I find the confirmation of the theory, and the actual response of vendor implementations, useful. Using empirical evidence based on a significant sample size achieves that. The alternative is to make the assumption that the theory and the vendor implementation is correct, and I only do this when i have no useful evidence to use for verification.
TFA doesn't mention drive vendors. Does "others" in the "fingering others" statement refer to other SMART metrics? I dont understand how recommending using some SMART metrics and not others is anything to be suspicious of. How could manipulation of which metrics to recommend help Blackblaze's business?