blah blah"