Appendix B: The Problem of Evil

I’ve been a bit overwhelmed by the responses on the Decline and Fall of Warlocks in Cataclysm series. There are a lot of questions and points raised which modify my original outline to the point where it the topics need to adjust to suit. “No plan survives first contact with the enemy” and all that.

As such, I have found myself writing posts which I think are related to the main thesis, but where the data doesn’t strongly support the central theme. Or, like in the following piece, I think the data is far more interesting for analyzing other classes than it is for Warlocks. I had hoped that analyzing the initial lack of appeal for Warlocks would yield some new insight into the class, but ultimately there are limitations on the publicly available data which outright prevent it.

This is a data-heavy post, which is pretty much the only reason I’m releasing it into the wild. All of the charts can be found in the 4.3.3 class distribution spreadsheet. If nothing else, this will hopefully prove a solid snapshot of class popularity at the end of an expansion.

APPENDIX B: THE PROBLEM OF EVIL

One common theory I’ve heard about the scarcity of Warlocks is that it’s because they’re the evil class of Warcraft. They personify the ends justifying the means. They rationalize using the tools of the Burning Legion against everyone, without worrying that perhaps this really isn’t a very good idea.

There are NO noble or virtuous Warlocks in lore, and Warlock characters aren’t portrayed as good or nice in the World of Warcraft. At best, no one trusts them, and at worst, they’re feared and reviled – and rightfully so! They consort with demons for fun and profit. They take delight in pain and torment. They are corruptors of the highest order.

And yet, they can also be the saviors of Azeroth.

Reconciling these two positions can, frankly, require some mental gymnastics.

Anti-heroes are a tough sell in a fantasy setting. Yes, there’s some appeal for those who don’t want to play a noble paragon or protector of the natural order, but it’s a limited appeal. A class based on the worst villains of WoW isn’t going to feed into people’s desire to be a hero.

I think there’s something to this idea that Warlocks are naturally an unpopular class at the character selection screen because they’re the bad guys. Even Rogues – the other anti-hero class – are a bit easier to recast in a heroic light. Sure, they’re ruthless and efficient, but you can picture them as secret government operatives, swashbuckling pirates, street urchins turned heroes. The noble Rogue is part of fantasy archetypes like The Gray Mouser or Bilbo Baggins; it’s part and parcel of the AD&D-inspired syncretic fantasy legacy Warcraft is heir to.

Warlocks are either necromancers, crazy conjurers, or wizards who crossed the line with the Dark Arts. They have their own place in a fantasy setting, but not as heroes. So it’s difficult, at character creation, to see how this character would appeal to a broad base of players.

The central questions of the previous posts in this series were concerned with the decline of the Warlock class over the course of Cataclysm. The revocation of the Simplicity Tax and additional complexity beyond the magic number introduced in Cataclysm created a situation of inelegant complexity without reward, which in turn led to a decline in Warlock popularity. These are based upon the significant changes to the class during this expansion.

The reason which I haven’t considered that the class’s character unduly affected it in Cataclysm is because I didn’t see any real change in the portrayal of the class to account for its decline. In other words – Warlocks didn’t get any worse in Cataclysm’s story. Warlocks still don’t have any sympathetic characters in Warcraft lore. All major Warlocks characters are unrepentant villains (Ner’zhul, Gul’dan) or they reform and renounce their fel ways (Drek’thar). None of this changed in Cataclysm.

That said, while I don’t think that the idea of the class caused the Warlock to decline, I absolutely agree that it doesn’t help its case to become popular.

Sadly, the data we have publicly available is limited and doesn’t let us look at things like: out of every character rolled, how many people choose a Warlock? How much time do players spend considering the class on the creation screen versus other classes?

The majority of the data we’ve considered so far has focused upon the Warlock class at endgame, level 85. It’s tempting to use popularity data from the leveling brackets (c.f. the second post in this series) to try to prove this point that players don’t choose Warlocks at the creation screen. Looking at the leveling graph again:

The important data point is in the 10-19 bracket, where players have gotten over the level 10 hurdle and are showing enough interest in the game to commit to more than an hour or two.¹ Warlocks show up with a resounding 21% deficit, and it gets worse from there. It’s safe to say that they’re not popular at creation, and that points to the class not having immediate appeal. They’re the house in the nice neighborhood which lacks curb appeal.

But what’s interesting is that they’re about as (un)popular as Priests, and more popular than Shaman from 10-19! Heck, even Paladins – the most popular class in the game – are unpopular at that level, though they quickly make up ground. If it’s the Warlock’s evilness which dooms them to unpopularity, why are traditionally heroic classes also unpopular?

That doesn’t make much sense, does it?

Well, no, it doesn’t really.

The data in the 10-19 bracket represents the current number of characters in that bracket, not the sum of all characters who have passed through that bracket. It’s correct to say that Shaman are not very popular in 10-19, but that might be because they’re all moving on quickly to other levels.² The shape of the line, and relative position, and how it ends up is more important for analysis than any specific intrabracket comparison.

Let’s look at this by directly comparing Warlocks to another popular class as they level.

Both classes experience a decline between their 30s-60s, but Paladins gain in popularity as they level. They’re almost twice as populous as Warlocks in Northrend.

When you hit Cataclysm:

Man, look at that hockey stick with the Paladin line! People love playing their Pallys at endgame.

Here are the numbers, which have also been added to the class distribution spreadsheet:

It’s tough to say, definitively, that people aren’t rolling Warlocks only because they’re the evil class with this data. It’s really tempting to say that because there’s such a small difference between Pallys and Warlocks in the lower levels, it must be because they like them equally. This is probably false.

The key isn’t in the 10-19 bracket and where they start – it’s with the overall population of each class. Character creation encompasses all characters, not just a subset of those leveling. What we see with the above graphs is that players enjoy leveling Paladins to the endgame and playing them there. They’re not rolling other alts, they level up and they play them. This is true to a much lesser degree with Warlocks, too – but the overall population of people who rolled a Paladin is much greater than the population of people who’ve rolled a Warlock.

In previous posts, I focused on endgame statistics – how many Warlocks were being brought to Heroic Raids, how many were getting 2200+ Arena ratings – which required focus on endgame, level 85 data. Determining if a class is over or underrepresented in Heroics/2200+ required a comparison between characters who could compete in an activity (because they were the correct level) and those who actually did. If we’d compared the total raiding population of a class to the entire class population, we’d draw the wrong conclusion. That’s not right.

So, let’s at total population data.³

Going back to the same source as the leveling data (Warcraft Realms), we get the following:

And let’s look at that as a graph:

There are some surprising results when you compare this data to the level 85 data for other classes – Hunters, Rogues and Shaman especially – but it’s not quite so revealing for Warlocks.

They’re still at the bottom of the barrel. This shouldn’t be surprising – leveling data puts them at low popularity, raiding data puts them scarce – but it’s good to see the data match up from different sources.

Let’s go ahead and compare the total population to the endgame populations.

This data is really interesting for other classes, but not really for Warlocks.

There are a few items to note:

• Hunters make up 12.3% of the overall population, #2 behind Paladins. Their leveling popularity translates into a sizable active character pool, but a comparably scarce population at endgame.
• Like Hunters, Rogues are more likely to be leveling than found at 85. This is probably due to the Legendary Carrot Effect. I think we’ll need to see Rogue data at different times to see if this really held true for Cataclysm.
• Shaman are more likely to be 85 than leveling.
• The massive DK population in the 50s and 60s is statistically a blip due to low populations at those levels.⁴

This, sadly, doesn’t really shed a lot of light on the question we’re trying to answer – is the reason for the class’s unpopularity because of their reputation as evil spellcasters, or because of other reasons?

We don’t really know. All we can say with confidence is that Warlocks are unpopular from start to finish.

There are a few ways we could test the theory of evil:

• Add a major sympathetic Warlock character to lore in the middle of an expansion (to isolate it from expansion-basedclass changes) and see if popularity rises.
• Conversely, add more negative Warlocks to lore and see if class popularity falls. Or rises! It all depends on the character. (…I do not really recommend this.)
• Change the introductory class text and reskin the class to be a Friendship Wizard. See if people reroll to play with Rainbow Bolts.⁵
• Survey Warcraft players to determine why they did, or did not, roll a Warlock.

Aside from those suggestions, I think this specific theory of evil driving players away is unprovable with existing data. It’s suggestive, and there’s an argument to be made for it.

But we need more evidence.

—-

(1) This data set is limited in a lot of ways – unfortunately we can’t see how many Warlocks really get rolled at the character screen. We’ll have to extrapolate initial character creation choice from the lowest level bracket, which isn’t perfect – but we’ll make due. ↑

(2) Scroll of Resurrection toons went through it REALLY FAST.↑

(3) This doesn’t represent characters from all active subscribers, but a representative subset. When you look at the population of active characters across tracked realms, the sample size is statistically significant, so we can make due with it. This is one reason why I tend to deal with percentages instead of absolute character counts in this series.↑

(4) This shouldn’t really be a surprise. Death Knights are well represented in the endgame, but their status as a “Hero Class” makes them ideal bankers on new servers. At least a life of service to Auctioneer Jaxon is a far better fate than serving the Lich King? ↑

(5) Mark my words, Friendship Wizards are going to dominate the DPS charts in the expansion after Mists. ↑