The “failing upwards” argument is kind of a joke. Key depletion exists to drag out our engagement, and without it, the real danger to Blizzard is that people would succeed upwards too quickly, finish their seasonal goals, and go on hiatus.
It’s a foundational element to gaming in general that the ability to repeat a challenge you just failed at provides more time to iterate on your group strategy and personal tactics. Virtually every game you play has this natural dynamic. WoW’s current approach severely interrupts the dynamic by saying, “oops, you failed, now go do a different dungeon at a lower level, and MAYBE when your key re-levels, you’ll get the same dungeon you just failed at and another chance to prove yourself.”
There is a legitimate argument for Blizzard’s key depletion approach, aside from the more obnoxious stretching out of player hours for Blizzard’s bottom line. The legitimate argument is that by repeatedly forcing players to engage with a variety of dungeons one level below their currently preferred difficulty, this provides a larger pool of groups in LFG for the mass of players working their way up.
So while the people at +14 find the +13 to be a chore that wastes a lot of time, the people at +12, needing a +13 to up their rating benefit from the larger selection of groups, and the opportunity to routinely group with slightly more advanced players. Not that I think it’s the best approach, or that the playerbase wouldn’t adjust to a better system designed to waste less time by letting people target the challenges they’re after. But the increase of the player pool in LFG is the real argument for it.
The argument that the danger is players “failing upwards”, on the other hand, is not a serious one. Additionally, if keys did not deplete, the risk to player time from running into someone that supposedly “failed upwards” would be even less - because upon seeing the poor performance, the group could boot the underperforming player and go again, without losing the key.
Again, the issue to Blizzard is close to the exact opposite of the “failing upwards” argument. Their concern is players succeeding upwards too fast, so they slow the rate at which players can hone their approach on the challenge they just failed by interrupting their attempts with the need to re-clear lower level dungeons.