Tuesday, April 15, 2008

The trinity is more MMO sin than holy

Last week I was in a pretty active discussion on Syncaine’s blog on whether or not good PvP and good PvE could both exist in the game. His point is that it can’t be done and my point was that it just hadn’t been done yet. I’m not going to rehash the whole discussion, but I will quote something I wrote that was a bit offtopic that mirrors my sentiments about what I think is wrong about MMO design philosophy:

[…]Why do you need a tank and clothies? Let me ask you, in real warfare – does your enemy just run up to one guy and let everyone else beat on them? No. Only dumb animals would behave in that type of manner. In the early days of MMO, developers had very limited AI so this mechanic of “tanking” a mob simplifies the AI required to make the encounter more real.[…]

In other words, the “holy trinity” of MMO game design was created because writing real AI was too sophisticated. For those of you unfamiliar with the trinity it follows this logic: A class to soak up damage, a class to deal damage, and a class to heal damage. The damage soaker controls the situation and takes all the damage, the healer heals the damage the soaker is taking and the damage dealer takes them out. There are variants, including crowd control and buff classes, but the basic design is around those three elements.

Early MMO developers needed a way to make NPCs very difficult encounters with a very limited AI. The result was to have a class whose sole purpose was to control or maintain “hate” or “aggro” of the creature and soak up the damage they dealt. They were given “taunts” and other abilities that didn’t scale damage but scaled some hidden mechanic only known as threat. This provided them a method for scaling an encounter to something difficult but without needing a complex AI for the NPC.

Of course, a truly intelligent NPC would evaluate threat in an entirely different way. It would ignore the hard to kill thing that does little damage and set priority by other criteria like distance and the source of damage or healing. And once they were hitting something, they wouldn’t be nearly as likely to disengage – turn their back – and run to a new threat unless they weren’t having much success killing what they were attacking . Groups of intelligent NPCs would behave very differently. One might go for or call for help, others might flank you or move into a more defensible position.

You could never have NPCs act so intelligently and have classes with such obvious flaws as exists in today’s MMO. Certainly a tank archtype would just be ignored and cloth types would be simply one-shotted. Truly intelligent combat would require a redesign and redefinition of roles.

So why change it? Because it’s flawed. And it’s boringly predictable. Remember that this design philosphy was the result of limitations in being able to script good AI. In recent years, the ability to program AI has come a long way and that is a problem that no longer exists. The reason for having such a system is antiquated and the only reason the system exists is because we haven’t evolved past it. In part, we haven’t evolved past it because we (as the gaming community) haven’t demanded that developers give us a better game.

How is it flawed? The biggest flaw is that it requires a specific group composition in order to succeed. A group of five or ten friends can’t just get together and succeed at an encounter. They need to worry about how they are putting that group together and consider group composition. It’s one thing if it’s “you be the healer this time” and it’s another thing alltogether when your role is defined by your class, gear and specialization.

If you like playing a tank and your best friend likes playing a tank – well, one of you is going to have to do something else when the encounter only requires one tank. In Warcraft, the problem is magnified in the end-game raiding scene because the ratio required for success in 5-mans is different than what is required for a 25-man. Every 5-man requires 1 tank, 1 healer and 3 DPS. A twenty-five man by contrast, needs 2-3 tanks and 6-8 healers. That’s up to 3 tanks fewer and up to 3 healers more than what would be needed if the 5-man ratio continued into the 25-man raid. It’s really no surprise that guilds struggle to find healers and ask the hybrids to spec for healing.

This whole dynamic just sucks if you get can get 4 or 9 or 24 people together and can’t go because you are missing that one type that determines your success or failure. It doesn’t matter that you have ten other people “on the bench” if they play a class that doesn’t fit in the group composition. And to make matters worse, the gear requirement will be steepest for the damage soakers (tanks) and damage healers since your group success is so highly dependent on their success. And if we do fail, who will we blame? I find it enlightening that one of my good RL friends quit his Feral Druid tank alltogether in favor of a Hunter he recently rerolled. One of his biggest reasons? He was tired of the scrutiny he kept getting from people. The other reason was because he was a complete non-factor in PvP.

Which brings me to the flaw in balancing PvP and the solo game. In the solo game, classes that can’t deal significant damage are frustratingly slow compared to those that can kill things rapidly. In the PvP game, these classes are at a very distinct disadvantage in a 1v1 game and only the healer is desirable in group PvP. If we increase the damage these classes deal, then we start to unbalance our PvE game. Syncaine’s observation that you can’t have good PvE and good PvP is certainly accurate if you presume that MMOs must continue to follow this antiquated forumla that has become the basic principle of MMO design.

The solution is to redefine these archtypes and start thinking outside the box. The trinity is broken and we need to start asking the developers of new games to think beyond the trinity. The wonderful thing is that there is no “best” solution, but lots of possible solutions.

All MMOs have their roots in the old pen & paper RPGs. The interesting thing about these RPGs is that the trinity didn’t exist in these games. We certainly had healers and warriors and cloth wearers, but a group didn’t need to consist of any specific combination of classes in order to succeed in a dungeon. One of my long-time groups that I played with frequently didn’t even have a cleric (healer) in the group.

I think part of the reason I have some hope for Warhammer Online is that it is based on a pen & paper / minature game that has it’s roots in these old games. A concept like “flanking” is irrelevant in WoW because the game lacks collision detection. I am by no means sold that WAR will be anything more than WoW, but I will say that the idea of collision detection brings a bit more realism to the game and I have to believe that a smart developer would build a smarter AI to take advantage of the nuances such a system could provide.

My final point is best summed up by something else I wrote in that discussion on Syncaine’s blog:

[…]The most common complaint that I hear from people is boredom. Bored with the grind. Bored with the lack of new content. Hmm. Well, if the content was a little more dynamic, then it wouldn’t get old nearly as quickly. WoW players don’t do things because they are fun or challenging, they do them because they have been taught that meaningless grinds provide them with items and gold. I say give them fun and challenge along with the reward and they would gladly embrace the change.[…]

No comments: