Now that everyone has stopped playing with their new holiday toys (yeah, right…), it is time for a long and somewhat discursive lecture on a blog that has been making the usual rounds in the MMO community, from developer Brian Green.
Green rightly invokes the inspiration for the MMO genre, the tabletop RPG, as the standard by which we can improve on the EQ/WoW formula, namely, doing away with the “trinity” of class design, or tank/healer/DPS.
This idea has some merit: the trinity design has been taken to very near its absolute limit, and further innovation will simply be refinements on a single theme. New, groundbreaking styles of gameplay necessarily will have to abandon this approach (which has proven to be both successful and enjoyable) or try different approaches which may lack the trinity’s broad appeal.
Green’s solution, however, is at once novel, inspired, and completely off-the-mark.
He first analyzes three proposals, each with significant drawbacks: either moving away from a class-based design to a skill-based design, or a multiple-role character, or a hybrid of the two.
The skill-based design, while effective, flexible and encouraging to new forms of gameplay, would be difficult to balance. A “flavor-of-the-month” system would ensue, where a creative and intelligent player would find a combination that provides a significant statistical advantage, and other players would naturally gravitate to playing that type of character, until the nerf-bat came… and then they would switch to next month’s overpowered skill combo. Skill-based games become a war between developers and theorycrafters.
The multiple-roles approach is one Square appears to be taking with Final Fantasy XIV. By using a staff as a weapon, one becomes a caster… but switch to a sword, and all of a sudden the player is a melee fighter. Initially, this sounds promising: it allows players to experience all the varied aspects of gameplay without having to reroll and relevel characters. But where is the individuality? If the only solution to flavor-of-the-month skill builds is to make everyone overpowered, and balance the game around that possibility, then the game becomes a cakewalk for developers and a joke for players.
Finally, the hybrid approach suffers from some of the flaws of both systems. This is more or less the approach taken by World of Warcraft‘s dual-spec feature. Players are allowed two sets of skill choices, but choices are restricted to one of three “trees” unique to that class. Most classes are “hybrids” and so can be tank/healer, tank/DPS, healer/caster, etc. Some classes, however, such as “pure DPS” classes, do not benefit from this scheme. Other classes, like paladins or druids, would benefit from a “triple-spec” scheme. Even still, some players would need two PvE specs and a third PvP spec to encompass all levels of gameplay. The idea of “balance” when a person can switch between pre-determined modes is still a problem, as is the problem of eliminating class “flavor” in favor of bland uniformity so that no one feels left out.
Green’s ultimate point is that MMOs should be more like tabletop RPGs. He says:
Given the history we find in D&D, we aren’t forced to abandon familiar combat systems in order to replace the trinity class design; however, we do need to change what we expect from individual characters. The two main concepts for guiding a different design are to eliminate specialized roles and allow the use of tactical options in combat.
Eliminating specialized roles would mean that the “role” would not need to be eliminated per se, but rather roles would need to be trivialized. For instance, if a dedicated healer character is no longer needed, then potions, bandages, and rest breaks would have to take the place of the healer. Players would avoid healing classes because being a glorified potion dispenser is not enticing. Currently, games like WoW entice people to be healers because a good healer is always in high demand and can write a ticket to the top-flight guild of his or her choice.
Similarly, many players might eschew tanking or DPSing if those roles are somehow able to filled by another. Why play a light-armored melee fighter if a heavy-armored one can do equivalent damage and have higher survivability?
Another problem with this set-up would be that fights would be somewhat trivialized to endurance battles between high boss hit-point pools and the massive amounts of damage, healing and mitigation capable by every party member. Consider something like a standard JRPG: the main “player” character is usually a heavy-armor melee fighter with high health, high mitigation, and the ability to use magic, especially healing magic (Cloud with the right materia, Terra or Locke with Espers, or Cecil). Even Rosa in Final Fantasy IV, the standard “white mage,” was capable of dealing not-insignificant damage. Most fights consisted of everyone autoattacking, unless the player faced a boss, in which case Curaga might need to be used occasionally. While this works well enough for single-player consoles, it would be deathly boring as an MMO.
Tabletop games are also a different beast than MMOs. The unique and salient feature of a tabletop game is that the NPCs are really “played” by a living, thinking, reacting human. MMO opponents, if not in PvP, are dumb robots. They may be exceptionally-well-programmed dumb robots, but they are still dumb robots. MMO battles are essentially a predictable, graphable numbers game, and as with any rule-based system, there is a way to exploit those rules to the maximum benefit of the player. Theorycrafters will always discover this, and they will always share this knowledge. Developers, to counteract this, must design encounters in such a way that pure numbers are enough, but make those numbers difficult enough for most people to reach that it requires coordination and effort to play the game at the highest levels.
So, if MMOs cannot be tabletop RPGs because of computer-controlled enemies in PvE, and skill-based, gear-based, or hybrid systems necessarily fail as substitutes for the “trinity” because they do not do away with the underlying mechanics, merely redefine the roles played in such a way that the “trinity” aspects are less noticeable, are we forever stuck with this mechanic?
Every encounter in every role-playing game is a numbers game: do more damage than the monster/enemy while taking less or healing through the damage. As long as combat encounters are defined in this way, the “trinity” aspect must be addressed in some fashion. Finding new ways of doing that is insufficient on its own; developers must find new ways of implementing the trinity aspects so that each fight tests the skills of trinitarians in new and unique ways. Attend.
If the goal of the tank is to hold aggro and mitigate damage, then the tank provides two relations to DPS and healer performance. The tank effectively sets a DPS cap for group-wide DPS: out-DPS the tank, and aggro gets pulled, leading to a wipe (potentially). The tank sets a “threat ceiling” that must always increase or else bosses will not die fast enough.
Since healer resources (mana, for instance) are limited, there is a finite amount of healing that can be done. If a tank fails to mitigate enough damage, the healers will be overtaxed and the party will fail.
Consequently, healers also stand in relation to the entire group such that if a healer does not pump out enough healing energy, people will die. Every lost member loses a total amount of mitigation or damage, making a wipe more likely. DPS stands in the relation to the tanks that DPS must always provide damage and debuffs to aid the tank, while working fast enough that healer resource management is possible.
This creates a “coordination tension” among players: every trinitarian role
must accomplish “its” job or the party fails. The problem from a class design perspective might be how to rebalance this, but from a player perspective, the problem is how to divide up the “jobs” in such a way that maximizes player enjoyment while still making encounters difficult enough to challenge players.
Green suggests that “eliminating explicit roles means that players are not forced into a specific role through class choices or game design requirements.” But what does this mean? Could a group of say, five rogues, walk into a WoW heroic 5-man instance and expect to down bosses? If so, then we need to have some ways for rogues to mitigate damage well… but if rogues can mitigate damage well, why would a party ever take a paladin or warrior to tank when they could have a rogue output rogue-like damage while avoiding more damage than a plate-tank could mitigate?
The answer is that no one would: a game simply cannot be “balanced” around an infinite variation of class or skill combinations, because some “job” would be left undone.
Every party, then, must strike some sort of balance. We approach the problem from the wrong direction if we think that the answer is to universalize certain features so that any party mixture is “viable.” Even in table-top gaming, a party consisting of all heavy-armor melee fighters would run into difficulty completing dungeons as soon as the DM decides that a trip-wire connected to an arrow trap should confront the party.
On the other hand, no encounter can be designed so that it is undefeatable. Our aforementioned party of heavily-armored melee fighters might choose to trigger the trap in a way that did not expose them to danger, such as throwing a rock through the trip wire and ducking out of the way. While this works in a table-top game because the DM can react to novel suggestions by the players in real time, it would be difficult for encounter designers to anticipate all of these novel approaches ahead of time. And even if they were to attempt such a Herculean task, the first forum post explaining how to do it would become the de facto solution from there on out.
The solution, then, lies somewhere between Green’s idealistic phantasy and the much more practical, cynical view of this article: abandoning the tank/healer/DPS “job” system is impossible so long as combat is defined as doing more damage than damage coming in or healed. Allowing each member to do the job of the other universalizes classes and makes distinctions between them useless, while trivializing gameplay. Allowing unique skill combinations to create self-designed hybrid classes creates class imbalance; it is the equivalent of letting players design class mechanics.
Allowing for on-demand job changes to ensure property party makeup proliferates bad playing and does not address the central problem, which is the avoidance of the standard party system. Redesigning encounter mechanics so specific party make-ups are not necessary would lead to certain classes being favored, or all encounters becoming so predictable that enjoyment was leeched out of them.
In order to meet Green’s diverse goals, the answer is to abandon “pure” job roles and to encourage designer-made hybrids. For instance, let us split the roles into five broad categories: the tank, the melee fighter, the ranged fighter, the magic user, and the healer.
Now, we must combine each of these: a tank/melee fighter is a heavily armored fighter, capable of putting out high damage or mitigating/countering incoming damage. The answer to balance is not to have them permanently select different skills or to obtain different sets of gear, but to have a single set of skills that are active and reactive to differing situations. If the fight is a burn phase and calls for high DPS, a certain set of skills will be used. If the fighter is taking large amounts of damage, defensive postures and skills will be used. A tank/ranged fighter would be better at applying debuffs, kiting, and positioning the enemy in such a way as to draw danger away from her party members. A tank/caster hybrid would focus on using defensive spells to mitigate damage, but would be capable of using offensive spells to output damage if necessary. A tank/healer would be capable of using healing power to sustain himself against large amounts of damage while outputting healing to party members, but might not be able to throw out a significant amount of damage.
Damage dealers would specialize in single-target, area-of-effect, or applying debuffs to enemies. A ranged physical/magical damage dealer, for example, might have the ability to do high single-target magical damage, but have high area of effect physical damage as well. Or perhaps a hybrid melee/caster class could apply debuffs as melee range, then move to a caster range, and use magic to cause high area-of-effect damage based on those debuffs.
Continue the example through each iteration of the combination: everyone is capable of fulfilling multiple roles within a single encounter, so long as they switch their skillsets on the fly. This would make MMOs less like RPGs (where abilities must be planned out beforehand) and more like action-adventure or FPS games where new abilities and tactics must be selected on the fly. The tank/healer/DPS mold might work for the first part of the encounter, but what about the second or third phases? In those, healing might be prioritized, or damage. As long as no one is locked into a single job within a single encounter, players are given the flexibility to be creative without having to over-generalize.
Class balance is likewise not a problem, because no one is a “triple-threat.” A high mitigation healer, for instance, is incapable of putting out too much damage, and so neither PvP nor PvE viability is threatened. Similarly, a high mitigation ranged fighter is incapable of putting out enough healing to sustain herself through a concentrated damage dealing assault.
By making class roles dependent on the button pressed, we test the skill of the player, not the player’s ability to “game the system,” as it were. By making certain skills proactive, and certain skills reactive, we test the player’s ability to pay attention and react in real time. By allowing players to fulfill multiple roles within a single fight, we allow for a broader perspective of encounter design, where aspects like positioning, player movement, fast reaction, etc., all become more important than gear and mastery of a single aspect of the game. It encourages a very wide variety of character types, prevents over-powered classes from arising, since each class is powerful in its own respect if played well, provides novel replay value and ease of direction for player advancement, and allows for ready identification of a player’s likely role within a party.
In other words, it meets Green’s design criteria while avoiding the idealism of his proposals. We do not abandon the trinity, but we redefine what each role in the trinity encompasses in terms of solo and party dynamics by providing a choice to the player, not at some pre-determined level, but directly in the encounter itself.