Blood Pact: Demonology and patch 6.2

After patch 6.2 comes out, we’ll look at how Warlocks are doing in Hellfire Citadel. I could post it now, but we’ve got a massive focus on the nerfs Demonology is receiving and whether those nerfs will be too little or too much. I want to wait to see some more of the Hellfire Citadel fights to see which specs adapt well.

For this week, I want to explore what was already revealed in the patch 6.2 Q&A with Lead Game Designer Ion “Watcher” Hazzikostas — not only that Demonology was too good relatively within the Warlock class, but also how it has design problems. If you want to rewatch the Q&A, the Demonology Warlock 3-minute question & answer occurs at about 38min 15sec in, and Watcher responded after the Q&A with a follow-up.

Demonology sits in the bottom half of the tier 17 simulation — but that’s using Patch 6.2 simulations, not current Patch 6.1 mechanics.

Bias and context

Before I get into looking at numbers, I want to talk for a bit about bias and context. Context is important when judging any set of data, whether it’s output by a program or by a bunch of players. Although the Warlock module is well maintained by Gahddo, and it’s fairly accurate as to how Warlock theorycrafters understand the current and upcoming patches, the numbers and bars in SimulationCraft are done with a certain chunk of assumptions.

Best in slot gear from Mythic is often used to produce the best theoretical results, and action priority lists are performed with zero player mistakes in mind. Simulations are performed on what’s called a Patchwerk test — no moving, no target switching, just a regular tank-and-spank fight — because that’s the simplest DPS test possible. Just outputting SimulationCraft results while including little if any of the context involved with the simulation is misleading.

Player perception is also a big bias when looking at spec performance data across a raiding tier. As Watcher mentioned in the Q&A, you get players who hear that the best spec is Demonology, and then play Demonology because of that, rather than playing the spec they truly enjoyed. So sometimes it appears like a spec is overpowered simply because a lot of people are playing it, so more end up playing it.

Context and bias shape the output of everything, so it’s important to keep both in mind when judging the specs in tier performance.

Just how many Warlocks were playing Demonology?

I keep hearing the same story about how those who switched from a favorite and long-played Affliction or Destruction spec, to a newer and confusing Demonology spec, got better DPS as Demonology than as the other two. But it’s harder to tabulate performance without wanting to take into account skills. We can talk percentiles of DPS using WarcraftLogs, but then we run into the issue of player perception bias. Is Demonology doing better because all the better players are gravitating towards it, or is it naturally just better? We can certainly use theoretical tests like simulations to argue how much of Demonology’s better DPS is due to the spec rather than due to masses of better players playing it.

But DPS isn’t actually the question here. There will always been a best Warlock spec, a middling Warlock spec, and a worst Warlock spec — that’s just how having three damage specs works. We want to know whether the Warlock specs were fairly equal on representation in raids, or whether one spec was severely over-represented.

I borrowed a little spreadsheet maneuver from Cynwise. I headed over to World of Wargraphs, specifically looking at the PvE representations of specs. The yardsticks that World of Wargraphs use are one or more Mythic bosses in both Highmaul and Blackrock Foundry, and global level 100s, all from the data that World of Warcrafts collects from the Blizzard Armory. We see the following for Warlocks in PvE content:

• Affliction Warlocks represent 0.7% of Mythic players, but represent 2.0% of all level 100s.
• Demonology Warlocks represent 3.2% of Mythic players, but represent 2.3% of all level 100s.
• Destruction Warlocks represent 2.8% of Mythic players, but represent 3.3% of all level 100s.

Demonology Warlocks are over-represented. Destruction is under-represented by a smidgen, but Affliction is almost relatively nonexistent.

That’s pulling from the Armory — now, let’s look at numbers of parses. We’ll take a trip to WarcraftLogs, mostly because I can’t find the number of parses on AskMrRobot’s logs. Again, we don’t care about the DPS done, merely the number of Warlocks playing as one spec or another. I’ve tabulated all the numbers of parses per spec per mode in a spreadsheet here. A few conclusions pop out for tier 17:

• Affliction has the least number of parses in all modes for both instances.
• LFR Warlocks prefer Destruction, with over half of all Warlocks playing Destruction.
• Normal Warlocks split fairly evenly between Demonology and Destruction in preference.
• Heroic Warlocks prefer Demonology, with over half of all Warlocks playing Demonology.
• Mythic Warlocks vastly prefer Demonology, with over three-quarters of all Warlocks playing Demonology.
• Overall, half of all Warlocks prefer Demonology, followed by a third playing Destruction, then Affliction.

What the fel — almost 4 in 5 Mythic Warlocks are raiding as Demonology. In most modes, half or more of all Warlocks are playing Demonology.

Even peeking at the DPS numbers doesn’t look good for spec equality, because Demonology stays significantly far, far ahead of the other two specs. Your top-of-the-line 95th percentile Mythic player gets the specs to look closer with Demonology still on top — much like theorycrafters predicted — but let’s look at a more average player. Heroic mode, 50th percentile is a good yardstick for an average but dedicated to raiding player, and there’s not really a debate on Demonology being better there. As you play with the different percentiles, you see a couple trends. The higher the percentile of player, the more Demonology does over Destruction and Affliction. The lower the percentile of player, the closer the middle spec gets to Demonology, but the middle spec never crosses or even touches Demonology outside of the very early first week numbers. Demonology is simply the best, all the time.

Why not buff instead of nerfing?

One of the arguments against nerfing Demonology into the ground asks, why doesn’t Blizzard fix Affliction and Destruction instead? Demonology is doing well, obviously, perhaps it’s the problem of the other two specs.

Except, well, it isn’t the problem of the other two specs. When looking at spec DPS across all bosses and all classes, Demonology is often #1 or #2, no matter which percentile or difficulty mode you’re looking at. There aren’t even fights where Demonology is just the pits at dealing damage — it is good, even very good, at across the instance. When looking at specific fights, perhaps Oregorger and Gruul are where Demonology isn’t top of the line, but in every other fight Demonology is in the top five. That’s overpowered.

We need to take a good look at ourselves and what we want from Demonology as a spec, in not only performance, but ease of learning, ease of mastery, or even just plain spec fantasy. We can’t be the best all the time.

So change Demonology now!

It’s hard enough to learn a new spec to play in a new raid tier because the one you used to play simply isn’t good enough anymore for what content you want to complete. But in that case, at least all the guides to that new spec are out there for you to consume and use. If your spec suddenly changed its mechanics completely in the middle of the expansion? That’s another story. At least after a new expansion begins, everyone is on the same page and gets to relearn their class. The middle of the expansion is not the place to begin relearning a spec; we have enough performance variety in the form of different fights that may or may not bug out in random ways.

Blizzard’s history with new class concepts also tends to start out overpowered, then gets nerfed down, and that wouldn’t solve Demonology’s current problem for at least another patch. It’s best to have the rest of this expansion to think about where the cruxes of the spec design problem are in Demonology, and use the expansion reset to start fresh with a new design aimed at representing, rather than complicating, our core spec abilities.