The Deluge

A monsoon is coming. We will soon inundate you with Mists of Pandaria information, starting with the upcoming media event and everything that follows. It's going to be a very exciting time for World of Warcraft, and we are all super impatient for it to happen.

But... we're not quite there yet. I want to make that clear upfront, because this blog isn't directly Mists of Pandaria related. You won't find any announcements here, just a philosophical discussion that you may or may not find interesting. If you're looking for thrilling announcements, you know what I'm going to say: Soon™.

Multiple DPS Roles

I said this blog isn't directly relevant though, because I want to discuss a topic that we did struggle with a lot during Mists development, and indeed through most of World of Warcraft. We have classes with multiple DPS specs, and for mage, warlock, hunter, rogue, warrior and death knight, there isn't even a melee vs. ranged distinction between those DPS specs. The question comes up all the time: "what is the role of these roles?" I don't think there is a right answer here, and we've even changed the design a few times over the last several years. Again, I'm not couching this in terms of an imminent announcement or anything. This is fundamentally one of those designs that could go in a lot of different directions. It's something we discuss a lot, and we figured given the strong opinions of our forum-posting community, many of you probably do as well.
A paladin can choose from among specs that let her be a tank, melee DPS or healer, and can shift around which role she fills in a raid or BG team from week to week. Through the Dual Spec feature, she can even do so within a single evening. If her group doesn't need another healer, or if she needs a break from tanking, she can become a DPS spec fairly easily without having to swap to a different character. A warlock doesn't have that luxury. Yet, the warlock still has three specs. Is the idea, then, that you are supposed to swap from Destruction to Demonology and back depending on the situation? Is the idea that you play Affliction if you like dots and Destruction if you like nukes? Or do you just switch to whatever theoretically does 1% more DPS for the next fight?
Players are sometimes cavalier about throwing around the claim that there's a "lack of design direction" when they want their character buffed. Of course, classes always have a design direction; players just sometimes disagree with it. My point is that just because we debate whether the current design is the best possible one doesn't mean there isn't a design at all. That distinction is important. And of course, we do have a directive for which DPS spec you should play: whichever one you enjoy the most. But that doesn't mean that is the best model or that it can't ever change. There are other models we could try.

Model One – Everyone is equal all the time

If your DPS and utility are the same across specs, then you just play whichever one you prefer. Maybe you like the kit of the Frost mage, or maybe you like the rotation of the Fury warrior, so you play them. As I said above, this has been the model we have used for a while now, with mixed success. The challenge is that "all the time" caveat. We can get all of the DPS specs pretty close together on target dummies, and indeed they actually are very close on target dummies today. Our encounters aren't target dummies though. Having some adds increases the damage of dot-specs. Having lots of adds increases the damage of strong AE specs. Having to move on a fight, and how often and far you have to move, can cause DPS to go up or down differently. Even if DPS is only off by a few percentage points, many players will respec to the one with the highest DPS (even if it's theoretical, even if for them they will do lower personal DPS than if they had stuck with a more familiar spec). A mage who just loves Fire might be frustrated if he ever has to go Arcane, while another player might be happy that he gets to try different specs for different fights.
The class stacking we've seen on the Spine of Deathwing encounter relates to the need for massive burst damage in a specific window, such that the difference between a one minute DPS cooldown and a two minute DPS cooldown matters. Even if we could make sure every spec had the same AE vs. single target damage, do we now need to also ensure every spec can do the same DPS in burst windows of various lengths? Is that even mathematically possible? Or do we just test every spec for every raid encounter of the current tier and tweak class mechanics around for whatever is the current status quo? That implies a high rate of change, and I wonder if we'd lose a little bit of the fun of experimentation and theorycrafting if it was basically accepted that you could take any spec to any fight and do about the same damage. It's more balanced, yes, but does it lack depth or flavor? Is it fun?

Model Two – Everyone has specialties and you match the spec to the situation

Under this model, we would establish spec specialties. For example, Arcane could be good for single-target fights while Fire is great at AE fights. Some of that design already exists in the game, but we try not to overdo it. If you really like playing one mage spec, or really detest constant spec swapping, then this model isn't going to be to your liking. Furthermore, we don't want to overstrain our boss design by having to meet a certain quota of AE vs. single target fights and movement vs. stationary fights and burn phase vs. longevity fights or whatever. It is also really hard to engineer these situations in Arenas or Battlegrounds (for example, both mobility and burst are extremely desirable in PvP), so in those scenarios there still may just be one acceptable spec.

Model Three – You swap specs to gain specific utility

If we used this model, then you might switch out to a different spec to gain a specific spell. Again, we have some of this today. A DK might want Unholy's Anti-Magic Zone for a certain fight. Hunters might go Beastmaster to pick up a missing raid buff. Mages might go Fire for situations where Combustion shines. Druids might go Balance when they need the knockback from Typhoon. A little of this sort of thing goes a long way though. As in Model One, not every player wants to have to swap specs. If you just like Survival, you might resent having to go BM to just to buff someone. If knockbacks are too potent, then it really constrains your raid composition and makes even casual guilds feel like they need to keep a stable of alts or benched players for every fight. If, for example, there wasn't a boss in the current raid tier for which warrior abilities really shine, then warriors start to feel like a third wheel, yet trying to make sure every boss in a tier has a moment for every spec to shine is a pretty daunting task.
The extreme case of this is the "utility" spec who does middling DPS, but brings a lot of synergy and utility that improves all of the other specs. This was the Burning Crusade model, where classes like shaman and Shadow priests were brought to raids just to make the pure classes (and warriors, who were always treated as pure classes back then for some reason) do better DPS. In Lich King, we changed the design to make different raid buffs and abilities more widespread and give groups much more flexibility in their raid (and to some extent dungeon) comps. We heard from Shadow priests that they wanted to do competitive damage, not just be there to make everyone else more awesome. But even today we get a lot of requests to improve the utility of someone's spec so that they are more likely to get invited to a group.

Model Four – There is just a best spec for PvP and PvE

This was the model of vanilla World of Warcraft, and we understand some players wouldn't mind it returning. In this model Arms and Frost and Subtlety (and other specs) were designed to be good for PvP, while others, Fury and Fire and Combat perhaps, were designed to be good for PvE. The PvP specs might have better mobility or survivability or burst damage, while the PvE specs have better sustained damage over the course of a 6-10 minute boss fight. A lot has changed since vanilla. We don't make many raid or dungeon encounters these days where DPS specs can just stand in one place and burn down a boss. Mobility, survivability, and burst damage can all be really useful on particular encounters, sometimes trumping the higher DPS offered by a competing spec. (There's that old adage that dead do zero DPS.) In addition, if there is a PvP spec and a PvE spec, then for pure classes that implies that your third spec lacks much of a role. (The good leveling spec? Is that exciting?) Furthermore, our Mists of Pandaria talent tree design explicitly takes away some of the tools from the traditional PvP specs and makes them available to other specs in the class. If this works out, then you can take your Frost mage raiding, or have an Arcane mage for PvP who uses some of what traditionally were Frost's control and escape tools. That's great if you PvP and love Arcane, or PvE and love Frost. It's less cool if you were the kind of player who was totally comfortable with the simpler (and possibly easier to balance) design of having dedicated PvP vs. PvE specs.

Model Five – Don't have multiple DPS roles

This is the most controversial model and the one that would require the most change, meaning we are almost certainly never going to do it. For sake of completeness though, you can argue that classes never should have been designed with multiple specs that fill the same role. In this model, either Arms or Fury goes away and gets replaced with something. (Archery? Healing?) Warlocks and other pure classes would need a massive redo to end up with say a melee and tanking warlock. Everyone becomes a hybrid. The hardest decisions becomes whether you want to be the ranged or melee DPS version of your class (like druids or shaman). This idea is elegant from a design perspective because it un-asks all of those questions about how much more damage pure classes should do than hybrids to justify their narrower utility. But, perhaps counter-intuitively, elegant designs often aren't the strongest ones (I could write a whole blog on that topic alone). Model Five is the kind of rhetorical question you could go back in time and ask before WoW launched, but not the kind of thing we could change today without taking an enormous amount of effort, to say nothing of the irate players who would feel bamboozled that we were so dramatically changing their character out from under them. I try to never say never, but this model isn't the kind of change you make in a mature game. It's here only for completeness and because I suspect some of you will bring it up.

But Which is the Best Model?

Hell if I know! I fundamentally believe that none of these models is, without question, the obvious right one. All of them have advantages and disadvantages, and there are probably other models you could come up with that are variants on these five, or perhaps even something new. Like I said, we're not announcing a philosophy change yet. If we get enough feedback for one model or another, we might eventually change our minds. Also for this blog we're going to lock the comments and ask that you post your replies in this forum thread. Just remember that even we don't believe that there is one correct answer, so please keep that in mind when you're composing your feedback.

Greg "Ghostcrawler" Street is the lead systems designer for World of Warcraft and holds the world record Wild Strike crit... at least until beta starts.