The occasional debate turns up about how well sims work. Not everyone thinks they do a good job of modelling how players will actually perform. So I decided to try a study that, rather than using results that MIGHT happen, instead used results that actually DID happen. How are the classes and specs doing in the current raid environment?
Anyone who enters a thread titled "Wall of Text" and expects a TL:DR version will be sadly mistaken. However, for such people that are looking for they key points, you might find the red text to be useful.
So first, this is how I got my information:
-- I went to worldoflogs.com and looked up Throne of Thunder 10N boss kills. I chose ToT because it was current content. I chose normal mode, because LFR is not the best place to reference for well-played characters, and there wasn't enough information to do a good 10H workup (sample size would be too small)
-- For each fight in 10N, I took each of the 23 DPS specs, and looked at the top 40 highest DPS(e) results from the entire site: American, European, Asian, didn't matter. DPS(e) is total damage divided by the length of the fight, so for example, if you died in ten seconds, your DPS(e) will be horribly poor.
-- I then averaged those results of those 40 players, fight-by-fight.
I think it's fairly safe to say that players who know their class, spec, role, and fight mechanics pretty well. That's why they're on the top. People may joke, but I think everyone knows you can't literally face-roll and do as well the players surveyed. Based on some of the DPS numbers and kill times I saw, they are probably fairly well geared as well. And if they posted on worldoflogs, then it suggests that they, or at least someone in their raid group, are taking their raiding seriously enough to monitor their results. All in all, they may not be the world's best players, but they're almost certainly good players, hopefully good enough to get some results for.
Here is the chart I came up with from the above information:
For each fight, I ranked the classes in order by the average results from that chart. As expected, the rankings had a lot of variability in different fights. That's good. That's to be expected. But then, I averaged the rankings to see which class/spec did the best, or worst, overall.
#1 Overall: Fire Mages
Fire mages were top DPS in two fights (Dark Animus and the Council) and 2nd/3rd in a total of 5 other fights. They were never ranked lower than 8th place on any fight. Fire mages put out an average of 209,747 DPS over the entire raid (all 12 bosses averaged) which is 20% higher than the median.
#2 Overall: Assassin Rogues
The ONLY melee in the top 5, Assassins topped the charts on five fights: Jin-Rokh, Ji'kun, Durumu, Iron Qon and the Twins. It looks like they are either the best, or very close to the best, and ripping single targets to shreds. Their overall DPS was barely a percent lower than Fire Mages. The only reason they came in #2 is, while they were top on the most fights, Fire Mages were more consistently strong, while Assassins slipped lower on a few fights. It was, however, a very close call.
#3 Overall: Affliction Warlocks
Aff. Warlocks came in 1st on Megaera and 2nd or 3rd in six other fights.
#4 Overall: Destruction Warlocks
It is no surprise to see these guys top the charts when the fight is littered with adds. They were #1 on Primordius and Horridon, and #2 on Tortos. Their rank slips, however, to mediocre (12th to 14th) against the Council, Iron Qon, and the Twin Consorts. It is worth noting that Destruction warlocks are the most inconsistent DPS in the study, in that their damage done varies fight-by-fight more than any other class/spec.
#5 Overall: Shadow Priests
Shadow priests were not top DPS in any fight. However, they never got lower than 12th, either. They consistently did better on multi-target fights than single-target ones, as well.
#5th Worst Overall: Feral Cat Druids
Cats had their place, doing better than most on Tortos, Ji-kun and Durumu. Perhaps they are better at fights where OMG THERE'S A BIRD GONNA GET IT GONNA GET IT is the line of thinking? However, they did poorly on all other fights, 14th to 20th.
#4 Worst Overall: Enhancement Shamen
Enh. Shamen did average overall on Dark Animus and it slides downhill quickly from there. They appear to be especially poor on fights with loads of adds, but did pretty poorly in single-target fights as well. Their damage is just overall low.
#3 Worst Overall: Arms Warriors
As an Arms warrior offspec myself, I was not surprised to see this. They came in 8th best on Megaera, a fight where you get to spam Execute 7 times, but that is the only fight where they did better than average. They were either third to last, or second to last, on the last six bosses and Horridon.
#2 Worst Overall: Marksman Hunters
As the only appearance of a ranged DPS in the bottom 5, I am sure MM hunters expected this result. Overall raid DPs, they were at 141k, nearly 22% lower than the median of 181k. Marksmen found themselves second to last on every single fight that Arms warriors weren't second to last, and in those cases, were third to last. They were outperformed on all types of fights, all types of mechanics, flat-out across the board. It should be noted that Beastmastery and Survival hunters actually did ok, being overall average in rank, so it's not the class. It's the spec that did poorly.
#1 Worst Overall: Subtlety Rogues
Subs found themselves dead last in every single fight. Every. Single. Fight. And, to make matters worse, clearly everyone knew it. You know how I said earlier that I took the top 10 DPS from each spec from each fight? That's not entirely true. Starting around Megaera, there weren't results from 40 different Subtlety Rogues getting kills. By the time I was looking at Lei Shen 10N, I had only ten total Subs to work with. Ten. Their averaged damage was also last by a whopping 40% lower than the median. I would still conjecture that those 10 Subs were good players, as they did make it to Lei Shen and they did kill him. But I think it's disturbing when even 10 comparitively good raiders are so dramatically out-damaged by literally every other DPS spec.
Now you might be saying "You cannot judge Subtlety Rogues due to their low sample size". But, I really think you can. I think that competent raiders just don't play Subs in PvE because they know how bad they are. I think raiders who try to raid at least semi-seriously choose not to play this class/spec that's so god-awful at PvE, and quite frankly, I think that's horrible. And you can't even say "it's because Sub is the PvP spec, just like Arms warriors who also suck" but I don't agree with that, either. Frost mages are one of the top, if not THE top PvP spec and they're doing just fine (not tops, but fine, 10th overall). Hell there are a ton of Shadow Priests and Destro Warlocks with 2200+ too and they're in the top 5 in this study. What's the excuse here?
"You shouldn't balance classes based on how the best players perform!"
Maybe not. But don't forget, these aren't world-first here. These are 10N results, not 10H. And, while basing tunes on only the top players might not be a good idea, basing tunes on players who are playing well for that class/spec is a fair decision. If a MM hunter, even playing to the best of their ability, is routinely being out-performed by everyone else, I think it's time to revisit that class for upgrades.
"It's not a competition!"
I admit, there are players who stick with their character because they like the feel for it. Fair enough, and more power to them. I still play Arms, after all. Nor am I insisting that every raid boss be killable using only low-ranked melee like Cat druids and Enh Shamen. But at some point, to kill a boss, you need a certain amount of damage in a fixed time. If one of your raiders, despite being a strong player in general and better than average at his class/spec, is not bringing their fair share to the raid, and progression is stalling, then decisions have to be made. And if those decisions involve kicking good players, I think there's a problem.
"Could this be gear scaling?"
Quite possibly, yes. I didn't look at the Armory of the literally hundreds upon hundreds of players involved. But I would also add that, if it was gear scaling that caused the differences, that that's still an issue worth looking into.
"Someone has to be last."
I agree. But that someone doesn't have to be last on every single fight, nor do they have to be last by such a wide margin, in a game that's supposed to have some kind of balance. Based on their consistently low ranks, I think MMs and Subs have proven they are in need of help.
"It is ok for some specs to be really bad. Dual spec takes care of that."
True, everyone has an option, but that option is not always damage. MM hunters can switch to either Surv or BM and do ok. I could switch my Arms spec to Fury spec and potentially even do better than average. Windwalker monks (6th worst overall) and retadins (7th worst overall) do not have an alternate DPS spec, and will be forced to either do worse than average because of their class/spec, or else tank/heal. Feral Cats and Enh Shamen would both have to switched to ranged to continue to fill their role. I don't think forcing someone to have a drastically different playstyle and fill a different role is the answer.
"5.3 will fix some of this"
Good! It is because 5.3 is still under development that I thought I'd make and post this. There is still time for changes to be made, but only if people are aware of the problems, when they exist.
MELEE vs RANGED
You've already seen that 4/5 of the top 5 are ranged and 4/5 of the bottom 5 are melee. But there's more to the story, actually. I averaged all the melee and all the ranged dps into "melee" and "ranged" and compared the results. Overall, there was an 8.91% total average DPS of ranged over melee, which shrunk to a 5.22% gap if you excluded the Subtlety Rogues. If you looked for fights where melee and ranged were within 5% of each other, you'd find 6/12 fights, that happened, which includes the final 3 bosses. Of the remaining six, melee did better than ranged on Ji-kun and ranged did better than melee in Horridon, the Council, Megaera, Primordius, and Dark Animus.
The biggest concern I have here is that, when ranged did outperform melee, it was often by a LOT. Ranged were overall better than melee on the Council of Elders, Primordius and Dark Animus by 18 to 20 percent each, which is a pretty big gap. By contrast, in the Ji-kun fight, melee outperformed ranged by only 6.73%.
I know that, generally speaking, ranged will probably always out-perform melee and that different fights will favor different people. But I would think that "bring the player, not the class" would be better served if the gaps were either tighter, or not so one-sided, more often than one-quarter of raid bosses.
THE MONSTERS THEMSELVES
The raid bosses with the lowest damage spread, by which I mean, the differences in DPS between the different specs was the smallest, were Iron Qon, the Twin Consorts, and oddly enough, Jin-rokh. In those three fights, two-thirds of all the measured players were in a spread of 18%.
The raid bosses with the highest damage spread, in which different classes/specs performed the most differently, were the Council of Elders, Primordius, and the Dark Animus. Two-thirds of all measured DPS were in a spread of about 40% or more. In these fights, "bring the player not the class" was the most demonstrably incorrect. And those of you paying attention note that these are the same fights for which ranged drastically out-performed melee.