And I think what often gets lost in the discussion is how badly matters are conflated, i.e., the singular good part of a past system element vs. its holistic merit.
Take TP for instance. We may associate it with things like Paladin wanting to alternate between Total Eclipse spam and self-healing in large pulls in Stormblood as not to run out of TP or MP, with Warrior having an opportunity cost for its self-heal (if and only if it was somehow the sole physical damage-dealer in its group), with casters having additional mobility through an effectively-no-cost Sprint to make up for not having the likes of Spineshatter, Shukuchi, or Shoulder Tackle, or physical jobs having a sharp limit to how much AoE they could put out and therefore what was nearly a doom timer double-physical-DPS dungeon pulls, etc., but those are more coincidence than part-and-parcel of TP, the main effect of which was simply to give physical damage dealers a less-leveraged, less-choiceful, less-relevant, and more-constrictive MP analog. Each benefit could be better achieved separate from putting physical jobs on a very limited, Attack Speed-punishing battery life.
Now, we could look at what was successful among iterations of MP, such as Ruin III on HW Summoner and how it created a much more substantial opportunity cost for SMN raises, reducing how much they'd have to pre-pay for their utility and therefore its burden on output in situations that did not need that raise. Perhaps something similar could have been done with TP, whereby one would have both net-negative and net-positive phases, the timings and lengths of which would depend on fight length. But that, too, would cost button-slots to then each be applicable to certain situations that could otherwise be spent on rotational depth usable in a far greater breadth of situations. SSS for mitigating the losses of melee-downtime? Spent instead on your high-TP-cost nuke that you use before first Invigorate and not again until the last expected minute of the fight. Etc., etc. Such "additions" from those lost systems or their reimaginings need to be seen in/through net value, not in isolation.
Or, take Enmity manipulation as it was implemented.
Start with the tanks. Up to 6 tank skills (7 if including ranged enmity attacks, 8 if considering that the "dps stance" could otherwise have been the default / a passive), 1 of which was used on CD in a circle-shirk, another used before the other's longer CD, and 4-6 more used only for the fight's opener and never again. It was bloated, to say the least. Again, we could revitalize this, but consider: to what end? What did we want to achieve by using discrete combos and oGCD CDs (over, say, fewer, more dynamic skills)? Or, if the point was just discrete actions in any form, why use discrete actions over, say, the timing of generic, more bankable oGCDs, the positioning from which to throw them out, etc.?
Or, far worse, consider its place on non-tanks. A CD to hit on CD. And if a single player lacks this CD or fails to hit it, all other players' hitting their CDs is wasted since the tank needs to waste their would-be rDPS to keep Enmity off the outlier.
:: Granted, we should probably also consider "Enmity" itself as implemented still to this day. Did we really want a single-metric nondecaying threat table over, say, impermanent targeting weights? Certainly, having some way to manipulate targeting is good, but Enmity as implemented so far has not only replaced opportunities for far greater net depth (via kiting, CC, directly blocking) but has done shockingly little to leverage itself. (Not that XIV's threat system is unique in that regard. Just worth pointing out for fairness to earlier attempts.)
Etc., etc.