Generally if I am going to be playing Dungeons and Dragons I am not very interested in a fiction where 20th level fighter is just a little bit better than a 1st level fighter at things like evading attacks, running and jumping and the like. I need a fighter who faces down pit fiends to feel like someone who would face down pit fiends.
I feel like we've done this a million times yet we've made absolutely no headway.
@Parmandur and
@Imaro
Imaro was heavily involved in my level 30 DC thread many years back when 5e was first released (which was unfortunately wiped out in the board changeover...it was an extremely productive thread).
Here are the issues at hand that make the comparison of 5e mechanics and fiction at end game incompatible with the experience of 4e mechanics and fiction at end game.
1) 5e uses some kind of extra-PC, objective baseline for the natural language assignment of the difficulty of tasks. Whatever that baseline is, its there, but we can fuss over what it is (which we did in that old thread). Is task x and y Easy/Hard for "average adventurer", "every-man", whom? Someone.
2) A large part of 5e's bounded accuracy is to keep challenges relevant for PCs for longer.
3) 5e's noncombat resolution mechanics produce a fiction whereby there is SIGNIFICANTLY more parity in feats of non-combat martial prowess (coordination/balance/explosiveness/endurance, etc) between "average adventurer" or "every-man" (whatever the baseline) and an endgame Fighter.
So in 5e, you're 4th level Fighter may have + 6 to Athletics. Meanwhile, your 20th level Fighter may have...+11?
4) Even if 5e's GMing ethos pushed back against the intentful design of (2) (it doesn't) and it included 4e's noncombat resolution ethos of "all on-screen obstacles are tier-relevant threats/dangers" (which, for the 4 millionth time, does not mean that the door to the guildmaster's vault in Heroic Tier is suddenly an Epic challenge at Epic Tier if the fiction of the guild hasn't changed), that would still mean that the 5e Fighter at endgame (who is taking on Ancient Red Wyrms in melee combat) isn't overwhelmingly more competent at feats of athletic prowess than the low level Fighter (who is taking on Red Dragon Wyrmlings in melee combat).
Say whatever you'd like about the anti-process-sim gonzo nature of 4e...but as a martial artist and athlete my whole life...that narrative...just does_not_compute. The parity in noncombat capability between the two is far, far, far too narrow.
The guy taking on the Ancient Red Wyrm better be borderline superheroic in their feats of athletic prowess. He almost surely needs to be north (possibly well north) of Captain America and Captain is well north of a 5e endgame Fighter!
Contrast with 4e where:
5) The guy fighting the Red Dragon Wyrmling (RDW) has somewhere around a +13 Athletics...while the guy fighting the Ancient Red Dragon (ARD) has around a +38 Athletics!
One guy is well south of Captain America. One guy is north of Capatain America.
6) Again, the ethos of 4e noncombat resolution is that "obstacles that are on-screen should be tier-relevant threats."
7) So that RDW guy is faced with climbing a steep mountain face with the most precarious of handholds in the cold winter (DC 22). Meanwhile, the ARD Fighter is climbing a sheer volcano, leaping from one impossibly distant handhold to another, while the actual Elder Spirit of the mountain is violently changing the shape of the face to try to shake the Fighter and assaulting him/her with scalding pyroclastic outflow from vents in the rock (DC 42).
The mechanics and ethos of 4e supports that fictional framing and that disparity (which supports the ability to face down an RDW vs an ARD). 5e's mechanics and ethos does not.
Which is fine. It what intentfully designed to produce that...because that is what 5e's vocal audience wanted.
But even if you want to genre drift play to mythic in 5e, the basic PC build tool and action resolution mechanics do not support that parity (the parity of, say, Jamie Lannister vs Beowulf)...having 25 % better chance to succeed at the overwhelming % of tasks just isn't anywhere near the requisite disparity.