Monday, March 14, 2011

Overcoming the (Un)holy Trinity

Syl has a really nice article about this online. Check it out.

You didn't, did you? Well, the basic point is that, while the holy trinity (tank, healer, dd) has its uses, it also has its restrictions. The wider point is that the entire character customization of current MMORPGs is debatable.

The Holy Trinity consists of at least two debatable parts: healing and tanking with threat. The dd part is alright; nothing to fix here. Let's collect a few strange facts:

- Healers can fix broken limbs with spells, while the guy in question is still continuing to fight. Bad enough that we have health bars, but from an immersion PoV, healing is absurd.
- Healers cannot be allowed to do anything else well - otherwise they are overpowered compared to other roles.
- Healers change every PvP battle 100%. Ever had the luck to play some battleground or warfront without healers, or against too many healers? It is pointless.
- Healers spend all their time looking at little green bars, instead of the action. The fact that getting out of the fire is hard is actually a game design problem; that Blizzard made it a symbol for skillful play is quite ironic!

- from an immersion PoV, tanking and aggro are absurd. Full Stop.
- Tanks cannot be allowed to do anything else well - otherwise they are overpowered compared to other roles.
- Tanks are often useless in PvP. If they are useful at all, then for specific tasks that only require a specific amount of tanks.
- Tanks spend all their time looking at one or more toes of their enemies.

The Holy Trinity
- Creating a group inside a game that is built with tanks and healers in mind can be an absolute pain. You either have too many or not enough tanks/healers. Talking about accessibility: Way before you tune the challenge level of your game down, think about this!

And what good comes out of it? Now, this is interesting!
- Assuming proper scripting, a very interesting encounter can be created with the help of the stability that is created by threat mechanic and healing from afar.

To understand this, you need to consider a MMORPG without any threat and without any healing! If you just scrap it, the monster will attack people randomly or, if it is smart, always the guy that does most and can sustain the least damage. If the monster takes into account distance, a rather unpredictable fight emerges. Without healing, the fight will also end in death for a lot of players or be a pushover; possibly both. Can this really be a design intent?

In some way the holy trinity, in combination with the threat mechanic, allows the designer to go wild with scripting. A pure tank and spank encounter is so trivial and static, that the designers do not have to fear any emergent behavior. They know exactly what the situation is like after whatever scripting event they implement. This alone can allow for a great game, like WoW has certainly been.

Bottom line is: There is a really good reason for the threat mechanic and healers. Even though there are so many bad side effects!

The solution?
There are two ways to solve this. One really tries to solve it, while the other one tries to circumvent it.

1) To solve it, it is useful to replace 'tanking/threat' with 'control' (not only for the player, but also for the game designer!) and 'healing' with 'support'. Assuming enough time to polish the game, it could be possible to pace it in a way that stunning, kiting, collision control, thrusting and (de)buffing can provide enough control and support. The problem really is the polish here. To make such a game would require absurd amounts of iterations. Make the game too fast and the whole thing doesn't work. Make it too slow and people become bored. Moreover, the perceived speed of a game is heavily influenced by the number of combatants. And to make diverse encounters you want this number to vary drastically.

2) However, there is another way: And that is to embed combat in a more meaningful game. In a game with non-instanced PvP/PvE, for example, the individual skill isn't as important as the pure number of players. Combine that with an activity, like venturing into a dungeon to find treasure to buy special stones in the player-driven economy to reinforce your castle and thus prepare for war against your classmates. You can embed combat into so much meaning that the combat itself doesn't necessarily needs to be as deep and challenging as today. The resulting game might be much more fun than four bars full of skills or pre-determined dance scripts.

An easy example are some facebook games. Even a simple, obviously absurd and utterly boring activity can become a billion dollar game, if it is embedded in a context that imbues it with meaning.


  1. A party of 2 or 3 true hybrids used to be viable option to run dungeons. Changing roles on the fly (usually due to the mana situation) was fun.

    Although I still specialize in undermaning dungeons, the removal true hybrid specs at early levels resulted in static roles and less enjoyable runs.

  2. Holy trinity is popular design choice for many reasons:

    1) It allows greater diversity (3 base roles! - instead of 1 -do it all)

    2) Creates a system where each player knows what his role and tasks are. By now everyone is familiar with this framework .Its de facto standard

    3) Allows easier scripted boss design . (imagine boss encounter without tank/offtank -it would be chaotic , and majority players are too dumb/slow to handle chaotic on the fly)

    4) Allows players with bad reaction and coordination to play - sitting in place and pressing 1-2-3 is the maximum taxing ability some players can handle

    For those reasons (and there are more) holy trinity will never go away

    To do away with it you have to homogenize classes to a degree.There is higher player skill requirement as well unless you trivialize content

    Having said all that I am big believer in a system which is balanced all way starting with 1vs1 and higher up, Biggest challenge is have enough diversity imho

    I am extremely curious how GW2 gonna turn out as they follow non -trinity design. For existing examples of successful non trinity design pve one needs look no further than left4dead. Heck its obvious gw2 heavily borrows from it with their rez/knockdown mechanics

    I also would like to 2 complete failures (from class design point of view) which were not trinity - Asheron's Call and Darkfall. In both cases all the classes completely degenerated into one uber hybrid which does it all with little variance ( limited to order in which you raised the skill and primary attack weapon of choice)

    One has to be keenly aware of the many advantages trinity system has and pitfalls of other systems

  3. I think tanking is a popular idea in games because of the challenge of writing a good AI for monsters. Players will almost inevitably beat the AI and trick it into doing outrageous things, trivializing encounters. Developers used to deal with this by simply banning players who did it, which is pretty lame.

    I think the holy trinity is a very useful crutch for balancing extremely difficult encounters. What I see a lot of people clamouring for, however, is a focus on options, tactics and fun rather than difficulty. If you decide that it's okay that there is an unbalanced group composition that can smash your dragon because smashing your dragon isn't supposed to be the only way to measure success, then you can relax on the balance/control end and the holy trinity doesn't seem so important.

    If some player figures out how to make the monster spin in circles attacking no one because he confused its AI so badly then he can post the video on YouTube, but other people won't feel like they have to repeat it to follow one true path to get gear.

  4. @Max

    while I agree with you that the holy trinity is basically 'a crutch' and removing it might require a lot more skill from players, leaving their 'comfort zones', I'm not sure I like to think of MMO players along these lines. I think it's a huge paradigm shift for sure, to focus on individual players rather than roles, but I don't believe this wouldn't be enjoyable or 'too hard'. we would really have a chance to test this first.

    I also dont agree with 1):
    if anything, I tried to show in my post that the trinity does exactly NOT guarantee diversity, and can even kill it. why only have 3 exclusive jobs when the same player could have a lot more diversity and flexibility?

    my point was really, that you dont need the fixed trinity to have different classes, identity or diversity. you can have these things without it. it's important to differentiate between 'class' and 'role' though - i never said wow doesn't provide role identity (it does that to extremes).
    and we already have class homogenization in wow too, that's the thing.

    Like Nils said, it would be a big challenge in terms of designing and balancing this properly; but I am convinced it's doable in a way that appeals to the most classic MMO gamer - if only they're open to try.

    gw2 might remove some roles but not necessarily class identity.

  5. It's true that set-in-stone roles help people to know what to do. But that doesn't mean that with no tanks or healers we'd have anarchy. Even now we don't have tanks and healers, we instead have the tank who tanks this and the one for that, raid healers, tank healers, raid healers who switch to the tank at this or that time. Even with roles we're not given perfectly clear directions.

    My point is that we can handle adapting to a situation. We can handle figuring out that it would work to have an archer at that high spot and that high spot, a tough guy in the middle to draw the initial attacks, and then some casters to slow it down and help protect whoever has its attention. These are all clearly defined tasks, or at least as clearly defined as a task in current raiding, but they are not based on a holy trinity.

  6. I think devil is in details:

    If there are clearly defined tasks
    (control/support/damage) how you pick who does them? What tools they have to accomplish the goals? How balance comes into play?

    As soon as you stated that there will be different roles in combat you have only 2 options : everyone can do every role , or there are specializations

    If there are specializations why your particular system is better than holy trinity?

    Hybridization might be not that bad approach, but playing such hybrids with challenging content has extremely high skill cap, so either you have content which is trivial, either is too hard for majority of player base.

    If we take one of the more diverse FPS out there (TF2) it went the paper/rock/scissors way to achieve diversity (each class has its counters and counters other ones)

    If you start defining system in details imho either you have diversity with trinity( or trinity like systems) either you have homogenized hybrids. You cant have diverse system where you are not shoeboxing players into specific roles

    The closest you can get to achieve both is imho easy ability to switch roles (where each player can do anything potentially , but not at the same time).

    p.s. Having said all that I think more homogenized systems are better if it has competitive pvp elements. I strongly believe lots of pvp balancing issues can be avoided if devs start balancing 1v1 instead of proclaiming (its a "group game" so 1v1 is never balanced).

    Scaling from balanced 1v1 to balanced any amount of players is easier than inherently unbalanced systems

    I believe early wow had actually much better approach to class balancing as hybrid specs were more viable.

  7. One of my best memories in WoW was running Deadmines at level 17 with four of my co-workers.

    We all received or purchased WoW for Christmas 2004 and we decided to each create a mage to compare notes.

    Five beginners, all mages, in vanilla Deadmines.

    After several wipes, we did make it to Mr. Smythe but his stun mechanism was our downfall. We set up tank switches, bandage rotation, and learned more about kiting that day than on all other days combined.

    Five beginners, no tank, no healer (other than applying bandages), and we got pretty far in our first dungeon.

    Where was our holy trinity?

  8. Actually, I think there's a great example that already (or at least used to) existed in WoW: Twinking. At level 19 or 29 or 39 players didn't have 4 bars of skills. Instead, players maximized the efficiency of their movements, and coordination between players.

    On top of that, in at least a few cases, the battles were part of a bigger picture. Not meaningful world pvp or something truly awesome like that, but simply reputation within the community. The twink community is small enough that exceptional or unusual players and guilds found value in playing well even in an instanced battleground. Many twinks ceased to see winning the BG as the goal, and instead sought out the twinks in the opposing team, finding greater value in simply killing them, as an example.

    I also miss vanilla WoW for the reasons stated in your post, Nils. There were many a time in a dungeon run when a rogue would tank a mob for a short period of time during an emergency, or a skilled mage or hunter would kite a mob because aoe tanks didn't exist at the time. WoW has gone away from such emergent behavior, and it's too bad. =/

  9. Your first suggestion sounds like what I would say :P I'm glad you didn't suggest hybridizing like many people do, I've always felt that mixing roles is the fastest way to destroy meaningful place and balance.

    I found it interesting that the problems that people have with the trinity tend to stem from its implementation, as if playing a healer *has* to be about staring at green bars and tanking *has* to be about staring at an opponents feet.

    I agree with your solution. Developers need to give combat more depth and the roles have to be refined at a fundamental level. I think that the push toward action gameplay will help to make combat more dynamic overall, but I also think that a better understanding of what "healer" and "tank" will be important.

    I'm not sure if balance would be as difficult as you suggest, especially if player's skill is a primary factor. The player's own skill could buffer the differences to some extent and the basic values could be figured out based on the stats needed to succeed comfortably in pve.