This is why I always said I played D&D -- not AD&D (TM), not even D&D (TM). The published rules were a starting point, to be modified -- and more importantly, to be ignored -- at the gamemaster's whim in the interests of making the joint storytelling exercise as creative and entertaining as possible. Players need to know enough to operate within the world intutively, and that includes having at least some sense of "oh, THAT was a good one" or "flubbed it completely" -- which is really what critical hit and fumble map to.
Given that this is heroic storytelling, it's not unreasonable to recognize those exceptional events by giving them dramatic descriptions/effects. Sometimes you fall on your face and the opponent gets to beat on you while you recover, sometimes Golfimbul's head goes flying across the battlefield and down a rabbit hole.
However, there is one specific situation in which the rules have to be more than a suggestion: Tournament play. When there are multiple groups of players who will simultaneously encounter the same challenges under the control of different gamemasters -- and will be judged on how they responded to those challenges -- it's important that the challenges, and their capabilities, be calibrated so each group has roughly equivalent experience modulo their own skills. That requires that the gamemasters agree on exactly how the world behaves, and that the players be able to predict the world's behavior even if they've never played with that gamemaster -- or each other -- before. In that one very specific case, sticking to the published rules makes sense.
(In fact, tournament module design is rather different from normal module design for that reason -- the players need to feel that they're in control while actually facing calibrated problems that give everyone in the party a chance to shine. Ideally, that should be true even if they compare notes later; real choices may be offered, with enough difference to say "oh, I wish we had/hadn't gone that way", but those choices need to not interfere with judging. Not easy to do well.)
I think you're being overly literal about the word "hit". In D&D, that's taken to mean "injure". Armor certainly makes you less able to dodge, but also makes you harder to injure; something that clongs off the armor with (at most) mild bruising isn't considered a hit for game purposes.
But, yes, D&D is biased toward the sword-and-sorcery genre, which generally is idealized/simplified. And that's fine, as long as everyone agrees on the expectations.
You CAN -- and some have -- work out a pencil-and-dice system which fully captures all the feints and parries and shifting of position and stance and ground conditions and visibility limitations and fatigue and sweat getting in one's eyes and that damned leather strap under the left armpit coming loose again. But realistically, most D&D players neither need nor want that kind of accuracy. They want the engagement to play out in a reasonable amount of time, with both their decisions and their dice rolls having reasonable influence, to achieve reasonable results... where "reasonable" generally means "what we'd expect in a book", not "what we'd expect on a 15th-century battlefield".
There are folks who want more realism, and that's fine too. But every roleplaying system has its strengths and its weaknesses, and needs to be judged by what the players of that system want it to be. D&D isn't inferior to Runequest, or superior, just different.
Which gets back to my statement about not playing exactly the published version; most gamemasters I've known liberally borrowed from every system they'd been exposed to, and then promptly ignored it all when it got in the way of telling a good story.