Lhivera’s Library

Adventures in World of Warcraft, Dragon Age, the real world, and beyond

The New Mage

Sunday, August 26 2012 at 11:09pm CDT

The Mage class got a significant overhaul for Mists of Pandaria, but if all goes well, it won't feel like a drastic change. Unlike the Warlock overhaul that involves significant changes in design, the Mage overhaul was instead an attempt to preserve the essence of the existing design while replacing a lot of the machinery under the hood. On the eve of the 5.0.4 release, I wanted to reflect on how the class is evolving, what I think has gone well, and a couple of things that I think can still be improved. This will, of course, be from a PvE perspective.

Homogenizing Capabilities

Two seemingly contradictory requests have been made, politely and less so, for years: for all three specs to work in both PvE and PvP, and for all three specs to have a distinct and flavorful playstyle. After years of trying to give each spec its own tools to compete in PvP, Blizzard seems to have reached the conclusion that there simply isn't any substitute for a solid foundation of control and survivability options; only once every spec has those can they successfully be built out from that point. And so they've taken two essential abilities that were previously only available to Frost, and one that required points in the Frost tree, and made them either baseline or available through the talent system. Every spec can now Shatter and/or Deep Freeze a frozen target, and every spec now has access to Ice Barrier.

This decision has led to some complaints about homogenization. Some people feel that the specs have lost their unique flavor; others feel that having each spec designed to be used in a specific aspect of the game is a more interesting design. There is some overlap between these positions.

I don't feel the latter is even worth addressing; it seems so self-evidently and completely wrongheaded to me that I wouldn't even know where to begin. It would be like someone coming up to me and suggesting that it would be better to design a meal around ground-up chalk mixed with mud; you don't argue with such a statement so much as sidle carefully and quietly away from the person who delivered it, making no sudden movements and avoiding eye contact.

As for the first position, there is some merit to it. The specs have been homogenized to some degree. I don't see this as a problem for three reasons:

  1. The increase in diversity that will result from seeing all specs in almost any aspect of the game more than outweighs the decrease that results from them sharing some significant Frost-flavored abilities.
  2. While the specs are sharing some additional abilities, they have also received additional unique abilities. The result, to me, is homogenization of capability but not homogenization of playstyle.
  3. With the talent system in the mix, we will see a great deal of variety within specs. Gone are the days where there are three specs, and every Mage of a given spec is functionally identical. A Frost Mage who chooses Frost Bomb, Scorch, and Invocation is going to play very differently from a Frost Mage who chooses Ice Floes, Nether Tempest, and Rune of Power. These differences are greater than we have ever experienced before; shifting a point from Netherwind Presence to Ignite in Cataclysm may technically count as a different spec, but it has no effect on playstyle.

All told, I think the developers have done an excellent job of giving us the freedom to play the spec we want to play, in the content in which we want to play it, while maintaining distinct playstyles for each spec — and providing us with unprecedented freedom to customize that playstyle.

Mobility and Mages

I wanted to quote something that Ghostcrawler posted recently in a Hunter thread, because it applies to all ranged classes:

Movement should be terrible for a ranged spec. Full stop. All of the various mechanics we put in from Spiritwalker's Grace to Aspect of the Fox are to make moving less terrible, but it should still be pretty terrible. (And if you think there are specs not penalized enough by moving, please let us know, though probably not in this thread. We felt that ranged movement got a bit out of control in Cataclysm so we've definitely made an attempt to scale it back.)

I bring this up because one of the primary complaints about Mage design during beta was that our Level 90 talents — or at least, the two that provide the most reliable and controllable damage output — restrict our movement. And yes, it's true, they do indeed restrict our movement — that's exactly what they're supposed to do.

Blizzard likes design that emphasizes the qualities of classes and specs. When a class or spec has strengths and weaknesses that the devs think work, and help to define the character and playstyle of that class or spec, they like to make those strengths and weaknesses more pronounced. This doesn't always work — as above, they needed to spread some Frost abilities around to make Fire and Arcane work in PvP and to allow for buffing Frost's damage in PvE — but when it does, it helps distinguish classes from each other and provide a stronger identity.

Mages are intended to deal heavy damage when allowed to plant their feet, and poor damage when forced to move. Invocation and Rune of Power reflect this design perfectly. Because both talents incorporate some necessary downtime, and because damage is balanced around reasonable — not perfect — uptime, we know that damage being dealt when the buffs are up is higher than the average around which we're balanced.

Bear with me while I illustrate this. At 20% haste, your absolute best possible uptime with Invocation is 40 seconds out of every 45, with five seconds of zero damage generation. Your average damage output with perfect usage is going to be:

40 / 45 * 1.25 = 1.1111 = 111.11% of the damage you do without the buff

Now we know Blizzard isn't balancing around perfect uptime. Let's say they decide that, given common raid mechanics, a reasonable expected uptime is 40 seconds out of every 55 seconds. Your average damage output is now going to be:

40 / 55 * 1.25 + 10 / 55 = 1.0909 = 109.09% of the damage you do without the buff

Just to be clear, I'm not saying this is the number they balance us around. I don't actually believe there is a specific number around which they balance us — I think the true picture is more complex and fuzzier than that. I'm just demonstrating how this works.

So if we do 100% damage without the buff, and we are balanced around this hypothetical average of 109%, our damage while the buff is up is:

1.25 / 1.0909 = 1.1458 = 14.58% higher than the value around which (in this example) we're balanced

You're going to see a lot of people complaining that we need to "jump through hoops" to maintain perfect uptime just to produce "average" DPS. Don't you believe it.

Jump through hoops? Yes, definitely. The hoops are what define the playstyle. If we didn't jump through any hoops, we'd all be spending all our time spamming a single zero-cooldown spell.

Perfect uptime? No. We have that straight out of the devs' mouth.

Average DPS? Only in a sense. The average, sustained damage we put out is balanced around decent use of these talents, yes, and should be expected to put us on par with, not above, other classes and DPS specs. But the DPS while the buff is up will be significantly higher than that — in our above example, nearly 15% higher. And that means we can take advantage of opportunities to focus downtime into periods of the fight where damage is not as important, and focus uptime into periods where high damage is critical.

I anticipate much more complaining about this as we enter Mists of Pandaria and more people think about choosing a Level 90 talent. But to me, these spells help reinforce the basic idea of how the class is supposed to work: we can be exceptionally mobile and slippery, and we can deal excellent damage, but we can't do both at the same time.

Arcane: Output Control and Mana Management

In Cataclysm, the developers tried to build a playstyle for Arcane that centered on managing mana and shifting damage around. The idea was that you should be able to play somewhat conservatively for long-term DPS, or you could burn your mana fast for burst on demand, sacrificing endurance and long-term DPS in the process.

Two things foiled this design: early balance adjustments made to allow Arcane to compete with Fire reduced mana costs so much that mana management was not nearly as important as intended. Later-tier gear boosted mana pools and regeneration rates enough to render mana nearly irrelevant. Arcane suffered from the same problem as healer specs: they simply outgeared their own design. Burn phases got longer, conserve phases got less and less necessary, and Arcane once again became known as a two-button spec. This was only slightly unfair; maximizing DPS required smarter play than most non-Mages seemed willing to admit, but playing poorly produced results closer to maximum than most Mages seemed willing to admit.

Arcane also suffered from relatively poor AOE and mobility, and Arcane Barrage, meant to be a signature spell, became less and less useful as the mana management game disappeared.

The Mists of Pandaria design is a second, much more successful attempt to develop this playstyle. Failing to manage mana will significantly hurt your DPS; doing it well will reward you with highly competitive results. Burst is now something you use when demanded rather than all the time, and it involves a tradeoff against your long-term performance. Every spell has an important place in the rotation; Arcane Barrage is the only means you have to control stacks, while Arcane Missiles provides both burst damage and longevity.

While many people aren't recognizing this fact yet, Arcane's mobility has been improved; damage while actually moving will be worse than Fire and Frost, but while they will burn their high-powered instants going into movement and need to start with their low DPS spells after movement, Arcane Missiles procs will allow Arcane to use its high-powered proc immediately after movement, quickly ramping its charges back up while bolstering its Mana Adept bonus.

Additionally, Arcane now has a potent one-shot AOE spell, now being capable of hitting up to seven targets with a 232% damage Arcane Barrage. Combined with the class's substantial cross-spec AOE arsenal, it should be able to put up a strong showing in multi-target situations.

The overall feel is similar, but more complex, more controlled, and more capable of handling situations other than single-target stationary damage.

A Criticism

There's only one significant error in the new design, I believe, and that's the lack of any way to discharge the Arcane Charge stack without hitting secondary targets. I might have suggested a conversion of the Fire Blast spell into a version that deals Arcane damage, less than Arcane Barrage would on a single target, and clears Arcane Charges. Arcane Barrage would still be the preferred spell in almost every situation, but should the need arise to clear the charge stack without hitting secondary targets, the option would be available.

Edit: Please see my comment below for further thoughts and a change of mind on this.

Fire: DOTs, Crits, and Reduced RNG

It seems pretty clear that Blizzard was mostly happy with the Fire playstyle as-is, and didn't want to change it much. Fireball still generates crits, crits still activate Pyroblasts, and DOTs still generate a substantial chunk of the total damage.

However, Blizzard did want to address two common complaints:

  • Fire's damage varies too much based on random events
  • Fire's need to wait for a random event to use DOT spreading makes its AOE unreliable

In addition, Fire's AOE advantage was simply unacceptable under the new design. If things like control, survivability, and single-target DPS were being normalized, Fire couldn't be permitted to retain an overwhelming advantage on AOE encounters.

All three of these issues have been solidly addressed. There were two main components to Fire's large RNG variance in Cataclysm: its extraordinarily high crit damage, and the need for two consecutive critical strikes to generate a Pyroblast proc. The former has been addressed by normalizing its crit damage to 200%, and by spreading Ignite across all damage, hits and crits, dealt by its primary spells, rather than concentrating it on critical strikes. The latter has been addressed by giving Fire the ability to use Inferno Blast to force a critical strike and complete the two-crit sequence.

Inferno Blast also solves the second complaint, allowing Fire Mages to spread DOTs as frequently as its cooldown permits. There's a trade-off here; if cleave damage is more important than single-target damage, it may be appropriate to forego a Pyroblast proc to ensure that Inferno Blast is ready to spread DOTs. But that's exactly the sort of decision Blizzard wants us to be making in combat.

However, with more reliable AOE comes less powerful DOT cleaving; bringing Fire down to something more akin to Frost's and Arcane's cleave and AOE capabilities required limiting DOT spreading to two additional targets.

Predictably, these changes have provoked some complaints; it turns out there were Fire Mages who really liked the high-RNG unpredictable nature of Fire. They enjoyed the excitement of catching a good sequence of large crits and turning them into a large Combustion. They liked having to react to, rather than create, Pyroblast procs. Unfortunately, or fortunately depending on your perspective, these people didn't really take up a defense of the playstyle in Wrath and Cataclysm while it was being so heavily criticized.

Personally, I like the result. I like randomness as well, and I think there's a place for a very random spec for those who prefer to play it. But for my personal preference, I like a little more deliberate interactivity, and I think the "get-a-crit-force-a-crit" design will prove attractive to many players.

I don't really see any way in which I would improve Fire.

Frost: Less Control, More Damage

Frost has undergone perhaps the most significant change; it has surrendered some (but not all) of its unique control, and all of its unique survivability, with some of it simply going away, and the rest being made available to all specs. At the same time, it has been retooled into a completely competitive DPS spec; anyone who claims that Frost is not a PvE spec in Mists of Pandaria should be on the receiving end of an ice-cold wedgie.

Building Frost into a competitive, scalable DPS spec took some doing; there was a lot of development iteration over various ideas throughout beta. When the first round of invites went out, the only spell Frost used in PvE that benefitted from Mastery was Ice Lance; this resulted in Ice Lance delivering frequent and massive damage against frozen targets, and Mastery being very difficult to balance against the other secondary stats. The devs quickly decided to spread Mastery around some more, solving several problems at once: the stat itself became more stable, Ice Lance damage came down from the stratosphere, Brain Freeze Frostfire Bolts became essential rather than providing a marginal DPS increase, and the Water Elemental became a much more significant portion of our DPS output.

Apart from its general noncompetitiveness in a raid environment, Frost Mages had a specific complaint regarding the spec's complete lack of cleave damage and its very poor AOE damage. Both have been addressed; in addition to the general class AOE toolset, Frost now has a very powerful AOE spell in the new Frozen Orb, and a decent two-target cleave available in glyphed Ice Lance. It shouldn't have any trouble keeping up.

One complaint I've seen a few times is that the new Frost rotation is simpler than the old. I don't believe this is true. The complexity people tend to talk about is a hypothetical decision-making process on Fingers of Frost, whereby your priority in Cataclysm was:

  1. If Deep Frost is available, cast it.
  2. If Brain Freeze is up and Fingers of Frost is up, cast Frostfire Bolt.
  3. If neither Deep Freeze nor Brain Freeze was available, and you have two charges of Fingers of Frost, cast Ice Lance.

The only problem is, you didn't actually have to do any of this. Extensive simulation work proved that it was both easier and equally effective simply to follow step 3 to ensure that you nearly always had one charge of Fingers of Frost; an add-on could simply light up if you had two charges to tell you to cast an Ice Lance. Once you were keeping a charge banked, you could simply use Deep Freeze and BF-FFB every time they lit up, without regard for your Fingers of Frost charges. Holding BF-FFB until you were sure you had an FOF charge available was unproductive; the vast majority of the time, the charge would be there, and when it wasn't, you were better off using the BF-FFB anyway rather than holding onto it and risking overwriting a Brain Freeze proc while waiting for a Fingers of Frost proc.

The new design is more complex, not less. You now need to maintain one of three Level 75 talents, and one of three Level 90 talents, while the only action that has been removed is the relatively simple act of pushing the Deep Freeze button approximately once every 30 seconds. There are now significant interactions between the procs and Alter Time (you want to use it to try to double them up), cooldowns and Alter Time (you want to make sure you don't use them when Alter Time is active), and the cooldowns and procs together (when glyphed, you want to use Icy Veins in conjunction with Frozen Orb to get as many mini Ice Lances as possible.

I'm very pleased with the way Frost has turned out. The pet feels more significant and powerful, I have some control over the mechanics of Brain Freeze based on my choice of Level 75 talent, Ice Lance and Brain Freeze are both very satisfying buttons to hit, and I'm still managing the part-random, part-controlled generation and consumption of Fingers of Frost procs. One of the things I enjoy most is Frozen Orb, which is now a very powerful spell, and something like a mini-DPS cooldown on a short timer — potent against a single target, spectacular against a group. The spec feels quick, responsive, and strong.

Another Criticism

I have only one real disappointment with Frost, and that's the use of Freeze to generate Fingers of Frost procs. I don't mean that it shouldn't generate charges; I mean that it feels clumsy and weird in a single-target rotation to use a reticule-targed, pet-based AOE spell to force a buff to appear on the Mage. I would very much like to see an additional Water Elemental spell, sharing a cooldown with Freeze, that produces a similar effect in a manner that doesn't feel so much like a Rube Goldberg machine. I like the ability to control procs a bit; I like interacting with the pet. I just dislike this particular method of doing those things.

Preserving the Essentials

I can't know this for certain, but the impression I have from observing the course of Mists of Pandaria Mage development is that the developers started by boiling the class and its three specs down to their basics.

  • Class: high stationary damage, low mobile damage, burst, control, evasion.
  • Arcane: moderate speed, controlled burst at a cost of sustained DPS, mana management.
  • Fire: low speed, crits leading into procs, passive damage over time.
  • Frost: high speed, pet, procs leading into crits.

They then took these blueprints and largely rebuilt the class and specs, attempting to produce a result that maintained these qualities but was built on a more solid foundation, without the patchwork of mechanics and workarounds that have accrued over the years as the specs were retooled and rebalanced over and over again.

The result is a class and specs that, to me at least, feel like modern, streamlined versions of themselves — sort of like climbing out of an antique VW Beetle and into the modern version. Yes, the vintage model had a lot of character and nothing modern can quite replace it, but the new model is a pretty great piece of engineering, and if you're going to share the road with lunatics in SUVs, you're a lot better off driving it.

I think Mists of Pandaria is going to be a good time to be a Mage.


Submitted by grizzlebeez on

First off, great post.

While I see your point regarding the blue post. I think you missed something to the argument of those of us on the other side. It was not a simple complaint of the talents restricting movement, it was a critique that they restricted it too much. I agree with the idea behind the blue post that movement should be a hindrance to range specs. However, we were already naturally hindered in that way. We didn't need additional aspects to furthur restrict us under penalty of DPS loss.

This was especially true for Rune of Power. If you look back at the majority of my arguments on the topic throughout the beta, they pretty much all centered on Rune. That spell seemed too punishing to me, but when I suggested things to edit it, I always kept the basic fundamentals of the spell in place. Even the standing in place part of the spell was fine, but it was executed in a way that was too punishing. Cast time, cooldown, and it restricts you to only standing in this spot right here.

I really never had a problem with Invocation, other than thinking it was lackluster. Just forcing me to evocate every 45 seconds was not something I found interesting.

Submitted by greyhat on

I remember a dev talking about the hunter class's original power resource (before mana) - it was generated by standing still. They dropped it because it simply wasn't fun or interesting. Rune of power is the exact same: not fun, and not interesting. I don't see why they recognized it before, but not this time.

Submitted by Lhivera on

I had a brief exchange with a dev about it this evening, and found his position compelling. Basically:

  • It no longer hits cc'd targets, eliminating one of the primary complaints (this goes for glyphed Ice Lance as well, and I expect probably all chained damage spells).
  • In questing, if this encourages you to pull single mobs away from groups when you're worried about handline multiple targets, that's a good thing, not a bad thing.
  • In raids, the type of situation where you really need to avoid hitting secondary targets (that aren't cc'd) is very rare. Dealing with such situations as an occasional challenge, as opposed to doing so as a constant frustration, is good gameplay.

I'm pretty well convinced that the design is fine as-is.

Submitted by Nathyiel on

Some reflexions ...

There's some problem with the Water Elemental. When I jump out of fire (fire is hot), I go just outside and WE following me. But I'm not far enough of the fire, the WE will slack in it with all bad conclusion. And the healing with Frostbolt is just a pity.

There's  aproblem between haste and Frost Bomb. A good example is the Madness fight. I have obtain a cooldown of 4.6s. And in general, it's begin to be problematic when the CD fall under 8s and is difficult to follow under 6s. If it's will not be problematic in T14, it will be for T15/T16. Some easy solution can be to fix the CD to 10s and to upgrade th damage with Haste like it's now for Mana gem.

Another think is the pet size: they're huge. WE isn't exception, especially if glyphed. All melee complain and it's a problem for clicking. 

On an interesting not, Frost can pull very huge AOE burst (like 100k+ DPS in 10s on small pack) with just Frozen Orb+Frost bomb+Ice Lance. I haven't found how to keep it going whan Frozen Orb is on cooldown.

The single-target rotation is pretty good more predictable with Frozen bomb, more random with Living Bomb and nervous with Nether Tempest. When Frozen Orb/Freeze is used, the game-play is very bursty. In between, you only spam Frostball and bomb/FFB with only rare Ice Lance. It's not bad.

I think that Frostbolt debuff can be better maybe with a Early Frost like mechanics but it don't need a lot.

My final reflexion will be for the Icy Veins accelerated casting animation. I still think that it can be better with a single cast animation and 3 followed bolt. Because actually, it's pretty ugly to the point that under Lust I only have one cast animation and one bolt.


Post-scriptum: like always, you did a great job.

Submitted by Circuitry on

I wholeheartedly agree on this, I've had a lot of trouble switching to raid members when I got to cast my Pet Freeze in 25s, it seems unecessarily challenging.

Submitted by Nathyiel on

A quick question about the interaction between glyphed Icy Veins and trinket/enchant proc.

The glyphed IV split the damage of some spell in 3.

Can Trinket & Enchant proc from each mini-bolt or only on succefull casting, like FoF ?

And is it normal that the frostbolt debuff is only applied once ?

In short, is the glyph changing only the spell grafics and damage calculation ?

Submitted by Lhivera on

Though each of the three bolts has its own chance to hit/miss/crit and roll damage, only the first has a chance to proc effects. I'll update this in the Compendium.

Submitted by Sw1tch on

I haven't been following the recent updates, so I'm still clueless as to this. Is RoP a viable option over evocation now? AKA, does the 15% constantly (assuming perfect uptime for conenvience sake) stay on par with the 25% from evocation whose only downside is casting for a couple seconds?

In general, does the 10% loss equal the loss from not damaging for a couple seconds every 40 seconds?

I really want to use RoP, I just want it to be viable as well.

Submitted by Citizenpete on

Hey Lhivera,

I have a strange issue regarding my own rotation:

Tonight I realized, that I´m playing different to what all the guides and simcraft say:

Priority example  (Simcraft):

Ice Lance <2 Stacks

Frostfire Bolt

Ice Lance

Frostbolt (Filler)

Regarding the simulationcraft action priority list, the following case could happen:

Ice Lance, Ice Lance, Frostfire Bolt

I would never do that! I only would execute Ice Lance or Frostfire Bolt during Frostbolt (Frost Bomb) cast (chaining)

Frostbolt, Ice Lance, Frostbolt, Ice Lance, Frostbolt, Frostfire Bolt

My question is: Is my playing style a possible improvement, deterioration or doesn`t matter?