First, you're not blind - you can test your product to see what the "plenty" is.
Second, many components have rated use, so it's easy to estimate mean time to fail and pick the one beyond the warranty period with whatever buffer you like. It's not like you need seconds level of precision here!
The level of consideration matches the level of argumentation, e.g., it's obvious you failed in your interpretative nitpicking on the word "mean" and think "reading about the normal distribution" means anything in this context.
Please design a physical product to reliably fail after a specific and precise amount of time (not usage, because that’s easier and not what you’re arguing), then come back and describe how easily you accomplished that feat. Everyone reading this thread who has worked in device design knows that your assertions are completely and utterly misguided.
Right after you explain how in this imaginary world of 0 knowledge where you're not even capable of translating usage into time companies set a warranty to 3 years (>legal min) instead of 13; and why there are warranty limitations for heavy use.
(and no, you don't need "reliably ... specific and precise", those are just artifical constraints you've added)
And don't speak for everyone, not everyone is so clueless re. business decisions just because they've designed some hardware.
First, you're not blind - you can test your product to see what the "plenty" is.
Second, many components have rated use, so it's easy to estimate mean time to fail and pick the one beyond the warranty period with whatever buffer you like. It's not like you need seconds level of precision here!