I've been thinking about this for a while, and while there's a general consensus that DRK needs help more than the other two tanks right now balance-wise, and there are a lot of threads making suggestions on how to change things, I think that (with no offense intended to people who have proposed the more sweeping design changes) there are comparatively few ideas floating around that are in line with SE's policy of mostly making simple variable changes to accomplish class balance. Going down the list of job skills, I'll try to identify where I think design and balance problems exist with the class, and how each change will address those issues.
Unleash
The problem: The primary problem with Unleash is that it is now an extremely counter-intuitive skill, and teaches new players some very bad habits. It's natural for a new player to look at the skill and think "Okay, this is my AoE damage skill, so I should use it whenever I'm fighting multiple enemies". The problem is that the damage on it is extremely low, to the point where single-target rotations are usually better than Unleash.
Before level 26, Hard Slash -> Syphon Strike is stronger than Unleash on 1-3 enemies, as strong on 4 enemies, and only becomes weaker at 5+ enemies (rare to see in dungeons at those levels).
After level 26, Power Slash combos are stronger than Unleash on 1-4 enemies, and only marginally weaker at 5+ enemies (again, rare to see in that level range).
After level 45, Souleater combos along with DADP on cooldown are stronger than Unleash on 1-8 enemies, and only marginally weaker on 9+ enemies (while providing small amounts of healing and mitigation that Unleash does not, and is infinitely sustainable, whereas Unleash will drain you of MP).
New players are very unlikely to realize the way that these numbers shake out, and so many will be taught to play the class poorly, using Unleash on pulls where it is ineffective for them to do so.
The secondary problem is that whether you're using single-target or Unleash at those levels, it's simply not effective enough. Until DRK gets Abyssal Drain, its damage on an AoE pull is nearly as bad as PLD was before they got Total Eclipse. When WAR gets 120 potency Overpowers right out of the gate, and even PLD has Circle of Scorn and Total Eclipse by the time they hit level 50, DRK is left in the dust and cannot contribute meaningfully to a party in any level 50 content.
The solution: Unleash should be changed from a Spell to an Ability - that is, it should be made off-GCD, with a cooldown from 2-5s.
The first thing this does is to teach new players how to play effectively - that they should be continuing to use their GCD attacks with off-GCDs woven in between them. Right off the bat, this clues novice players in to the idea that DRK is the tank class with a high APM and many off-GCD skills to be used.
It also brings DRK's performance in low-level dungeons up to an acceptable level. 50 AoE potency, added to DRK's single-target combos spread around a pack, puts them roughly on par with WAR (and at level 46+, PLD) in terms of damage on most AoE pulls.
At the same time, despite being oGCD, this doesn't cause any imbalances to DRK at high-level. The MPotency ratio on an oGCD Unleash would be roughly similar to a non-DA Dark Passenger, which most people agree is a pretty poor return on damage for your MP.
Power Slash
The problem:
When Stormblood launched, the enmity multipliers on Rage of Halone and Butcher's Block were raised from 5.5x to 7x. Power Slash remains at 5.5x.
The solution: This should be bumped up to 7x, to reach parity with the other tanks.
Presumably the counter-argument here is that the Dark Arts effect on Power Slash was buffed to double the enmity bonus, taking it from 5.5x to 11x.
However, this is not necessary to balance tank enmity. PLD has Shield Swipe (450 enmity-potency, 15s cooldown) and Circle of Scorn (300 enmity-potency, 25s cooldown) to bolster their single-target enmity, even without touching Rage of Halone. WAR has Onslaught (a ridiculous 1000 enmity-potency, used mostly during Berserks but potentially on a 15s cooldown), and still generates BG if they use Butcher's Block, making it a relatively light DPS loss.
Bringing non-DA Power Slash up to a 7x multiplier to match the other tanks is absolutely fair. DA Power Slash being raised up to double that at a 14x multiplier is not exactly an insane boon to DRK, compared to the tools listed above that the other tanks have - and frankly, being able to pump out that much enmity (at the cost of a substantial amount of MP/damage) should be 'a unique thing DRK can do', similar to any number of unique perks that the other tanks have.
Blood Price
The problem: Blood Price's effect on MP regeneration is almost unnoticeable in single-target prior to the level 66 trait (and is pretty bad even with that trait), and is incredibly weak even in AoE pulls until you're being hit by 8-10 enemies.
At the same time, though, we all have to recognize that Blood Price in its 3.x incarnation was wildly overpowered, and we can't go back to that again.
The solution: While under the effect of Blood Price, MP costs should be reduced by 25%.
This has the effect of making all your MP go about 33% farther with Blood Price up, which is a nice, useful effect, both in single-target and in AoE, without giving you infinite MP in AoE, and without giving you an insignificant amount of MP in single-target. It's equally useful on small-pull AoE with 3-4 enemies, and on big-pull AoE with 10+ enemies. And maybe more importantly, it actually makes Blood Price feel like a Grit-centric equal to Blood Weapon, rather than a terrible 3rd Place ribbon that just makes it feel even worse when you can't get BW up.
Dark Passenger
The problem: The MP cost on Dark Passenger is too high for what it does, and has largely been excised from the hotbars of a lot of DRKs, without a particularly visible loss of efficacy.
The solution: I'm not alone on this one, but it bears repeating anyway: The MP cost on Dark Passenger should be reverted to its pre-SB state, and put back down to (at level 70) 1200 MP.
This is a very slight increase to DRK damage in single-target: Basically, every 30 seconds, you can exchange half a DA worth of MP (that is, 70 potency worth) for a 100 potency attack. 30 potency every 30 seconds - somewhere on the order of 0.3% extra damage, but it would make the class feel much nicer to play.
And of course, in AoE, it would make DP an unambiguously good skill to use, with or without DA. While it's not terrible in AoE in its current state, and is a little underrated, it also wouldn't be imbalanced for it to be a little cheaper: 3600 MP for 240 potency is not going to make Abyssal Drain obsolete or anything.
Dark Mind
The problem: While Stormblood did PLD the mercy of eliminating specific damage-type conditional cooldowns by making Block apply to any sort of damage, DRK feels left behind in that regard, stuck with a major cooldown that can only be used on magic tankbusters. It also costs MP to use effectively, and can't even really be described as "overpowered" for the attacks it works on, due to the short duration, but is merely "powerful" in those situations.
The solution: As I see it, there are two solutions here.
One is for SE to say "No, this cooldown stays magic-only, but we promise that Deltascape is the new model for content design, where the majority of the tankbusters are magical." In this case, the damage-type dependency can stay, but the effect (and DA effect) should be boosted. The base mitigation of Dark Mind should be increased from 15% to 30%, and the Dark Arts effect should be to instead double the duration from 10s to 20s. It isn't uncommon to run into situations where spikes of damage, or consecutive cleaves/tankbusters, are spaced 10, 12, or 15 seconds apart. In many of those cases, (free, damage-boosting, universal) Vengeance is a better cooldown than (MP/damage-costly, type-specific) Dark Mind, because of the extra duration. If it's going to remain such a specific mitigation tool, then it should have power to match.
The second solution is what I'd prefer. We have at least one all-magic fight in Deltascape (Catastrophe), and another fight where the vast majority of tank-threatening damage is magic (Exdeath and Neo Exdeath). Despite this, Dark Mind doesn't feel overpowered in those fights - it just feels appropriate. Therefore, I really can't see any reason why Dark Mind should be restricted to magic damage in the first place. Turning it into universal mitigation, at 15% free/30% DA, would basically just put it on par with one of the many tools in WAR's suite of cooldowns (while I'm at it, yeah, Raw Intuition should just be changed to 20% damage reduction from the front - but that's a tangent). The weakness of its short duration remains, for situations where tankbusters are placed 10-15s apart, but it doesn't simply become a dead cooldown on any fight where SE forgot to make the tankbusters magical (see, so far in SB: Susano, Halicarnassus).
Shadow Wall:
I'm just mentioning this so that people don't think I forgot it: I think that Shadow Wall is fine, so long as Dark Mind is brought up to par, and TBN stays the way it is. Comparing it directly across from Vengeance or Sentinel, it comes out the obvious loser, but that isn't a problem so long as the rest of DRK's toolkit makes up for that difference.
Plunge
The problem: Where to begin? The animation lock is too long; DRK should not have any skills that eat up both 'oGCD slots' when they're already the most oGCD-heavy class in the game. The range is too low; every equivalent gap-closer has a 20y range. There are places in fights that seem specifically designed to make Plunge an exercise in frustration, with knockbacks happening 25 seconds apart compared to Plunge's 30s cooldown.
The solution: Plunge needs buffs on just about every front.
The animation lock should be reduced - the target here is that if a DRK is standing in melee range (and doesn't need Plunge for the mobility), they should be able to double-weave Plunge and Dark Arts in the same oGCD, with Blood Weapon up.
The range should be extended - just bump it up to 20y, like every other gap-closer.
The cooldown should be reduced, from 30s to 20s. This has three effects: First, it makes Plunge more usable as a gap-closer in fights that have multiple consecutive knockbacks. Second, it gives DRK a small but necessary DPS boost - Plunge currently tends to make up between 3% and 4% of a DRK's total damage, so increasing the number of Plunges by 50% would raise DPS by 1.5-2%, closing a lot of the (unnecessary) DPS gap between DRK and PLD/WAR. Third, it aligns Plunge's cooldown to Blood Weapon. Just as PLD will always get one Spirits Within inside of their Fight or Flight, and WAR will always get one Upheaval inside their Berserk, because those skills have 30s cooldowns compared to their respective buff's 60s cooldown, Plunge at 20s would always line up so that every second Plunge lands inside of a Blood Weapon.
Sole Survivor
The problem: Sole Survivor has a weak effect, is hard to use, typically can only be used during really inopportune moments, and is completely unusable in many fights without adds.
The solution: Get rid of Sole Survivor as a debuff. Just make it an instantly-poppable heal, for 20% HP and 20% MP.
20% HP is a weaker heal than Clemency, even without Convalescence or Requiescat. It's about as strong as Defiance Equilibrium, and is weaker than Berserk-Defiance Equilibrium. The cooldown is twice as long as Equilibrium, and (IIRC) %-based heals like that don't produce enmity the way that Equilibrium does.
20% MP is less than a single Dark Arts/TBN. It's nice to get, but doesn't really do that much for you.
DRK has the fewest/weakest self-healing options of all the tanks, and has a very limited ability to generate MP on-demand if they can't do a single-target melee rotation, despite the fact that all of their AoE and ranged skills require MP. Sole Survivor being a universally-applicable solution to both of those issues closes the gap a little bit in a class-unique way.
Quietus
The problem: Quietus's Dark Arts effect is terrible, and should never be used, under any circumstance.
The solution: On the other hand, there's an obvious hole in DRK's AoE toolkit. During a dungeon pull, when longevity is a factor, non-DA Quietus really is a godsend, giving you enough MP to keep going. But during AoE sections of raids and trials, you often find yourself in a situation where you really don't care how long your MP lasts - you just want to get as much AoE damage in, as quickly as possible.
DA Quietus should not restore any MP on hit (unless you're running it with Blood Weapon), but the potency should be raised up from 160 to 280 - a boost of 120, slightly lower than Dark Passenger's DA effect, to the same overall potency as Decimate (at, let's be clear, a much greater cost than Decimate, considering the Blood loss as well as the opportunity cost of not getting the MP regeneration, to make up for DRK having better oGCDs for AoE).
Bloodspiller
The problem: The weird way in which Bloodspiller has different potencies for unbuffed, Grit, Dark Arts, and Grit+Dark Arts is clumsy as hell, and makes the skill a nightmare to balance since SE seems to not adjust all the aspects at once. And since TBN ties the potency of Bloodspiller and the potency boosts of Dark Arts together in a way that means that the relative value of each is affected by the other, this is a skill that's going to see a lot of tweaking over the lifetime of the game.
The solution: "Ignores the damage penalty inflicted by Grit." It's really that simple.
In Summary
The cumulative effect of these changes would be to:
- Make DRK more intuitive, and more effective, for new player and/or at low-level synced content, which are two major failings of the class's current design.
- Bring DRK up to par with PLD and WAR in terms of enmity generation, by making non-DA Power Slash less horrible, DA Power Slash less underwhelming, and by giving them a usable oGCD enmity skill as an answer to the other tanks' respective abilities.
- Bring back a couple skills that were overnerfed in the transition from Heavensward to Stormblood, making it feel rewarding to use DRK's entire toolkit, without simply buffing raw power to the point where those skills become overpowered.
- Improve DRK's self-sustain (from 'almost nothing' to 'worse than the other tanks, but with lower DPS costs'), and either remove the unique susceptibility that DRK has to content design with physical-only tankbusters, or at least give them commensurate power in the instances where their magic-only toolkit can be brought to bear.
- Improve DRK's DPS, and the alignment of some of their cooldowns with respect to both other cooldowns and to some fight timings that currently mess with a lot of DRK's usefulness. In truth, these changes would probably still leave DRK as the lowest-DPS tank, and further adjustments may more may not be necessary on that front, but they would close the gap pretty significantly, and would improve the overall flow of gameplay.