oh boy, well I was hoping this wouldn’t have to be my soapbox, but I guess I might as well explain how warcraftlogs (and to a more relevant example, where blizzard tuning) arrives at its metrics. Before I start, I will say that I’m in no way saying that we’re fine, in fact warlock is underperforming in almost every metric (dps, survivability, pvp, hps) and should be tuned higher, but in no way is 40% ever realistic.
I am using mythic data because it is more representative of people that are able to play their class effectively and at a high-level. Unsurprisingly, heroic data is actually much, much closer on average level than mythic data (see here (https://www.warcraftlogs.com/zone/statistics/38#aggregate=amount&difficulty=4).
Wall of text inc.
First off lets get the story straight, usually for tuning at the upper-end limits, blizz wants to increase the floor of the average percentile of throughput in relation to all relative classes; This is accomplished a couple different ways:
- You can take the overall metric of dps ON AVERAGE over the span of all raid encounters and make a judgement as to the overall aura damage that must be increased to bring averages closer together. For example, if you look at this chart, upper end thresholds are not taken into consideration as averages are more easily attainable targets, and talent tuning is left to major patch content (anything that is a .5, .7 etc.):
(https://www.warcraftlogs.com/zone/statistics/38#) - Internal metrics on dps are slightly different based on context and factors that designers consider when creating talents and elaborating on the niche for each spec. There are ways they accomplish tuning by reducing major key talents that are outperforming their desired throughput based on fights or overall (see recent lightning rod nerf, which was targeted, but in no way is the complete tuning for elemental). I would not be surprised if affliction got specific context tuning for single target WITH aura buffs to incentivize people to play it higher-end mythic encounters this coming reset, or whenever they make tuning adjustments.
ANALYSIS
Typically, warcraftlogs aggregates its data based on a couple different factors:
- Normalized scores
(https://www.warcraftlogs.com/zone/statistics/38#) - Overall aggregated data that shows dps metrics on an average level
(https://www.warcraftlogs.com/zone/statistics/38#aggregate=amount)
Both of these examples are using normalized (as in averaged out based on all percentile ranges) of the expected dps outcome of classes. In no circumstance is tuning EVER related to upper thresholds as these are extreme outliers. This is also the case for anyone that uses a simulation software that aggregates the thousands of simulations to create a median dps throughput that it spits out at you. - Per boss metrics are usually the best case scenario in which your classes niche can be harnessed and put into context. Usually any class or spec that is best on every fight is severely overtuned, there is no class that fits this at all at the moment, not even elemental which has extreme funnel damage, which is in itself niche (although more relevant this tier).
FURTHERMORE there are a couple contextual variables to consider in these issues:
- Class representation for pure dps classes is always skewed to whichever spec is performing higher in any given circumstance. For warlock this is denoted by destruction being the better spec for progression, and demonology surpassing the other specs in single target fights. This means that stronger players are inherently increasing the normalized data for both these specs higher than affliction to some degree.
- Simulation data is placing our specs in relatively close proximity to one another, aside from demonology which is over 5% ahead in raw throughput if you can match the expected dps output from the sim.
What does this mean
Taking a look at the spell data from the sims, as well as the context within the raid, warcraftlogs has a decent representation of where we are at an “all bosses context”. You take the average aggregated data from the best non-exploit spec (elemental) and then divide it by the aggregated data of affliction. To most nuanced TC people, obviously this way of looking at buffs/nerfs is rudementary and misses context, but unfortunately this is how blizzard has historically incentivized participation rates in classes and specs.
This nets out to 1,345,983/1,121,749=1.20…
In a nutshell, ele is technically outperforming aff by roughly 20% (IN AN AVERAGE CONTEXT) in raid.
HOWEVER
Given the context above, I’m pretty confident in saying that affliction is actually underperforming the metrics that are given on an overall basis and that the threshold would be lower by X%. In my mind, I’m pretty confident in saying a +/- of 5% after doing all the fights on mythic, and focusing on the niche that affliction has (which is single target funnel, albeit more variable than other ranged).
TLDR:
No, our spec does not need anywhere close to a 40% increase to “beat other specs” even if you were take raw throughput data on average. We would be over 20% stronger than the highest spec if you took into account no context and used aggregated data at this current moment.
If anybody wants to chime in or back up their end with some data, I’m all ears.