Having returned partly to WoW (especially its PTR, for limitless spec playstyle testing) for the Legion pre-patch and adjusted specs and talents, I have to say the thought that most struck me is that talents aren't necessarily a good thing. If there's a choice that's so good that the spec (in our case, class or job) feels somehow incomplete, less distinct, less coherent, less opportunistic, less challenging, or just less compelling without it... that should be baseline. It's even worse when you have something that carries more interesting gameplay in the same mutually exclusive selection as something that is more mathematically sound, or even just a second, nearly as interesting choice. Against the more powerful choice, both of the interesting ones are lost. Against another interesting choice, one of the two is lost.

The general advantages that I hear people mention about a talent system is that:
- It allows variance in skill floor/cap while allowing multiple actual player skill levels to participate in a larger range of content.
- It allows variance in aesthetics or playstyle (other than skill floor/cap).
- It allows variance in niche/capabilities.
- It gives a sense of progression.

The problem is that for the first to really work, the most difficult choices have to pay off. But, but by nature of the same things that make them more difficult, they are more finicky, and regardless of player skill may often be more condemned by encounter mechanics. This isn't worth taking as a rule of thumb; it will always depend on the particular talents themselves and the particular fight, perhaps even the particular gear levels, raid composition, and fight pace. But it is an accurate, if imprecise, generality, if I may count 8 years personal experience and archives of ElitestJerks and Icyveins as any authority. (The worst being when people condemn a given spec before even mastering it, because when played non-optimally, it doesn't pan out against others that have a narrower range between optimal and non-optimal play.) The same is of course true with the second goal, and the issues it faces. We want real variance, but there is balance to consider as well.

That it not to say that it is impossible to create choices that pay off as intended in almost every scenario. It's just very difficult, and may need to include safety measures that make the tooltips seem "somehow less clean", the design "a bit more cowardly", "less straightforward", "more casual", or "forgiving" to borrow the words of other players on the PTR looking at where those designs have been implemented. (That's not to say that these were all negative opinions; most I spoke to [yes, anecdotal] thought these changes inventive, useful, and wise, but somehow they added a new, and thus kind of weird fluff or underweave to the systems they affected.)

The biggest problem to me is in the third goal, niche/capability adjustment, especially when talents or similar choices can be freely swapped or realloted. At that point, what you really have is all the talents, on a shared cooldown with certain additional locks. And if you're tied into the wrong one when those locks click shut, you're output will almost always be condemned. After all, the goal here is output itself to match a given situation. A wrong talent should definitionally be punitive. (Now, that doesn't necessarily have to be the case, depending on the particular encounter, and the balance between single-target, cleave, AoE, direct, DoT, snapshot, and live dps across your baseline spec — or even across all other talents, but that too would require a specifically aimed balance.) This I personally find de-immersive, especially when those locks or even the sheer UI- (or alternate QoL)-based ease of changing talents is too tight. It's clunky, and it doesn't feel so much like you're a character as a mere avatar. Take from that what criteria, or even butthurt, you will.

As such I can't help but feel that any capability changes should be first and foremost adjustments to and meant for playstyle, the consequence an attractor but seemingly an afterthought to design (but without balance oversights), and have reduced enough consequence over the span of an entire encounter (e.g. a boss fight, especially if with distinct phases preferring distinct skill-sets) to make the choices of equal enough value at least for casual players (or preferably leading to allotted roles rather than pure AoE, pure ST, or pure cleave focus, etc.).

Now for the kickers:
It is not necessary to create talent systems to introduce new playstyles. You simply need an ability set that can produce multiple means of optimal or very near optimal play. A Monk might have one way of playing, under particular criteria and priorities that fits together to what one would call the class, but then have another that's partly unable to be maximized simultaneously, causing you to choose one gameflow or the other, that parses very simularly, and then perhaps a third, fourth, or even fifth that are blends of the two, but nonetheless feel distinct, following their own core rule, that parse almost equally well. This is what I'd call lateral complexity. Each playstyle has slight advantages and disadvantages, and the super-optimal playstyle within a given fight, under a given composition, at given stats, with a given raid dps pace, would likely be a particular line-up of multiple of them. But that should be within, say, 8% of someone playing just the playstyle of their choice, or, say, 5% of the playstyle most generally viable for the fight, or a blend of play-styles that doesn't quite match the actual rhythm of the fight. [INDENT][INDENT]Perhaps the biggest advantage of this is that while we still want the playstyles to score similarly, they're not technically obliged to, because at the least the job still has something else it can rely on. It's multiple specs in one. If you get bored, you start learning how to master some other component. The skill floor remains the same. The skill ceiling increases tremendously.

As such, I personally believe that specs are best used only when either the shades of playstyle are too close and the overall range can't be extended otherwise or there's just too much button-bloat to be solved through additional means of direct control, etc.

Talent systems, when imagined universally (e.g. the idea that every class must have 3 choices to pick 1 from at levels 15, 30, 45, 60, etc., or 1 per level to a maximum of Y choices, etc.) leave themselves in much the same rutt as our original armory system has done for us. Consistent structure may be attractive, but it is limiting, and is not necessary. It may be better, when allowing talents, to allow any number of selections and any number of choices in each.

Finally, we already have the basis for some very unique takes on player customization if we are willing to refresh some of our legacy systems, and supply new criteria to them. The last thing the game needs (apart from wasted or short-lived content) is to borrow a hackneyed concept in an obvious form when it could do it more refreshingly, and quite possibly better.