What he was pointing out is that by paying attention just to the failures--which are much, much, more likely to be reported--we can lose sight of the sample size. If you hear from 10 people who tell you their Suunto malfunctioned you might not buy one. Because you don't know about the 999,990 people with flawlessly operating Suunto computers, for an admirable 99.999% reliability. So, yes, we do need to concern ourselves with the non-failures.
As a sample size, yes, hence my wetsuit analogy. If you're implying (maybe not) 3-nines reliability - I don't think so. Given just the feedback on this thread, you'd be lucky to see 85%/90%. One poster/shop owner just clamed 50% failure out of the box!