I'm not having any luck making sense of the Mention function, so please forgive the no-web-savvy formatting here...
Blue, I have not yet had an opportunity to play any games in 5th edition, while my main experience of 4th edition was thinking (as a big fan of the original DragonLance setting, who'd always wanted a pre-Goldmoon game to be viable) that it was really wonderful that the medicine skill was finally allowed to patch up people who'd been injured by things other than caltrops (and yet I was as frustrated as anyone else by how the hardcover books were released pre-playtest, causing errata files to approach or even exceed the page counts of the books in question). I never enjoyed 3rd edition much, because it rapidly grew to include far too many extra books with extra content to be included at any given player's whim (making it more time spent on sourcebook-studying than on actually playing), although I did run Mike Mearls's "Iron Heroes" for a couple years, ages ago. So my main impression of 5th ed is that it's far more similar to 2nd ed than to either 3rd or 4th, which feels to me like a positive thing. But maybe I'm mistaken.
Fanaelialae (and Saelorn), as I understand it, the reason why Dexterity is an overpowered stat is because the basic structure of the stats was put together piecemeal during the original development by Gygax. And we wound up with "physical power" being divided into two stats for offensive and defensive applications, while "physical finesse" was left as a single stat which thus rapidly became twice as powerful as it should be, once ranged-combat-expertise and evasion became factors in the rules. Similarly, the Wisdom stat has no real definition at all, since it was originally envisioned as "all the brain stuff that isn't raw Intelligence", but then Charisma got thrown on so that telling honeyed lies and threatening to beat a man to death with a chair can be resolved using the same core stat. After which, Wisdom wasn't quite Willpower, or Perception, or Savvy anymore. It was just "the other stuff I guess". Which became a tradition unto itself...
If I were ever to take the time to build a "retro-clone" game from scatch, the fist thing I'd do is make a list of all the (metaphorical) sacred cows inherent to D&D's structure, and look for a more-functional way for the rules to handle each one. Armor class is of course the biggest elephant in that room (since its function was never re-addressed when the game shifted fom the assumption of "one hit equals one wound equals one kill unless his armor gets in the way" to "rolling damage rolls is fun so now every combatant has hit points so that damage rolls are meaningful for handheld weapons as well as siege weapons"; but I would also rework the basic line of stats to be far better balanced: Strength + Durability, Coordination + Reflexes, Intellect + Willpower, Awareness + Equilibrium (the resistance to emotional manipulation). Pairs of active and resistive traits, for physical power, physical finesse, mental power, and mental finesse. Balanced from the ground up. With the added bonus of humans being modeled the same as the other races, with an inherent strong and weak suit, because they tend to be very stubborn (+WIL) but easily goaded (-EQU) into unwise action.
And yes, my list left out Charisma. Because the method of how one approaches their talkmonkeying should be a far more significant factor; Intimidation should be a matter of Willpower (since it comes down to a contest of wills), Persuasion a matter of Intellect (figuring out what approach will appeal to them), and Deceit a matter of Equilibrium (to play on their emotions without letting your own control you).
However, D&D is built on the basic structure of "the traditional tradition of traditional traditionalism says we've always done it this way", rather than on the basic structure of "game balance is a good design principle". So I'm working with the game I've got.
Also to Saelorn, I find it rather amusing that you used the phrase "initiative pass", as the game currently being played by my players-group is Shadowrun 5th ed (I play an ork dog-shaman called Black Lab), and yes it would be a pause for D6 rolls on each tick, but since only hypothetical creatures with a combined DexMod-type and Haste-type bonus of +14 or more would be able to take a tun on evey tick, I see this as a feature rather than a bug, as the per-tick increments of movement (which will vary per-tick based on each D6's result) create a battlefield that shifts and flows in a manner far more similar to a kung fu movie than to a game of Monopoly.
Also, I am aware that Cecil + R(L?)ydia + etc used the ATB subroutine as a behind-the-screen element of the game's progamming, but it was in Chrono Trigger that I was able to see the mechanism being played out in the view of the player-interface, and thus had the chance to truly appreciate its function (as I saw the impact of each SpeedUp pill that I gave to Robo). And I acknowledge that action-speed vs action-impact is a facet of game balance which my structure would DEFINITELY need to address before it would be vible for anything beyond preliminary playtest.
FrogReaver, my system has movement earned on each tick occuring before (and separate from) any actions earned on that tick, so they are factored as separate resources both earned over time. Admittedly, the ability to get more actions does yield the ability to get more movement (by way of the -20 of the Initiative score after each action renewing access to the movement earned in the 10-29 range), and I do have concerns about the game balance consequences of that facet of my structure...
TallIan, I have always felt that the decision made in 3d ed to standardize the ability scores' mechanical impacts down to a single column of data across all six stats created a missed opportunity to individually balance the effectiveness of the stats in comparison to each other. So yes, a lot of good game design does get hamstrung by WotC's decision to pretend that the inherent imbalance (of the traditional traditionalism's traditionally traditional six stats) are best addressed by sweeping them under the rug.
Now, I could be wrong about this, but I think that SAD and MAD refer to dependence on single/multiple ability scores for a character's tactical functionality, and I openly admit that I do not yet have any good answers for those concerns. It is my hope that in the course of playtesting this structure, I'll be able to gain some insight however.
And for the purposes of that preliminary playtest, I think I'll use the suggestion which vincegetorix gave in the original reply, bonus action during the same instant as primary action, reaction being available once after each of your turns, and readying an action putting your own movement on hold until your action is triggered or canceled.
Such is my thinking.