Psychochild's Blog

A developer's musings on game development and writing.

17 December, 2009

Article: Rethinking the Trinity of MMO Design
Filed under: — Psychochild @ 7:31 AM

I have a new article posted at Gamasutra: Rethinking the Trinity of MMO Design. It is intended to be an article in the vein of my How to replace levels posts. This time around I’m looking at the “holy trinity” of tank, healer, DPS as the core design of a game and what could be done to create an alternate system.

I’m interested to read your thoughts on the article.

Next Post: »


  1. Have you looked at Mortal Online yet? It’s being done by a Swedish company called StarVault. It’s a skill based MMO and a lot of the mechanics are in the vein of UO, but it does contain ‘realistic’ combat in the sense of having friendly fire and cover mechanics. Hide behind a tree and the archer can’t hit you unless he moves.

    It’s still in closed beta (NDA lifted) and looks to have a lot of promise. Lots of issues to work through yet, but I am hopeful.

    Comment by Joe — 17 December, 2009 @ 9:17 AM

  2. Excellent article, it’s going to take me a while to think through a full response.

    In the meantime a discussion over at Epic Slant is quite interesting:

    Have to chuckle at the fact that someone at gamasutra has called you an idiot already. Forum pvp is the flavour of the moment it seems. Maybe Ghostcrawler could tank it for you ;-)

    Comment by Stabs — 17 December, 2009 @ 10:54 AM

  3. Interesting article. A few thoughts:

    * In D&D, heavy armour doesn’t actually necessarily result in greater damage mitigation. A thief (or later a rogue) with high enough dexterity could be just as hard to damage (i.e. have as high an AC in modern terms) as a fighter in plate mail. Likewise, with the right spells up, a magic-user (or later a wizard) could likewise be just as hard to hurt in combat as anyone else. I don’t think the underlying assumptions that rule today’s MMOs (e.g. you can either wear heavy armour, or use spells, or have stealth, but not more than one of the above) really arose in D&D until 3rd edition, by which time MMOs had already established their current form. D&D was, until fairly recently, designed for a high degree of flexibility in character builds, such that no two wizards were really very much alike at all; I think the rules from 3rd edition on have actually been influenced by the popularity of MMOs, and not the other way around.

    * In terms of shaking things up, I’d propose a few more drastic options. E.g. remove taunts but implement collision detection, such that “tanking” means physically putting yourself in front of a foe. As for healing, why not try restricting heals to out-of-combat only, or at least drastically reduce the number and/or power of in-combat heals? With no taunts per se and no (or less) in-combat healing, there’s no more need for the holy trinity; instead players would have to rely more on positioning, crowd control, debuffs, and focussed fire to mitigate damage. That could result in a far more tactically-oriented game than what we have now.

    Comment by foolsage — 17 December, 2009 @ 11:32 AM

  4. I think you missed the central point of the trinity. The trinity exists to answer the question, “who does the AI attack?”

    If you figure out who the AI is going to attack, it makes logical sense to have that target be as hardened as possible, or manipulate the situation such that the AI attacks the hardened target. That results in the creation of tanks, from whence the rest of the trinity flows.

    Even in fights where the rules are different will see this dynamic emerge (unless the target is pure random). For example, take a look at the Faction Champions fight in WoW. FC has 10 enemy mobs who don’t use normal threat. Instead, they attack people based on three attributes:

    1. Defenses (armor/resistance). Individuals with lower defenses are attacked first.
    2. Proximity. Closer individuals are targeted before farther ones.
    3. Health Deficit. Individuals who have lower health are targetted.

    Even in this fight, the trinity emerges. We use a tank that starts the fight with half health to get initial attention.

    Second, this fight is not very popular with many players (though I enjoy it). People *like* the regular trinity gameplay. The basic structure is predictable, but also allows many individual twists, which makes each fight different and more interesting than the last. A less predictable basic structure might seem more appealing, but also might lead to many fights which end up more or less the same in the end.

    Comment by Rohan — 17 December, 2009 @ 1:45 PM

  5. Joe wrote:
    Have you looked at Mortal Online yet?

    Not much. I’ve heard a lot of people are getting excited about it. These days I’m only playing one game while trying to focus on my own projects. I suspect that Mortal Online is going to have a hard time because it appears to be targetting the same audience that Darkfall is. Not sure that niche is big enough to support two larger games like that.

    Stabs wrote:
    Have to chuckle at the fact that someone at gamasutra has called you an idiot already. Forum pvp is the flavour of the moment it seems.

    Unfortunately, I don’t see that post; it must have been removed. But, forum PVP has always been in fashion. But, instead of tanks, healers, and DPS you have targets, ass-kissers, and flamers.

    Thanks for the link to the Epic Slant forums. Haven’t visited there as often as I might like to.

    foolsage wrote:
    In D&D, heavy armour doesn’t actually necessarily result in greater damage mitigation.

    All other things being equal, it does. A Fighter could have just as high of a Dexterity score as a Thief, so the Fighter wearing heavy plate is going to have better AC (or defense) than the Thief in leather.

    …with the right spells up, a magic-user (or later a wizard) could likewise be just as hard to hurt in combat as anyone else.

    I mentioned that spells and magic items can help fill in the defensive role for different players. That’s one element that makes D&D much more flexible.

    In terms of shaking things up, I’d propose a few more drastic options.

    Personally, I agree we do need to go more drastic. But, people have been proposing drastic changes for decades now. I wanted to show that you don’t have to throw everything out and start over. I thought it was amusing that we could go back to the source of the current system for inspiration.

    implement collision detection

    The internet still lags, sadly. Last night in LotRO I was tanking an instance on my Champion and locked up for about 10 seconds. If the only thing preventing the squishy Minstrel and Lore Master from getting eaten was me moving to block the monster, they would have been hurt during that time. Given that I had built up enough aggro in a more traditional trinity system, they were fine during that time.

    Rohan wrote
    I think you missed the central point of the trinity.

    Not really. I didn’t spend a lot of time on the definition because the article was long enough as it is. Yes, you’re right, it makes sense to harden a target and get the A.I. to attack that. The problem is that this has stuck us in a rut for a while now.

    There are also the second-order effects to consider. Like how Tanks can’t really do good damage, otherwise why bring fragile DPS if Tanks can take hits and do damage? DPS needs to be fragile, otherwise why bring a specialized tank if DPS can do damage and take hits? Plus there’s the assumption that healers have to be weak in order to make them get hurt fast if they get catch aggro away from the tank; this is why you won’t really find healers in heavy armor in current MMOs.

    People *like* the regular trinity gameplay.

    Some do, yes. But, new mechanics can give a new game a unique selling point. I think it’s also accurate to say that players in a game designed around the trinity don’t like fights that violate the assumptions of that design. The fight you describe makes it so that the people who are supposed to be least able to take the hits do take the hits; that’s obviously not really much fun unless the player has some way to increase defenses while being hit. Since the design of the trinity enforces that only tanks can take sustained damage, this is generally not possible.

    Comment by Psychochild — 17 December, 2009 @ 3:06 PM

  6. I’ve read this article about halfway, and I begun to think of my first MMO. I came to the party late, but I think it has given me an unbiased look at the current generation. My first one was Matrix Online. Now I actually played in Closed Beta, which compared to final release, was in my opinion awesome. From Beta to release, it was nerfed.

    Since combat was interlocked, you and your opponent could only damage each other, the tactics changed considerably. So you had 3 different type of classes. Inside DPS, Outside DPS, and Healers. Inside DPS would most likely be a tank, but you wouldn’t design the character that way, as many abilities for characters were to do damage for the melee class, and few were designed to block attacks or debuff the opponent. Outside DPS, the black mage could basically be a direct damage dealer, an area of effect, or a debuffer. Then Healer would be around the opposite of the black mage.

    There was actually another class, a Bard, buffer and debuffer, but it was seldom used. Now you could also switch out your abilities at will, just go up to a phone booth and you can click off an ability and select another. Now there were macro’s in place to allow me to change within seconds which class I needed to be. Didn’t work on clothes though.

    The difference between Closed Beta and final release was multi-classing. Closed Beta, it was easy. I was a spy/SMG Specialist. Using Sub machine Guns in the matrix mode looked awesome. As a spy, I could do sneak attacks, or blind my enemy with a flash bomb, and then take them out with my guns. I also had some light Tae kwon do, in case my guns were knocked out of my hands. Few people actually knew about the spy tree, so it made me kind of a unique character. And perhaps my character was a little strong, but it allowed for some interesting combinations. My opinion is that the final release was an unfinished product, and they prevented multi-classing to that degree, so either I was a spy or a soldier, not both. Which practically made spy useless.

    One element you mentioned, which I did notice in MxO, is healing. We are chasing a big boss through the streets, that can kill you in one hit. It was actually kind of exciting. But when you died, you have people rezing, even people of opposite orginizations (I was Zionist, and a Merovingian rezed me). But while I got experience for doing damage, the healers and rezers did not. And I don’t know if WoW, or any other game ever trying changing that paradigm, but as a potentially future developer myself, I would like to see that changed.

    Comment by mad_cat — 17 December, 2009 @ 3:32 PM

  7. Psychochild wrote, “A Fighter could have just as high of a Dexterity score as a Thief, so the Fighter wearing heavy plate is going to have better AC (or defense) than the Thief in leather.”

    Actually, that doesn’t work. Armour imposes restrictions on how much Dexterity bonus a character can apply towards AC. So e.g. when wearing very heavy armour, a character can only apply 1 or 2 points of Dex (or Int) bonus to AC. Without getting into enhancement bonuses and shields, armour won’t get a character past 20 or so AC. Contrariwise, a completely naked rogue with a Dex of 30 would also have an AC of 20. Amusingly, it’s often easier for characters to achieve very high AC by using only very light armour, than by wearing heavy armour, for exactly this reason – e.g. Monks in v3 and v3.5 had AC bonuses every 5 levels, and could apply both Dex and Wisdom to their AC – so a high-level monk wearing cloth armour usually had one of the highest possible ACs.

    You’re quite right about lag making collision detection problematic; I can only hope that latency will decrease enough and/or other technology will solve this problem.

    Psychochild wrote, “Yes, you’re right, it makes sense to harden a target and get the A.I. to attack that.”

    I think this makes a great deal of sense for players to attempt, but I also think it’s fairly ridiculous and unrealistic behaviour for the AI to follow. The more intelligent a foe is, the more likely they should be to ignore the hardened target and attack the highest-priority tactical target (healer if one exists, dps otherwise). Taunting foes is fine and dandy when the foes are mere animals, but sentients shouldn’t be so easy to control in this regard. That’s always bothered me.

    Comment by foolsage — 17 December, 2009 @ 3:40 PM

  8. foolsage wrote:
    Armour imposes restrictions on how much Dexterity bonus a character can apply towards AC.

    I guess I didn’t make explicit that I was using 2nd edition as my reference. Third and fourth editions do have maximum dex bonuses applied for the class of armor. I actually do prefer 3rd edition for playing, but it did introduce a lot of mechanics that could complicate the discussion, such as the Monk example you give. While monks at high levels are killing machines, they start out fairly humble; as a side note, I’ve never been a fan of the “you must suck at low level to get great power at high level” type of balance.

    I also think it’s fairly ridiculous and unrealistic behaviour for the AI to follow.

    Most game encounters aren’t intended to be simulations, they’re more like puzzles with partially hidden information. A group’s goal is to do as much damage as possible while not drawing aggro away from the tank. So, people come up with systems, plugins, or the designers just add an indication to show how you are doing progressing on the puzzle.

    You are right in that there is often some type of high priority target in a group. This is why the trinity design rarely works in PvP, and it can be a jarring experience for people who “try out” PvP in a heavily PvE game with the trinity design.

    Comment by Psychochild — 17 December, 2009 @ 4:31 PM

  9. Psychochild wrote, “I guess I didn’t make explicit that I was using 2nd edition as my reference.”

    Ahh, fair enough. 2nd edition had great flexibility but wayyyy too many loopholes IMO. But ok, gotcha.

    I also never really liked the “start out as a commoner” model, although it is of course a common trope in fantasy literature. I find it gets pretty tedious to start yet another game by fighting rats and work my way up to boars in the hope that someday I’ll be able to go one round with a dragon.

    Psychochild wrote, “Most game encounters aren’t intended to be simulations, they’re more like puzzles with partially hidden information.”

    Very true. Often the desire to make a “fun” challenge in-game leads to fairly unrealistic behaviour – e.g. the well-known constraint that causes AI to (generally) be extremely stupid. I was just noting that tanking as a concept really doesn’t make a lot of sense, in and of itself. We’ve come to accept this model of how combat works, and even to expect it, but it’s still a bit silly.

    And yeah, tanking generally fails, HARD, in PvP. Players who come to rely on it are in for nasty shocks. That’s one reason I’d love to see collision detection, latency permitting of course; it would allow similar behaviours to be effective both in PvE and in PvP, meaning skills would be more transferable than they are now.

    Ah well, maybe someday. ;)

    Comment by foolsage — 17 December, 2009 @ 4:57 PM

  10. Sorry in advance for the long response – this is one of my favorite topics, so I got a little excited.

    First I’ll reply to this:
    I also think it’s fairly ridiculous and unrealistic behaviour for the AI to follow.

    You also have to remember that the AI is a computer – and no matter how complex you make it, someone can figure out how the AI works and teach everyone else how to exploit it.

    If it attacks healers first, I’m going to boost the defenses of my healer and make him tank.
    If it attacks the guy with the least health, I’m going to cut my tank’s health before we get started.
    If it attacks guys on red shirts, I’m gonna get my best defender a red shirt.
    If it attacks on proximity, I’m going to change where my tank stands.
    If it attacks psuedo-randomly based on a time-based seed, I’m going to keep trying and resetting until I figure out the seed and calculate when to attack to get my “lucky rolls.” (Eventually there will be a 3rd party hack for this.)

    No matter what the method of determining targets, players will find the way to exploit it.

    Okay – now on to the good stuff:

    Why does the trinity exist?

    MMO Players are always moving (sometimes very slowly, but always moving) towards the optimal solution.
    MMOs are meant to be perpetual (play forever) so players will eventually figure out that optimal solution.

    In a game where the following 4 are true, the trinity (or as close to it as possible) is the optimal solution:
    1) The goal is to get an opponent’s variable to 0, while keeping your own variable above 0 (first to 0 is out of the game).
    2) Specialization (in reducing the enemy to 0 or preventing yourself from being reduced to 0) is possible and has benefit.
    3) The enemy is a computer (and thus the AI is exploitable)
    4) (optional – only needed to add healers) Player specialization allows PlayerA to improve the survivability of PlayerB above what PlayerB can reach on their own.

    * The optimal situation for any fight: PlayerA, solo, one-shots enemy. To increase the number of times that happens, PlayerA would want to maximize the rate at which they decrease enemy health, sacrificing all other stats if it needed to do so.

    * If one-shot isn’t possible, but the ability to manipulate AI is (see above about how players can ALWAYS manipulate non-random AI), the best solution for PlayerA is to stay at max DPS (for all fights where it is possible), but get the AI to attack PlayerB in cases where it isn’t.

    * If the amount of damage PlayerB loses by specializing in survivability is greater than or equal to the amount of damage a max DPS player (PlayerA) can do in the added time, then the optimal solution is for both be max DPS and have the same “cookie-cutter” class. (Because they have more creatures they can 1-shot while solo, and are equal in teams. To have PlayerB survive would reduce PlayerB’s solo game, with no gain in group play.) Note: This is a possible game, but not as fun.

    * However, if the amount of damage PlayerB loses by specializing in survivability is less than the amount of damage PlayerA can do in the added time, then the best option is to have PlayerB specialize in survivability, but only to the point needed to defeat the enemy. The stronger the enemy, the more PlayerB will need to specialize in survivability. To beat the “toughest enemy we can” means PlayerA specializes in Max DPS, sacrificing survivability (glass cannon), and PlayerB specializes in Max Survivability, sacrificing DPS (tank).

    * If you want to take down an enemy even more difficult than that, you need a PlayerC who can increase PlayerB’s survivability beyond the maximum that one class can do for itself. This is commonly done by healing – but there are other ways to fill this role. Note: It is possible to have a game without this. In that case players will either reach their limits far earlier in the game, or they will learn to stack up multiple tanks to defeat larger foes.

    * Of course, adding more DPS makes the fight faster, so that’s why you’ll generally see 1 tank, 1 healer, and many DPS.

    Commonly, in fantasy MMOs, PlayerA maximizes their damage by increasing direct damage, or using DoTs, or even reducing an enemies defenses, but these are not the only options. A summoning class could add more pets to the fight. A class could increase damage done by other players, and do their DPS by proxy, through a pet or other ally. Even if it’s lazers, bullets, or cream-pies instead of fireball, it’s still “maximize damage at the expense of everything else.” It’s still a “glass cannon.”

    Commonly, in fantasy MMOs, PlayerB maximizes their defenses by mitigation (armor/shields/blocking), but dodge/parry also increase survivability, as does a self-heal, as “spell reflect”, or increasing PlayerB’s own total health, or an immunity bubble, or reducing enemy strength/damage done, slowing the enemy, even crowd controlling enemies so they can’t do any damage at all. It can even be about movement speed and location in game-space. In EvE is it about having enough capacitor to run shields, staying out of range (but still targeted) so the enemy can’t hit you, and being fast enough to move out of range if they do. It doesn’t sound like tanking, but it is — it’s “maximizing defenses as much as needed, at the expense of DPS.”

    Again, it doesn’t have to be “sword-and-board.” The lazy part is not the trinity – it’s in thinking “tank == heavy armor”. In Alganon, we have a Frost Magus who casts spells that boost his defenses and make him a regen-tank (in addition to damage). Our Ranger tanks can kite the enemy, staying out of harm’s way. We’re currently working on a swashbuckling assassin who use dodge/parry/reflect instead of damage mitigation, and we’ve got plans for a class that lays down areas of defense and uses his physical location (movement in game space) to change his defenses and allow him to tank.

    Commonly in fantasy MMOS, PlayerC boosts PlayerB’s survivability with heals, but there are many other ways to do it. Shields, bonus HP, increase dodge, anything that PlayerB can use on themselves, PlayerC can stack on top and be a “healer.” Crowd control is a fun method here, but it’s still the trinity.

    It is common for the player who maximizes their defenses (the tank) to focus on controlling the AI, but that is not always the case. Many early MMOs had no “taunt,” and many current ones require the tank to use proximity and collision detection to stay as the primary target. It is nice, however, because it gives them something to do other than “be hit,” but it doesn’t have to be that way. A ranged pet class that puts its survivability in the pet would be perfect to make into an aggro control class – when solo, it keeps the enemy on the pet. In a group, it keeps it on the tank. Adding this doesn’t mean losing the trinity – it’s still the trinity.

    I think the key is not “getting rid” of the trinity, but embracing it – make it more fun.

    I love to put a different metaphor on an existing role, like an magic-user that can tank, a soldier that increases DPS by proxy, a ranger that heals. I love to create fights that flip the mechanics on their heads, where high defense or “not being the target” becomes a bad thing, or fights where you heal the enemy to death. These mechanics can be fun, and original, but I know that it’s still the trinity, and that’s okay by me.

    Comment by Hue — 17 December, 2009 @ 6:02 PM

  11. Oh, and the “tl;dr” version:

    1) A small number (3+1 optional) elements that are critical to MMOs form a system where the trinity is the optimal solution.
    2) Players always move towards the optimal solution. They will exploit and beat the system if they have to.

    Thus – you can have an MMO with the trinity, or you can fight your players and hope either you, or they, quit before they kick your ass and put the trinity in your game anyway.

    Comment by Hue — 17 December, 2009 @ 6:14 PM

  12. Interestingly enough, your trinity alternative sounds somewhat similar to David Allen’s recently launched game Alganon. I haven’t played, nor do I really intend to, but here is my understand based on what I’ve read.

    Each of the four classes has one of four primary roles: Warriors – tanking, Ranger – crowd control, Healer – healing, Magus – damage. Each character then has a talent tree that allows them to spec into the other roles. So if you need a tank you know that any Warrior can do it, but so can an Ice specced Magus.

    I don’t know to what extent a class’s dual roles, as they call them, are simultaneously available. Does a tanking spec Ranger still do some amount of crowd control, more like your trinity alternative? I can’t say.

    For better or for worse I’ll let the other areas of the game speak for themselves, but the class system is fairly interesting in theory if not in reality.

    Comment by Gary — 17 December, 2009 @ 7:40 PM

  13. Great article.

    Wizard 101 is an interesting example of how to do things differently. It theoretically has a “tanking class” (Ice school) and a “healing class” (Life school), but the heart of combat consists of players casting buff spells on each other and debuffing the enemies, with the goal of allowing one player to cast a spell for huge damage. For example, a Death wizard, Storm wizard, and Balance wizard grouping together could cast Feint (huge damage buff), Elemental Trap (elemental debuff on enemies) and Tempest (big elemental damage spell). The neat thing about it is it requires player to work together to maximize damage output.

    Rather than having a single heavily-armored class, the game allows players to cast defensive buffs on whoever has aggro, effectively allowing anyone to become a temporary tank. And many schools have a self-heal spell, reducing the need for a dedicated healer.

    It’s really a very creative system. It promotes interactions among group members, but it does it in a completely different way from the D&D standard.

    Comment by Tolthir — 17 December, 2009 @ 10:21 PM

  14. Good thoughts Psychochild, I have been thinking for a while now that there needs to some shake up in the combat system for MMO’s. I hear a lot of people talk about an upcoming game being a WoW beater and they always fall flat, and I think a lot of that has to do with a very similar game play mechanic between the challengers and WoW. One interesting concept that I think could use more exploring is a system that was used for sword fighting in Pirate of the Burning Sea, which was mostly based on ship to ship battles but had some boarding combat and fighting on land. The system they used relied on two attributes health and an avoidance attribute. Instead of having a fixed dodge change or armor modifier based on dexterity each player had a bar that showed there ability to dodge and respond to attacks called balance. Players had to manage their balance well because the health pools were relatively small, once you could no longer dodge or parry attacks you died pretty quick. Each time you dodged or parried an attack you lost a little balance making you easier to hit, similarly different attacks affected your balance differently, there were attacks that didn’t do much damage but caused your opponent to lose footing and thus a lot of balance (aka a swipe at the ankles). On the offensive side you had to be more measured in your attacks as well, the more aggressive you were the more quickly it eroded your balance.

    This combat system isn’t perfect, but I think, with some work and tweaks it could make for an excellent game. As I have envisioned it instead of class differences being based around skill sets and allowing certain characters to wear more durable armor the classes would revolve around how they manage their balance. Different classes would regenerate/lose/maintain their balance at different rates in different situations depending on the fighting style of their class. Each class would approach fights from a very different perspective and depending on enemy behavior would need to take on different roles in the fight. Sometimes having an expert swordsman who has perfect balance while standing toe to toe with an enemy would be desired, while other times having a barbarian, who is completely balanced while seeming to charge head long after enemies, yet can be caught flat footed if made to change directions would be better. Just something I have been thinking about, would love to hear your thoughts on it.

    Comment by Barret Hudson — 17 December, 2009 @ 10:58 PM

  15. W101 really is a great game. I’m also fond of Puzzle Pirates, where combat is a very different animal.

    I’m a bit curious as to why Battletech hasn’t been mentioned yet. When every fight is a battle of attrition (no healing), tactics, careful range manipulation (positioning) and information parsing, you get a very different sort of combat. I’m very interested in the MechWarrior reboot, since they make the claim that they want every weight class of mech to be useful. The “standard” assumption that “All Assault Class, All The Time” is the way to success is something they are trying to get away from. A player in a light mech (recon, evasive, sniping) should be just as interesting to play (albeit very differently) as a hulking assault behemoth. It’s a totally different paradigm from “level 80 beats 80 level 1s (or 4 level 20s)”, placing much more emphasis on player skill and tactics.

    Of course, lag still rears its ugly head there, if you’re talking MMO design. That and the “client in the hands of the enemy” mantra might be the biggest obstacles to making combat more tactical and skill based, which would open up more design space for different combat modes. Well, that, and the player base’s resistance to turn based combat. Dofus, W101 and Atlantica Online are all more tactical thanks to being turn based, and it really does change things around.

    Then again, even in something like Disgaea, I’ll use “tank, ranged DPS and healer” tactics since it’s all based on depleting HP. Put a well defended character in a choke point, stick some ranged damage dealers behind him, and have a healer on tap, and you effectively have a solid plug that the AI crashes itself into, to its demise. As such, when combat is mostly about depleting HP and combat rapid healing is possible, you’ll wind up with the trinity roles in one way or another.

    Combat in Advance Wars is interesting. It places a high premium on aggression, since damaged units can’t hit back as hard; they are hurt, which reduces their firepower. Disposable units and large scale warfare work well there, but could that be a good angle to look at MMO combat? What would it do if injured characters (whether that means low HP or actual injuries based on tactical considerations) were less effective in combat, or maybe just different? Could their role shift over the course of a fight, where a hurt tank shifts to support by kiting a ranged attacker? Could a healer out of mana go berzerk with righteous indignation and suddenly become a fierce melee DPS engine… until it keeled over, exhausted (or dead)? Could a ranged attacker shift to healing when very limited ammo runs out?

    Back to giant robots and attrition, Front Mission 4 had “repair” Wanzers (mechs) that effectively functioned as healers. It changed the dynamics of the game, since earlier iterations had been very Battletech-like in their attrition-based combat. Repair is a significant “fudge factor”, allowing for more mistakes to be made (important when some tactics are *random* in that game) while still permitting mission success. In Front Mission 3, combat was shorter and more unforgiving. Is there a way to introduce a fudge factor for experimentation and tactical flexibility *without* healing or repair? Increased avoidance and fog of war are one possibility, as is informational denial (messing with opponents’ abilities to target/fire/etc.

    Sorry, this is a little disjointed and blathering, but I really like thinking about these things. The underlying assumptions about what constitutes “proper” MMO combat are not sacrosanct, and it’s nice to stab at the sacred cows to see what kicks back.

    Comment by Tesh — 17 December, 2009 @ 11:00 PM

  16. Hue wrote, “You also have to remember that the AI is a computer – and no matter how complex you make it, someone can figure out how the AI works and teach everyone else how to exploit it.”

    This is certainly true, and I don’t ever forget it, sadly. It’s also true however that the current AIs are pretty primitive, by very deliberate choice, and that more complex AIs are harder to overcome through such simple tactics. Current AIs provide very very simple tactical problems and rely on trial and error on the part of PCs to solve the problems. Once solved the combats are generally trivial – thus the existence of “farm status” on even the toughest fights. PvE fights only require tactical thinking by and large during the learning stage, not during execution, and that’s disappointing to me.

    I advocate more realistic AIs and the elimination of a lot of the tactical shortcuts we use today, e.g. the “holy trinity”. I firmly believe that reliance on tank/healing/dps is oversimplified blather and panders to the lowest common denominator, making combats far too simple (note – “challenge” and “simplicity” are not on the same scale). I do think that challenges need to be gauged to the audience, but I think the current model in MMOs largely fails in this regard; I’d prefer players have the ability to set their challenge levels more dynamically. I’d also strongly prefer to have pretty much all PvE fights be less predictable. Don’t you get bored with combat? I surely do, even in games I really enjoy. Why is that? I think it’s because the fights are simplistic and invariable.

    Humans are incredibly efficient problem-solving machines and I respect that. No matter how games are programmed, people will find the easiest ways to solve problems. However, the more randomized the problems, and the deeper the possible complexity thereof, the better able games will become to challenge novices and experts alike. Right now they really only succeed at challenging novices; the challenge to experts is forming a group and getting everyone not to screw up as we all progress through the completely predictable prescripted fights. That’s… honestly kinda sad. Where’s the adventure, where’s the heroism in tank ‘n’ spank? Where’s the adventure and heroism in learning every stage of a preset battle (either by repetition or by looking it up online) and then trying to lead others through the steps? It’s paint-by-numbers heroism, which isn’t remotely heroic at all.

    Tl:dr version: I don’t agree with your summation. ;) I think the holy trinity is only one of many possible approaches and I think its drawbacks considerably exceed its advantages. It’s a shortcut to true in-the-moment tactics and players have grown used to it, but it’s doing more harm than good IMO. I’d love to see some games break free of the constraint and experiment a bit. Change is good.

    Comment by foolsage — 18 December, 2009 @ 12:48 AM

  17. Tank You! Oh, No, Tank YOU!

    [...] Rethinking the Trinity of MMO Design (Psychochild) [...]

    Pingback by Tish Tosh Tesh — 18 December, 2009 @ 6:04 AM

  18. This goal of trying to replace the holy trinity is pointless given the assumptions of combat and group based play. Hue’s post breaks down the math as to why in any kind of combat, specialization via tactical roles is the optimal solution. If your game is based on group combat, but you don’t support roles, you don’t have a role playing game.

    The only possible way around roles is to change the victory conditions. Moving the enemy health to zero can not equate victory. Given the assumption of combat based gameplay, the only other way to achieve victory (achieving victory is the entire point of combat, no? unless you play to lose?) is to make victory conditions strategic, and reduce or remove entirely the importance of tactical victory. If you make strategy the core gameplay type at the individual player level you end up with a very different kind of game.

    The entire idea of replacing the holy trinity is, as you touch on multiple times, one of balance. There is this vague idea on the part of designers and some players that balance is important. Your reasons why the trinity is bad are nit-picky and plain wrong. They read as if you came to the conclusion that the trinity is bad, and struggled to justify your conclusion emotionally. The core reason why you think the trinity is bad is that it is necessarily unbalanced. I think the worship of balance is a very American ideal, and it’s seeming importance to some players and designers is evidence of a cultural bias. Balance is not important in and of itself.

    Fun is important.

    Comment by null — 18 December, 2009 @ 9:26 AM

  19. For sake of clarifying somewhat, “holy trinity” in Everquest terms actually didn’t include a DPS class at all. They were effectively optional and you never really required a specific DPS class, just some variety of one. The trinity was not about roles, but about specific classes — Warrior, Cleric, Enchanter. Without those three the group was crippled in some fashion. The rest of the slots could be made up of anything, optimally DPS just to speed things up.

    This is something that game design has actively stepped away from, and instead reduced the ‘required roles’ to simply Tank/Healer/DPS as opposed to Tank/Healer/Support/[DPS]. The issue is that DPS classes all do the same thing (damage) and support classes such as Enchanter offered a relatively unique toolset that was required for effective play in group settings.

    The term was re-purposed as a role-defined expression with later games.

    Not a big deal or anything, but if you’re actually going to attempt some pseudo-academics and chart a history of a term, do a little more footwork.

    Comment by HT — 18 December, 2009 @ 9:29 AM

  20. Losing Aggro

    [...] 18 12 2009 Brian “Psychochild” Green has a good new article and related blog posts up about rethinking the trinity of MMO [...]

    Pingback by p@tsh@t — 18 December, 2009 @ 12:34 PM

  21. It was a very good article, a perceptive description that I could relate well to as a player. I think what could be mentioned is that the reason why the trinity is so popular is because we need A.I. enemies to be stupid to have a chance to defeat them, and the trinity design makes them dumb enough to handle. Without hate control we have no way to have tactics or defend against an attack all, especially if the mob A.I. is smart enought to mimic players.

    The formal and abstract, rule-based system lets us survive. When it isn’t, you have PvP, in which a lot of players will die quickly and repeatedly, and not get so much out of the game.

    Comment by Dblade — 18 December, 2009 @ 2:31 PM

  22. Slashdot has a post about the article:

    True to form, however, the comment writers don’t seem to have actually read the article. ;)

    Comment by Psychochild — 18 December, 2009 @ 7:31 PM

  23. null wrote:
    The entire idea of replacing the holy trinity is, as you touch on multiple times, one of balance.

    I don’t agree it’s entirely of (gameplay) balance. I think there are a lot of other problems as well. Read Tesh’s article linked just before your post, “Tank You! Oh, No, Tank YOU!” His entire experience was ruined when the only tank they could find was poor at his/her role. It’s kind of brutal that the experience of so many people can be halted and/or ruined just because they can’t find one specific role necessary.

    I think your arguments are a bit misguided here, because it’s more that people resist going away from the trinity design because of balance. Current game design is balanced around the concept of the trinity, and breaking that requires a lot of re-thinking of the rest of the design. One goal of my article was to show that you don’t have to throw everything out.

    HT wrote:
    They were effectively optional and you never really required a specific DPS class, just some variety of one.

    Notice I didn’t say “DPS class”, I said “DPS role” which includes multiple classes. In older EQ some of the roles were embodied in a single class (healers were always Clerics because Complete Heal was the best healing spell bar none), but other MMOs went away from that pretty quickly. For example, DAoC had multiple types of healers and tanks and DPS even on each “side” or faction. DAoC even blended different roles pretty well (with my preferred class, the Thane, being a bit of a Tank and a bit of DPS); the game’s focus on PvP (via RvR) forced some changes in the trinity, though.

    Not a big deal or anything, but if you’re actually going to attempt some pseudo-academics and chart a history of a term, do a little more footwork.

    This isn’t “pseudo-academics”, this is design analysis. Much more interested in reality than academic detail.

    Modern trinity-based games (which would primarily be WoW for the last five years) the holy trinity refers to tank/healer/DPS roles. These games have a much wider variety of roles between the classes, even to the point of having classes that can specialize into different roles and even now able to switch between specs easily. I did include EQ for a bit of completeness, but a lot has changed since it was released a decade ago, so I didn’t focus too much on the details in the article. As you might have noticed, it already ran 4 pages and extra length in the sake of excruciating detail wasn’t a priority. ;)

    Comment by Psychochild — 18 December, 2009 @ 10:45 PM

  24. Awesome, gonna go read it now :)

    I’ve thought a lot about the Holy Trinity style of gameplay and why it’s so fundamental to MMOs. I think it’s just plain hard to come up with a different style of play without really developing some serious AI or overpowering classes.

    I think the perfect genre for removing the Holy Trinity would be a sci-fi MMO. I like the idea of standing and shooting at creatures, equally, and not doing the silly melee tank, medic healer and ranger laser DPS.

    Comment by We Fly Spitfires — 19 December, 2009 @ 3:16 AM

  25. null wrote, “If your game is based on group combat, but you don’t support roles, you don’t have a role playing game.”

    I’m puzzled by your definition of roleplaying. For most people, myself included, roleplaying means putting yourself in the role of your character – i.e. experiencing the game’s setting as your character would, and reacting in kind. This has absolutely nothing whatsoever to do with having roles that people have to fill in combat.

    null wrote, “The core reason why you think the trinity is bad is that it is necessarily unbalanced.”

    You might wish to read the article again. :) He listed 5 disadvantages to the trinity: 1) groups are restricted to certain compositions (and thus it can be harder to form groups), 2) trinity-balanced games don’t always support other gameplay styles like soloing, 3) limited roles restrict player choice, 4) systems tend towards complexity, and 5) the trinity doesn’t make much sense in some settings. Of those 5 reasons, only one (#4) really has anything to do with balance, and even there, balance isn’t the core issue; rather, player understanding of the options available is the issue.

    Your objections to his reasons why the trinity isn’t optimal seem founded in a misunderstanding of what he actually said.

    Comment by foolsage — 19 December, 2009 @ 4:45 AM

  26. First: I thought the article was a good overview, and hit the high points in terms of both history and challenges.

    I would like to just mention my own perspective on the topic, see if there is any feedback. I’ve felt for a while now that the Holy Trinity of classes (as well as many other aspects of MMORPG design) are most directly a result of the assumptions in the core system: i.e. HP/AC-def-THAC0-etc/XP, etc.ils down

    When the combat decision tree of combat boils down to: “whoever reduces their opponent’s hp to 0 first, wins”, there’s not a lot of room for much beyond the 3 core roles that have shook out over time. To win, I must reduce the opponent’s hp total faster than they can reduce mine… full stop.

    Characters can achieve that by 1) being able to take more damage/be harder to damage to begin with, 2) be able to restore their own and/or others hit points, or 3) reduce the opponent’s hit points quickly. (The 4th option that most can think of, “crowd control”, is essentially option 1, IMO… you are “harder to damage” because the opponent can’t attack at all, is slowed, is forced to use ranged attacks due to immobilization, etc.)

    Not many other explicit options available: not that creativity can’t build on this, tanks vs. crowd control being a good example of different means, same ends… but it generally boils down to the same 3 basic options, in varying proportions.

    So, I would suggest that truly breaking from the Holy Trinity would necessarily imply changing the core mechanics in a meaningful way, and that, conversely, changing core mechanics/systems would automatically lead away from the current big 3. Even something as simple as having two “hit points”-like scores instead of just one might be sufficient to change the status quo. Introducing multiple victory conditions, instead of just the current one (kill or be killed) might be another avenue of exploration.

    Anyway, what I’m driving at is, isn’t the Holy Trinity really an automatic outcome of the core mechanisms/assumptions that underlay the genre at present?

    (Sigh, I could write 5 more pages and not be done… I really should start blogging again…)

    Comment by DamianoV — 21 December, 2009 @ 11:53 AM

  27. Why dismiss collision detection so casually by saying the Internet has lag? That’s a bad excuse when almost every FPS Internet game has collision detection. I suggest looking at Team Fortress 2 as an example of a non-trinity team game with classes.

    Comment by Nelson — 21 December, 2009 @ 1:05 PM

  28. DamianoV wrote:
    isn’t the Holy Trinity really an automatic outcome of the core mechanisms/assumptions that underlay the genre at present?

    Yes and no. As I point out in the article, D&D didn’t have such a strict setup and it was entirely possible to run a campaign without a healer or someone in heavy armor. You had to adjust your tactics and perhaps the GM had to keep party composition in mind when designing challenges, but it was possible. Try doing a typical full group or raid encounter without a tank or healer in a recent MMO focused on the trinity.

    Many games also weren’t designed with the trinity in mind. In Meridian 59, people don’t build themselves to be “tanks” or any other aspect of the trinity; this is probably since PvP is a large focus of M59 and the trinity doesn’t work in PvP. Looking at the original EQ, Warriors didn’t have a specialized taunt ability originally. (At least not that I remember. I think it was added later.) Players dissected the aggro mechanics and figured out a way to manipulate the system to accomplish those goals.

    We’re to the point now, however, that most games focus the trinity design because that’s what people expect. Look at how common a “taunt” ability is for tanks in games. Aggro management is a large part of modern games. To paraphrase Richard Bartle, designers ask how they want the trinity to be implemented instead of asking if they should use the trinity design in the first place.

    Nelson wrote:
    Why dismiss collision detection so casually by saying the Internet has lag?

    Probably because the game I run, Meridian 59, has collision detection and I saw that it’s regularly frustrated by lag. FPSes may have collision detection, but in most cases trying to use collision by staying in one place to block an opponent will merely result in the death of the player, probably by a sniper if not by the person they’re trying to block.

    It’s almost as if I’ve played games and thought about it! ;)

    Comment by Psychochild — 21 December, 2009 @ 4:38 PM

  29. isn’t the Holy Trinity really an automatic outcome of the core mechanisms/assumptions that underlay the genre at present?

    I think it’s clear that a group needs a combination of offensive and defensive skills, but it’s unclear why the defensive team needs to be composed of a tank and healer. Some other defensive roles from various games:
    -Crowd controller (EQ enchanter)
    -Defensive buffer (CoH force field defender)
    -Enemy damage debuffer (EQ shaman’s slow spell)
    -Kiter (someone who attracts the enemy and then runs quickly, like an EQ druid)
    -Cannon fodder summoner (like a D&D undead summoning spell)

    And that doesn’t even include things like physical blocking, using terrain defensively, and other possibilities.

    Also, the offensive team doesn’t have to be composed of “DPS classes” in the traditional sense. For example, some games have group combos that do massive damage but require the coordination of several players to execute. Wizard 101 allows players to cast damage buff spells on each other, effectively creating their own combos.

    Basically, there’s room for lots of creativity in combat design, as long as developers are willing to risk trying new approaches.

    Comment by Tolthir — 21 December, 2009 @ 8:47 PM

  30. The way we always played D&D, the monsters would have ripped the heads off the DPS if they could have got to them. When coming up against opponents who could lob spells, the first people they’d go for would be the party’s magic-users. An AI that says “hit whoever is doing the most damage to me” works pretty well for monsters, and “hit whoever could most cause me to lose this fight” works pretty well for smart monsters. The job of the fighters was to stop the monsters getting to the people they wanted to kill by standing in the way, physically blocking access. Geography played a part in combat. The reason fighters were hit was because they were in the way, not because they were taunting the enemy (most of which couldn’t understand what they were saying anyway).

    Spells in D&D were like nuclear weapons, though. Most of the time, the casters would be standing around using missile weapons: they had big spells, but they only had a finite number of them. If your magic missile could take down one enemy in one session, you wanted to make sure it was the right enemy; if you only have half a dozen heals, you don’t want to waste them patching people up for minor cuts and grazes, you want to save them until emergencies.

    The result was that in D&D the fighters tended to do most damage and get hit the most, with the mages saving themselves up for bosses and nasty surprises. Thieves were used for scouting, checking doors, disarming traps and so on. Clerics were second-tier fighters until called upon to heal.

    D&D, though, has telescoped time. You move from encounter to encounter very quickly; furthermore, the encounters can vary considerably and don’t always involve combat. In a virtual world, the fun isn’t figuring out how to open some puzzle door (the solution is on the Internet anyway); the fun is much more in the combat.

    In part, the combat in graphical worlds is a legacy of their textual origins. In text worlds, everyone was in the same location. No-one was “nearer” the enemy than anyone else. This meant that there was no such thing as physical blocking. In graphical worlds, we can have collision detection but the DikuMUD gameplay doesn’t/couldn’t use it. As a result, we get these contrived fights where threat is measured as an abstract quality with little relation to actual threat. The guy in the steel suit is saying rude things about me, I guess I’ll just beat ineffectually on him while those other three guys are hurling fireballs at me and igniting my skin.

    If collision detection played more of a part, then combat would be more of a thinking-on-your-feet kind of thing. Boss encounters would depend more on party make-up and positioning than learning the script. That’s what I’d go for, but then I’m not a fan of fixed character classes so I would say that.

    You have to ask, though: would this actually be any more fun?


    Comment by Richard Bartle — 22 December, 2009 @ 2:10 AM

  31. While far from a perfect system, I think Guild Wars does incorporate a lot of features that pull it away from the Holy Trinity model. Tho, surprisingly, you find a lot of players in GW still trying to use the model.

    1 – no aggro/deaggro skills. No class has a taunt, or a way to reduce aggro/hate. One of the few ways to tank is to pull monsters onto a sharp corner with the tank body blocking, with the rest of the party flush with the wall behind. Its an exploit of the AI pathfinding, but at least it involves some use of terrain, and can only be used if the terrain is right.

    2 – little crowd control. The only real “traditional” CC is snaring, and a few roots. No mezzing/fear/etc. Interrupts are good for stopping big enemy skills/spells, but you essentially have to use them more like in PvP. In hardmode, you can take advantage of the “improved” AI’s ability to recognize being in AoE damage to use such skills to scatter the enemy from a target.

    3 – can respec in any outpost/town. You can change all your stats/skills except what your primary class when not in the combat areas. If you are a Warrior/Elementalist with some points in Earth Magic to buff your armor/damage mitigation, you can switch the points to Air Magic for something more offensive. Or switch your secondary and be War/Necro to perhaps get some lifestealing. Sadly, you rarely see people try to adjust their builds for specific missions/areas (at least in PUGs). But, pretty much every profession has the ability, through the different attribute lines, or secondary prof, to fill different roles, or the same roll in different ways.

    4 – most monsters are built like players. Mobs have skill attributes and a hotbar of skills. Although power creep has happened in the game with more and more monster only skills, and PvE player on skills, for the most part the hardest monster groups to fight are those built with skills balanced like a PvP enemy might be. You can rush the enemy healers/protectors, but their melee and other DPS will try to hit your own squishies. Of course, the AI is never at good at utilizing their skills as enemy players might be.

    Of course, with 1300+ skills and 10 professions (tho, really the first 6 cover most everything), you get a lot of degenerate gameplay. The Holy Trinity may not be used as much, but players find ways to make 8 player, with 8 skills each, team builds which are too devestatingly effective (usually boils down to 6 synergetic damagers and 2 healer/protection). Or silliness like the perma-Shadow Form nonsense that’s been allowed to exist for too long (basically, total immunity to damage and spells).

    Comment by Joiry — 22 December, 2009 @ 8:53 PM

  32. It would indeed be more fun… for some. I’d love more dynamic, tactical combat, but I know I’m in a minority. I suspect the more germane question would be: “Is it fun enough to enough people to be profitable?” Perhaps that’s a bit pedantic, but in a world where MMO design is expensive and time consuming, there really isn’t a huge indie fringe playing around on the edges of the game design map. (Unfortunately, methinketh.)

    Comment by Tesh — 22 December, 2009 @ 8:58 PM

  33. An Ownership Society

    [...] one right answer. (Yes, that’s an obvious statement, but I do feel it needs to be noted. Challenging the status quo of MMO design is sort of a hobby of [...]

    Pingback by Tish Tosh Tesh — 22 December, 2009 @ 10:42 PM

  34. Damion Schubert posted on his blog about this article: The Tank is the Trinity.

    It didn’t send a trackback for some reason.

    Comment by Psychochild — 23 December, 2009 @ 5:04 PM

  35. Here is my take on the whole Trinity thing:

    In summary, more copying of Diablo 2 and less of EverQuest, please.

    Comment by Melf_Himself — 23 December, 2009 @ 9:14 PM

  36. Joiry (31) and Tesh (15) made some good examples that there are already alternatives to the “holy trinity”.

    The idea of attrition, and thus having a focus on avoiding damage, makes players think much more than a tank & spank encounter. Which is horror if tanking fails, and boring if it works perfectly in addition to that.
    We must get rid of the MMO-mentality ASAP: “You have heavy armor, therefore thou shalt not deal damage! But you in the smooth silken robe, you cannot take a hit, but thou are destined to nuke em all!”

    I wonder if Guild Wars 2 expands the successful formula to solo friendly play, without making gameplay too dumb. Basically, being able to fight mobs on your own, without party/npc support. I dare to say there is no place for straight “push red bars up” healing classes and tanks that do little damage there.

    Comment by Longasc — 24 December, 2009 @ 7:00 AM

  37. I really liked your article. Here’s my opinion for what is worth.

    Trinity style play is based on a highly mechanical puzzle oriented approach to combat. Once a trinity style battle has been scoped out, it’s beaten (essentially you use rotations and stop/go at particular script points). At this point repeated battles become reflexive and cease to engage the problem solving part of your mind. This is when the battles stop being engaging and start becoming a grind (a simple set of mechanically repeated optimization tasks = assembly line work). PvP is the opposite. It is lively and organic.

    Because PvP is driven by a recognizable and lively human intelligence, PvP battles are more engaging in the long term than PvE fights. Again, in general it is more engaging in the long term to fight against a lively intelligence than it is to either dismantle or be dominated by a mechanical intelligence. Ideally your AI’s should mimic the lively intelligence of an actual player (mistakes and varying levels of competence included).

    If you create a good set of group PvE AI’s that can really start to mimic the feel of live human opposition, you are in a prime place to prototype new game mechanics, play styles, and classes. I pretty much guarantee that the emergent play style and classes will be something different from and more interesting that the standard trinity with higher level strategic thought and tactics replacing the familiar rotations and pool/timer management of current MMO play. The emergent play would likely be a mix of organic self and group oriented play. It would feel more like you are fighting alongside someone towards similar goals in a real battle.

    A lot of the criticism of better AI design arises from the straw man example of a player fighting a perfect mechanical intelligence. Well, oddly enough, that’s not what players are asking for. What they are asking for is a lively human like intelligence to fight against. That is much different than the god knowledge mechanical intelligence in Daimon’s blog ” it would be trivial for designers to code monster AI that did the smart thing every time. Kill the healers, then kill the DPS, from highest DPS or squishiest to tankiest. If someone casts crowd control, don’t let them cast it twice. Ignore damage output, and target troublesome classes. And above all, don’t divide your damage – have the whole group of enemies burn down one target. Yes, this strategy is effectively what PvP groups do in the arena.” Those are in fact some of the strategies that players attempt to employ in arena play, but imperfect information, imperfect coordination, emotional factors, and mistakes in threat analysis keep this from happening. Ideally, the information used by the AI’s should be based predominantly on information that would be available to a player in the AI’s position (god information and perfect coordination makes the AI seem mechanical). Like players, the AI should base its actions on emotional and rational factors. How rational an AI is can be varied with level. So in the early levels where players are just learning the mechanics, the AI’s should still be fairly rudimentary. As the player levels up, the challenge of the AI should as well. If you want to involve the players more, you can have the NPC opponent’s broadcast any significant actions (Take out that xxx now!) or modifications to their threat tables (e.g. Squad leader is taking fire!). A side benefit of this approach is that PvP combat will not come as a tremendous shock to players who have mostly been playing PvE.

    Comment by lucien — 31 December, 2009 @ 5:46 PM

  38. I should state that you don’t have to start off with a highly developed AI. All you need is a good basic PvP game engine, and good human PvP players that can bias their play style a variety of ways. So, working alongside your PvP group, your first task is to try to develop the environment and some classes that have varied abilities/play styles and whose interaction generates interesting game play. After that, have your PvP group play against a number of players of varying backgrounds (PvE, PvP, casual, etc…) and ability. These players don’t have to know they are playing against human opponents. Have your PvP group vary their play styles, personalities, group coordination, and skill level. Find out what the test players actually like playing against, then go back and break down what your PvP group was actually doing and see if you can duplicate it in code. Iterate the environment, classes, and AI based on this kind of prototyping and you have an excellent chance and finding the game design that leaves the trinity behind. The trinity was developed by accident. The next breakthrough will probably also arise through accident or through intelligent prototyping and experimentation.

    Comment by lucien — 31 December, 2009 @ 9:55 PM

  39. OMG, why do MMOs suck so bad?

    [...] legitimately fascinating concepts proposed for MMOs; for example, I've suggested eliminating levels and the holy trinity, but I've not heard of anyone taking that to use in an MMORPG, let alone offering me a consulting [...]

    Pingback by Psychochild's Blog — 11 January, 2011 @ 3:05 AM

Leave a comment

I value your comment and think the discussions are the best part of this blog. However, there's this scourge called comment spam, so I choose to moderate comments rather than giving filthy spammers any advantage.

If this is your first comment, it will be held for moderation and therefore will not show up immediately. I will approve your comment when I can, usually within a day. Comments should eventually be approved if not spam. If your comment doesn't show up and it wasn't spam, send me an email as the spam catchers might have caught it by accident.

Line and paragraph breaks automatic, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Email Subscription

Get posts by email:

Recent Comments


Search the Blog


November 2019
« Aug    



Standard Disclaimer

I speak only for myself, not for any company.

My Book


Around the Internet

Game and Online Developers

Game News Sites

Game Ranters and Discussion

Help for Businesses

Other Fun Stuff

Quiet (aka Dead) Sites

Posts Copyright Brian Green, aka Psychochild. Comments belong to their authors.

Support me and my work on