Discussions of transparency, audibility and what's "good enough" always fall apart because the gray area between clearly-audible and clearly-inaudible is vast. Within that area are elements that can only be heard under certain circumstances, only with high-quality speakers, only by those with better-than-average hearing, only when listening intently, or only heard by trained ears.
Because we can't always be sure that no one else can hear what we can't hear, our strategy is to maintain the highest accuracy possible, to carefully attend to details that "probably" no one will notice, to aim for a higher quality than we think we need. It's why we mix with 32-bit data, even though what happens down in those lowest bits is almost never audible. It's why we intently listen to and correct the tiniest details that probably don't matter to the big picture.
Because we reflexively aim for the highest possible quality without asking ourselves how important it really is, many of those efforts are unnecessary. But I liken it to consistently using your turn signals when driving...if you do it by sheer habit, even when yours is the only car on the road, that reflex may someday save your life.
OTOH, a common trap for audio people is to get stuck in trivia while remaining oblivious to bigger mistakes. it's like a drunk driver making sure his oil is clean. Or obsessing over what dither algorithm to use while obliterating dynamics through over-compression.