Why I Hate Hate

January 15, 2008 at 1:28 pm (Brainstorming, Game Design, Game Mechanics) (, , , , )

There is a particular game mechanic that is now used in almost all MMORPGs (and a few other game besides) for controlling group combat.  It’s variously called “aggro,” or “threat,” or “hate,” and it’s what makes a monster attack one player rather than another.

And it’s holding the genre back.

If you’re reading a blog on game design, you probably already know how this works, but I’ll explain it anyway.  Each monster in the game keeps track of all the players (and sometimes other creatures) that he doesn’t like.  Moreover, he quantifies how much he dislikes each of them: he tracks every bit of damage you deal to him (and maybe to his friends), every negative effect you use against him, every good turn you do to any of his enemies (like healing them–or yourself, if you’re his enemy), and anything else you do that could be construed as hostile.  These numbers get fed into a giant equation that tells the game exactly how much he dislikes every particular person around him, and he attacks whomever causes him the greatest angst.

So, first of all, allow me to observe that this is a fairly pathetic attempt at artificial intelligence.  In the typical case, it’s probably better than attacking a random foe–but not by much.  And it’s more exploitable, too.  It arguably might reflect the way we expect certain dumb beasts to react, but it’s awfully simplistic even for that, and it certainly doesn’t reflect an opponent with any strategic intent.  It completely ignores teamwork (both for itself and its opponents), risk/reward, and even the relative hardness of different targets (like who’s wearing heavier armor, and who’s low on health).  This represents a foe who is acting entirely on impulse, without the slightest rational thought.

But that’s not the main problem here.

The real problem is that the games take this a step further, and actually encourage players to exploit the weaknesses of this AI.  In fact, many games present players with a wide selection of abilities designed specifically to manipulate this system.  There will be “taunt” abilities that make monsters angrier at you without actually harming them in any way.  There will be “hate-reducers” that encourage the monsters to go after someone else for no apparent reason.  The “tanks,” whose job it is to defend the softer members of the group from enemy attacks, often rely entirely on manipulating the hate system to do their job.  The designers no longer have the slightest belief that the monster is making a good targeting decision–in fact, the design assumption is that the monster will usually be manipulated into making the worst possible decision.

Do you see what has happened?  What doubtless started out as a quick AI hack has become enshrined as a primary game mechanic.  It can no longer be removed from the game, because other parts of the game aren’t balanced without it–some don’t even make sense without it.  If you changed the AI in one of these games, many abilities would no longer make sense, tanks couldn’t perform their role, and existing gameplay would fall apart.  Even if you eventually put it back together, the strategy of the game would be completely different (not that different is bad, but that’s an indication of the level of dependency here).

Now, if ever anyone working on World of WarCraft (or EverQuest, or whatever) decides they want to replace the AI with something else–even in a narrow scope–they can’t.  Not without throwing ordinary gameplay and game balance out the window.  Want to have a special boss that’s smarter than a regular monster?  Can’t do that.  Want to add some PvP?  Enemy players aren’t going to respond to taunts.  Want to create group monster behaviors for large-scale combat?  The players’ hard-earned abilities are broken.  Once you’ve exposed the inner workings of your AI and let players directly mess with its internal state, you’ve lost modularity–the rest of the system doesn’t work unless you keep that AI just as it is.

That doesn’t mean that these games don’t occasionally put in PvP or smarter monsters–but when they do, ordinary gameplay rules don’t apply, and many character abilities and roles around which the game is balanced no longer work.

And the fact that PvP is automatically unbalanced, and that I can never face an opponent with even a modicrum of strategic ability, is quite frustrating.  These games have limited their gameplay to a narrow, shallow, time-worn trail in order to keep using this one mechanic that wasn’t even a good idea in the first place.

It’s crappy for balance, too.  The frail “back row” characters’ strategy supposedly consists of being as effective as possible while keeping their hate below a certain magical threshold.  Creep right up to the edge of that threshold, and you’re playing perfectly.  Set one toe beyond that line, and suddenly the battle’s going badly.  The effects of your actions are extremely unstable.  There’s no flexibility or forgiveness built into the system, which means the difference between perfect gameplay and a rookie mistake is one tiny error.  The fact that you can’t actually see how much hate you’re generating or what the threshold really is certainly doesn’t help this, but even if you could, that’s still a stupid, pointless, and frustrating reason to lose a battle.

And for all its simplicity, this system doesn’t even succeed in being obvious or intuitive.  A new player is not naturally going to infer that he can help his team win by spending less time attacking the monsters and more time twiddling his thumbs.  There’s no rational thought behind that.  This isn’t even a rough approximation of either realism or strategy; it’s a secret voodoo formula that only makes sense to the initiated.


So what should these games do instead?  Am I advocating that we overturn all the precepts of MMO gameplay?  Abolish the tank?  Remove the reasons for attacking one player over another?  Make everyone the same?

Of course not.  But the gameplay shouldn’t revolve around playing your opponents for utter fools, either.  It shouldn’t require you to dance on the knife’s edge of disaster.  And it shouldn’t require you to be an expert on the secret inner workings of the game in order to understand the basic principles of strategy.

A balanced game has stable gameplay when all parties are trying to win.  The standard model of hate doesn’t allow the gameplay to remain stable when you face any opponent except one that you can jerk around like a wooden puppet.

Here are some more flexible options:

Battle Formations.  The distinction between front-row soldiers and ranged support wasn’t invented by MMORPGs, you know.  In real life, it works because the front row is physically between you and the enemy back row, and trying to cross that line entails anything from exposing your back to retributive attack to physically pushing your way through a sea of bodies, depending on the circumstances.  Add in rules like player collision, flanking, backstabbing (not just for rogues) and other such things, and it should be possible to build a system that naturally favors a front row/back row distinction.

Targeting Disincentives.  Of course, video games don’t have as much detail as real life.  Positions and postures are less precise, detailed collision models are computationally expensive, players tend to have less situational and spatial awareness than real life fighters, and so forth.  If you take the same basic idea and abstract away the physical realities, you end up with a system of disincentives.  Tanks stand next to back-row characters and activate some ability that causes nearby allies to take less damage, diverts damage from them to himself, or somehow penalizes enemies who attack the allies under his protection (e.g. counter-attacks).  Set it up right and you can make it so that attacking the enemy tank is tactically sound, even if your game doesn’t use the concept of position or distance at all.  This abstraction also makes it easier to create magically-themed and long-range tanks, should you find that idea appealing.

Targeting Incentives.  You can also do the reverse–generate good reasons to attack the tanks, independent of the other targets.  For example, give the tanks bonus damage when they’re not being attacked, or strong but easily-interruptible abilities, or let them cast debuffs that can be mitigated by attacking the caster.  This doesn’t usually scale as well to different battle sizes (since you need to worry about multiple tanks ganging up on one target), but it can work in at least some situations.

Coercion.  Some people, after hearing me rant about the hate system, assume that I’m against any abilities that coerces your opponent to behave in a certain way, but that’s not true.  Abilities that restrict what actions your opponents are allowed to take–like, say, disallowing the target from attacking anyone other than you–are completely valid (and already exist in most of these games, in the form of “stunning”).  Just make sure that their effects are defined in terms of what the target is or is not allowed to do, rather than in terms of how much the target wants to do something, and you can apply it to any AI you want, or even enemy players, without difficulty.  You’ll need to play with durations or intensities or something in order to keep the effects of these abilities smooth, rather than creating sudden breakpoints as in the hate system (stunning already has this problem), but that’s certainly a solvable issue.

Debuffs.  You can also just let the tank do their job of protecting the party by using abilities that directly reduce the enemies’ offensive capabilities.  In this case, the “tank” might no longer be called a “tank,” and he may not be the one who gets attacked, but that’s actually one of the most appealing things about this option:  you can balance the game so that it is completely tactically viable to attack the “tanks” (debuffers) and completely tactically viable to attack other members of the party.  Killing the debuffers first gets you back to full offensive capabilities quickly, which may or may not be your highest priority, depending on the situation.  Behold!  Strategic variability!

There’s no single option here that’s clearly the best route, but there’s no shortage of possible replacements for the current standard.  I think people continue to use the hate system because it’s easy to implement, and it allows players to feel smarter than the AI, but mostly they’re familiar with it, and players are familiar with it, and to do something new and different entails effort and risk.  But this system has kept games bound to a cheap, transparent hack of an AI for years, as it becomes more and more inappropriate, and its continued use is blocking much of the interesting potential of these games.  For the continued improvement of such games, this model has to die.



  1. N'Dak said,

    This is why when Star Wars: Galaxies first came out, it was amazing. You could do different attacks that prevented your opponent (NPC or PC) from using certain moves. The AI of the mobs was good enough to where it would run after anyone in the group and it was up to everyone to keep the enemy NPC on the tank; which could be almost ANY class. Then WoW came out and there was the combat upgrade where they reworked the entire combat system to make it like WoW. That’s when I quit.

  2. spellman23 said,

    Yay. Thanks for the solutions. I’m partial to formation issues myself, but that’s probably why I tend to play FPS and RTS games. Requiring units to take time to rotate and target or turn to move? Many awesome things.

    Also of note, already in some respects is this in place. By exposing your avatar too deep into the enemy lines, it provides more room for enemy units to surround and attack. If you instead have a line of “tanks” standing next to each other, they are trading less melee opportunities for higher risk of AoE attacks hitting them all. However, extending this system is something I’d love to see in a MMORPG someday. Of course, there’s the issue of the avatars not being able to react very well to turn and such, but then I guess it would evolve into more avatar microing of not just skills and targeting, but also to avatar movement, placement, and orientation. Movement and placement are already implemented to an abstract extent, but again, more would be much more interesting.

    I’m still waiting for an epic battle with huge firing lines like the opening of the Lord of the Rings movie. Huge armies where strategic placement makes a difference, not just the tactics.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: