Thank you all for your fast answers.
I try to refrain speaking about "best" or "worst", but I must confess, that I initially chose "Frost", because it was shown way up in some simulated dps rankings, and because I was advised to use that spec for leveling.
What I do try to find out is: how close is the damage output really between the 3 specs, and what other reasons are there to play a certain spec.
Of course the main reason is: to play a spec which you enjoy the most, while playing, but for some players it might be most enjoyable, to play always the one spec, which does the most damage.
@Akraen you stated:
"I'd never disparage any mage from playing any of the specs, and I think only the top mages would even notice the DPS difference among them. In my 25H guild, we have 1 fire, 1 arcane, 1 frost, and 1 mage who changes specs constantly-- we all trade places for the #1 dps slot."
I would like to hear from the one mage in your raid, who constantly changes specs - how are the DPS results, and why does he change specs? Does he change between the different encounters?
"(...) but raidbots and rankings are not that. You need to check boss kill times, damage on necessary targets, and weigh how much is padding versus effective DPS. It doesn't change the general Arcane>Fire>Frost order, but it does bring it more into a level playing field to take it all with perspective."
Would you mind elaborating, how "boss kill times, damage on necessary targets" and "padding" affect the logs / dps rankings? What do you mean by "padding"?
"It doesn't change the general Arcane>Fire>Frost order, but it does bring it more into a level playing field to take it all with perspective."
At the moment I feel, I can't take that perspective where it is leveled, that for example the top 2 ranked fire mages did >940000 dps on the shamans, the top 2 ranked arcane mages did >775000 dps and the top frost mage did 713682 dps. Would you mind explaining on an example / log or something?
I feel like the best way to answer your questions is to use a similar Q&A style to Akraen's post, so I'll give it a go.
What is wrong with DPS rankings?
DPS rankings don't compare specs meaningfully. The top parses right now for all bosses involve a number of gimmicks that make them useless for comparison.
You mentioned Dark Shamans. Top parses right now involve intentionally leaving grunts alive until you pull, and killing them right after entering combat with the boss. This inflates numbers greatly, while having no meaningful impact on the boss kill. Also, Dark Shamans is a boss fight that starts with 4 mobs, 2 of which are beasts. Top fire mages are often trolls, and take advantage of the 2 dogs for their 5% racial, generate a gigantic combustion and spread it across 4 mobs. This is a special case that can only be done on Dark Shamans, and to a certain extent, protectors.
Similar "number tricks" can be done on many bosses. On Nazgrim, many guilds choose to pull the Ironblades (warrior trash mobs) with the boss. Galakras has a trick for disabling the cannons without pulling trash, which you then kill during the boss fight. On spoils you can open additional small boxes which contain Burial Urns, and you don't kill them, but let them spawn a lot of little sparks that you can attack for numbers.
Top ranks also have serious bias towards spec DPS variance. For an extremely simplified, somewhat misleading picture, supppose mage X has a range of 90K-110K DPS, versus mage Y being 70K-120K DPS. Mage X has a higher average of 100K dps versus mage Y having 95K DPS, but when it comes to top parses, you are more likely to see mage Y on top because of the potential 120K DPS number, which mage X has no chance of reaching.
Why do kill times matter? How do they affect DPS?
A simplified answer is that "The faster you kill the boss, the higher your numbers will be." Similar to what Soggs said, the faster you kill a boss, the higher proportion of the fight you spend under powerful cooldown abilities such as Bloodlust, and Icy Veins/Arcane Power/Combustion. Have you ever tried using Time Warp and cooldowns to kill a quest mob, and notice how your DPS is through the roof because it died nearly instantly? Faster boss kills essentially do the same thing.
What is wrong with Noxxic?
Noxxic mindlessly takes results from Simulationcraft, and generates guides/references from them. They have no experienced players of each class to review the results. This means that their guides don't reflect realistic situations, and depend completely on the accuracy of Simulationcraft's "single target stand and burn DPS" evaluations. This also means that valuable components to raiding, such as DPS during movement, multitarget cleave DPS, AOE DPS, burst damage on priority targets, survival, are all completely ignored by Noxxic. If you look at Noxxic's guides closely, you'll see that they ignore all movement, survival, and utility talents.
Simulationcraft is a great tool, but it is only as accurate as the information you feed it. You mentioned Frost seeming to be the strongest, and it is no surprise to me - Frost's Simulationcraft module has been much better developed than the other two modules, and comes closest to what it can do in real raid situations. This does not reflect real raiding at all.
So what is the BEST DPS SPEC?
The general consensus is that there isn't one. On each fight, there will be certain elements that favor one spec or another, but it also depends on the strategy being utilized by your raid, and tiny details in how they execute it. For example, tanking Fallen Protectors together will favor a mastery-based fire mage due to powerful ignite cleaving, while tanking them separated by a few extra yards will favor a haste-based fire mage.
The only advice we can give, is that you should play what you're comfortable with, and adjust it according to the needs of the group. That will give you the
BEST RESULTS.
