A comprehensive explanation would probably take well over 1000 words, so let's start from a simple illustrative example:
At present, DRG has 9 single-target weaponskills. Since, in content, they will never be used in any order except True Thrust, Disembowel, Chaos Thrust, Wheeling Thrust, Fang & Claw, True Thrust, Vorpal Thrust, Full Thrust, Fang & Claw, Wheeling Thrust and that order will repeat infinitely, they amount to a single decision, to be made once every 5-10 GCDs. And is a single choice even really... a choice?
Not that I'd recommend it, but all 9 off those keys could be handled with equal depth... by, effectively, "auto-battle", with you just weaving in Life Surge before Full Thrust, all other oGCDs as needed, and so forth.
So at present you have 9 keys that do no better a job than 0 keys except in that the finger-dance aspect of hitting a different key per GCD helps keep your timing grounded (while punishing those with manual disabilities, but that's besides the point).
But let's say we instead use... 1 key, but this time allowing for a decision in every possible GCD rather than only once per 5-10 GCDs, with accordant changes in the toolkit. Your "auto-battle" (term used only to avoid confusion with "auto-attacks") would sequence down one line or the other of combos, but now those combos have been opened up. You can now swap between combos at any point along the combo, choicefully favoring either long-term damage or direct burst. You've only one key, swapping between "Finesse Mode" and "Power Mode", for instance, but it'd already allow me more control over my combos than you had with 9 keys previously.
Let's add in the AoE weapon skills to be comprehensive. Presently, we have 3 of them. They all amount to a single decision. The system that binds these 3 actions into that decision acts merely to increase the punishment for having less than 3 GCDs to spend in that particular line of action, making the points of decisions between AoE and ST less modular or fluid. It adds only one manner of restriction-as-optimization -- that Life Surge should be held for Coerthan Tempest.
Let's convert those 3 buttons, for a total of 12 buttons, into our a single button for a total of 2. We now have "Power Mode", "Finesse Mode", and "Piercing Mode", where one is the default and then swapping into either other would make the previous mode take that key's former place. I.e. if Finesse is the default, Power and Pierce are my alternatives across buttons 1 and 2, respectively, and if I swap to Power then Finesse takes the place of button 1, and if I swap to Pierce then Finesse takes the place of button 2.
Voila, I can now charge up -- at whatever varying efficiency we see fit by which to add further constraints, or hopefully thereby add depth to the system -- a Coerthan Tempest of an Impulse Driver -> Disembowel.
I've more available timings at which to swap things around, and thereby gameplay concerns and optimizations, yet I'm using a sixth of the previous number of keys.
But that's not even making full use of these 2 keys or modes. Let's add a further bit to the swaps. Let's say... you can hold down the key to consume MP but generate a further effect on the next attack.Alright, now I have greater combo complexity than exists in the current build of Dragoon atop, potentially, all the mobility of Elusive Jump, GCD-optimization akin to Life Surge but far more frequently and fluidly available, and window optimization in ways novel to Dragoon. I've still used only... 2 buttons.
- If I charge up while swapping into Power mode, that will lend the attack further potency on its direct attack (such that there's even more synergetic value in comboing from Impulse to Disembowel to Full Thrust, for instance). You could make this percentile (favoring MP consumption on Full Thrust), or Flat (favoring certain other effects). You could have this bonus linger to some extent into a further attack, synergizing with a quick swap into Pierce thereafter. You could have the charge-up affect different skills differently. There's tons of stuff you could do here, and yet it doesn't cost a single button extra.
- Let's say for now that if I charge up while swapping into Pierce, the previous would-be attack is retained, but will now dash forward and cleave (allowing me to mass-DoT, or mass-vuln, albeit at lesser duration or effectiveness). The cleave range generated per moment of charge depends on the affected skill's potency, so it isn't used for Full Thrust spam constantly. In fact, if I want to maximize it's use for mobility, I'd pop this for True Thrust or Doom Spike only.
- And just to spitball a finish, let's say Finesse gives me some added Attack and Movement Speed, lingering; essentially allowing me to bank the uptime consumed on some fancy spear-flips to really milk a buff window starting as those spear-flips end.
And that hasn't even touched on any of the other obvious means of increased depth: skill uniqueness and the undermechanics of combat have yet to be utilized. At present, all but two skills still offer nothing but raw direct potency, and the last two at best provide that potency in slightly less direct ways (DoT damage and potency-as-percent-of-window). They don't have to be that barebone, especially if undermechanics are used in kind.
What if armor were a thing, for instance? Should True Thrust, which appears Dragoon's most straightforward deep-striking attack, really have no advantage in piercing armor over the other skills? What if there was at least a modicum of physics? Should Vorpal do nothing to, as per its appearance, push enemy bodies or blades to each side? Should Full Thrust not toss lighter enemies skyward? Should Disembowel's double-slap, as if parrying, honestly have no effect or its turn, as if ducking under a misdirected blow, not synergize with the first half of its animation? What we had a modicum of AI, rather than merely a CD schedule, a movement speed, and a target-directing table with zero interactions beyond damage-times-modifier in, enmity out? Should the wild, flurried VFX of Chaos Thrust really have no impact on mob behavior or behavior-informing metrics? Should Fang & Claw's dodge component really be only for show?
Naturally, to take advantage of those different skills' unique effects, you want more buttons. Heck, you might even want all 12 of them. But that's only effective design when you let each amount to at least a decision each for every given GCD, rather than a third, fifth, or ninth of a decision as offered every 3, 5, or 10 GCDs. Having a lot of buttons is fine, but it ought to come with commensurate depth, rather than just depth in pretense -- the reality of which, by being separated despite marking no separate decision, is squandered.