Something I've noticed lately is a tendency to evaluate firearms based on so-called "torture tests". If a gun can survive an arbitrary test like a fall of a hundred feet it is somehow more "reliable" than a gun that can only fall 20 feet.
But what does it really prove? If, in the middle of a fight, I drop my gun over a 100 foot cliff will my attackers stop while I climb down to retrieve it then wait for me climb back up to resume the fight? Hello! Disarmed guy! Unless those thugs are in a class of stupid beyond all other classes, they will take advantage of my butter fingers to kick my clumsy ass over the cliff after it.
Similarly, a guy who calls himself "ZombieTactics" on Calguns.net linked to a video where some guy reportedly buried his gun for two years, dug it up and then proceeded to fire 500 rounds through it with only a few failures.
OK, admit it: A guy calling himself "Zombie" anything getting excited over stuff that's been buried for two years really is kind of funny.
Nevertheless, what does it prove? If my gun gets buried in a pile of dirt am I going to wait for two years to dig it up? Not likely unless I am dead in which case I won't care. So what is the context of this "test"? Does it prove that, after the Apocalypse, I happen to to find a Glock with 500 rounds of ammunition buried in someones backyard it will mostly work? The possibility exists I suppose but I don't see why I should even care.
Reliability only exists in a context. A gun is one component of a system and the suitability -- of which reliability is a part -- of any particular gun can only be evaluated relative to its function within that system and the environment it is expected to operate in.
One of the important factors that separates engineering and the sciences from wishful thinking and fantasy is a grounding in observable reality. The, often unstated, assumption is that reality exists and it is knowable. Context is not everything but without it, separating any bits of truth from You Tube fantasies becomes an order of magnitude harder.