Dev Watercooler -- Threat Level Midnight

Threat revisited

One of the fun things about working on an MMO is that the game design will evolve over time, and you have the opportunity to make changes to reflect those design shifts. (And yes, we know that it can sometimes evolve too quickly).

Back in December, I wrote a blog post about our vision for how threat should work. Since then, the game and the community have continued to progress and the designers have found ourselves changing our minds about the role of threat. Enough that we’re planning to apply a hotfix this week to change how threat works.

Why have threat?

Threat’s role, just so we’re all on the same page, is to make fights more interesting. Tanks spend a lot of effort staying alive, but they aren’t under immediate threat of death one-hundred percent of the time. Plus, their staying alive is also dependent on their healers and other external cooldowns. We have always been concerned that if threat was not a big part of tanking gameplay that tanks might get bored just waiting around until it was time to use a cooldown. Likewise, if DPS and healers had no risk of being attacked themselves then the sense of danger facing a powerful creature could erode. Furthermore, every character’s toolbox includes some cool survival and utility abilities and the game feels more shallow if those are exclusively used for PvP. It’s fun for a mage to Frost Nova an attacker and Blink away. It’s fun for a hunter to Feign Death. Yes your life would be a lot easier without threat mechanics, but our goal isn’t to make fights as easy as possible. Our job is to make fights fun. Having too much to manage might not be fun, but it’s also not fun to be bored.

That’s been our traditional argument for threat needing to matter. Here is the case against it:

Why not have threat?

Throttling

• As I said in the previous blog post, it’s not fun to feel throttled. It’s not fun for the Feral druid to stop using special attacks in order to avoid pulling aggro. It’s fun to use Feint at the right time to avoid dying, but it’s not fun for Feint to be part of your rotational cooldown. We want you to spend most of your effort trying to overcome the dragon or elemental, not struggling against your own tank.

Tanks are busy

• I’d also argue that our encounters aren’t really boring these days. We ask tanks to do a lot -- everything from picking up adds, to moving bosses around, to staying out of fires, to providing interrupts, in addition to the classic tank roles of staying alive and generating threat.

Threat stats aren’t fun

• We put threat stats (hit and expertise for the most part) on tanking gear, because without those, tanks would be limited to choosing from among mastery, dodge, and parry. (In the current state of itemization, you are rarely choosing more Strength, Agility, Stamina, or armor.) Druids can’t parry, and even for the plate users, there is a tight relationship between dodge and parry, and even mastery for the warrior and paladin. That gets us dangerously close to the old model of stacking a single uber stat (like Stamina or defense), which makes gearing choices too simplistic for tanks. Did something drop? Okay, put it on. (Contrast this to a DPS caster who might want more or less hit or might favor haste over crit, etc.)  
  
  We want threat stats to be interesting, but the reality is that they aren’t. Any decent tank will usually choose survivability stats over threat stats. Back in the day when taunts and interrupts could miss, you could argue hit was marginally useful. But in a world where hit is really just for generating threat, it isn’t very exciting and tanks get understandably emo when we put too much on their gear. (DKs are somewhat of an exception in a good way -- more on that in a sec.) We do see some players try and get excited about threat stats or even proud of their ability to generate threat, but overall we feel like threat stats are a trap, and it’s usually the case that improving your survivability will have a better net impact on your group’s progression.

We don’t need a more complex UI

•  We have threatened for years (see what I did there?) to build in some kind of threat tracking tool into WoW. But is that really good for the game? Do we really need yet another UI element for players to look at instead of looking at the actual game world? We know many raiders in particular use third-party threat mods today, but that has really been borne out of necessity rather than a sense that watching threat is super compelling gameplay. (When we say “super compelling gameplay” you can mentally replace that with “fun.”)

Dungeon Finder

•  I know this bullet will be a point made by players critical of this change, but I would feel remiss in not bringing it up. We want it to be a positive experience when Dungeon Finder matches experienced players with newer players. The skill and gear of the former can help make up for that of the latter. Who better to teach you boss mechanics than players who have done the fights before? Even better, the gear of a veteran tank can make up for the less powerful gear of a beginning healer (which doesn’t necessarily mean a noob -- it could be the alt of a very experienced raider).
  
  However, this system fails and often spectacularly so when it’s the tank who is the undergeared player. Even if a competent healer can keep the undergeared tank alive, the fully raid-geared DPS spec is going to constantly be on the verge of pulling threat. That’s not an issue of skill. It’s just numbers. It’s also not a problem that is easy to overcome for either the overgeared DPS or the undergeared tank -- it’s just not a lot of fun for anyone.

So now what?

Given all of that, and watching how tanking has unfolded in Cataclysm, we’ve gotten over the concept that threat needs to be a major part of PvE gameplay. We have therefore decided to buff tank threat generation in a hotfix this week to where it’s generally not a major consideration. We expect the community to gradually stop using threat-tracking mods as players realize they don’t need them.

It’s an important distinction that the concept of “aggro” will still exist. If a DPS spec attacks an add the second it shows up, then the creature is going to come at her. However, if a tank gets an attack or two on a target, then the target should stick to the tank. Worrying about who has the creature’s attention should generally only be a concern at the start of a fight or when additional creatures join the battle. Worrying about a warrior or DK (the classes with nearly non-existent threat dumps) creeping up on tank threat after several minutes will almost certainly not be an issue any longer. (And if it is, we’ll have to make further adjustments.)

We like abilities like Misdirect. It’s fun as a hunter to help the tank control targets. We are less enamored of Cower, which is just an ability used often to suppress threat. We like that the mage might have to use Ice Block, Frost Nova, or even Mirror Image to avoid danger. We don’t like the mage having to worry about constantly creeping up on the tank’s threat levels. The notion of aggro (who the target is attacking) is a keeper. The notion of threat races (who is about to pull aggro) is going to be downplayed from here on out.

Upcoming changes

Here are the specific changes you’re likely to see:

• Hotfix: The threat generated by classes in their tanking mode has been increased from three times damage done to five times damage done.
• In an upcoming patch: Vengeance no longer ramps up slowly at the beginning of a fight. Instead, the first melee attack taken generates Vengeance equal to one third of the damage dealt by that attack. As Vengeance updates during the fight, it is always set to at least a third of the damage taken in the last two seconds. It still climbs from that point at the previous rate, still decays at the previous rate, and still cannot exceed the current maximum.

Long-term changes

You could argue that once threat is very easy to manage that a warrior tank could just go AFK. In reality, given today’s boss encounters, an AFK warrior would end up standing in the wrong place, missing a tank transition, or otherwise do something or fail to do something that wipes the party or raid.

That said, we ultimately don’t want tanking to be just standing there soaking boss hits and we would like to have more stats on gear that tanks care about. To solve those challenges, we want to shift more tank mitigation to require active management. We’ll still give all the tanks emergency cooldowns like Shield Wall and Survival Instincts. However, we want to move the shorter cooldowns like Shield Block, Holy Shield and Savage Defense so that they work more like Death Strike. Blood DKs have a lot of control over the survivability they get from Death Strike, but as part of that gameplay, they have to actually hit their target. The other three tanks will get similar active defense mechanics. This doesn’t mean everyone needs to use the DK model of self-healing, but they can use the DK model of managing resources to maximize survivability.

Death Strike consumes resources to help the tank survive. We toyed at one point with the paladin Holy Shield being a Holy Power consumer and we think we could do so again. Heck we could make Word of Glory the thing you’re supposed to do with Holy Power, so long as we balanced all tanks around that idea and didn’t feel it infringed too much on the DK mechanic. We could make Shield Block cost rage, and change Protection warrior rage income such that they had to manage rage, the way Fury and Arms warriors now must do. If tanks generated more rage from doing damage and less from taking damage, then hitting a target becomes very important, but for mitigation, not threat management reasons. This is a bigger change than it seems though. We don’t want a model where the Prot warrior ignores Shield Slam, Devastate and Revenge (since threat isn’t a big deal) in order to bank all rage for Shield Block (because survival is). Imagine a rage model where you always had enough rage for your core rotational abilities (they could be cheap or even generate rage), so that you could funnel most of your rage into Shield Block when survival mattered and Heroic Strike when it did not. Redesigning Savage Defense to make it a rage sink is an even bigger change, but we think there is an opportunity there to make the rotation more interesting for druids (and all tanks really). Their rotation would help them achieve the goal that usually matters the most to tanks: living.

This is the kind of design for which we’re really going to need a lot of feedback once it hits. We can implement and verify empirically how much threat a tank generates, but it’s hard for us to replicate the experience of all of the various raiding groups and dungeon parties out there. We invite you to try out the immediate and eventually the long-term changes when they are available and let us know how they feel. Do you miss the threat game? Are you bored when tanking now? Conversely, with the changes, is tanking more fun for you? Does this new implementation of Vengeance feel better? Some systems design calls we can make just by processing numbers, and some are more squishy and involve a lot of gut checks and wishy-washy “but how does it FEEL?” language. Messing with this kind of thing is definitely somewhere in the middle.
 

Greg “Ghostcrawler” Street is the lead systems designer for World of Warcraft, and lead eater at the dinner table.