Warrior, Wizard, Thief: Class, Balance and Archetypes in Computer RPGs
Two central issues for any fantasy RPG are: how well does it represent the archetypes (warrior, wizard, thief) and how well balanced are the archetypes in relation to one another in terms of gameplay.
If you're not familiar with the term archetype, it basically refers to a deeply ingrained intuition about how a certain type of person or thing is. The term was popularized by Carl Jung to refer to deep psychological structures, but it has been extensively adopted by RPG gamers to describe certain kinds of player characters.
In relation to fantasy RPGS, archetype loosely refers to the way that genre stereotypes are represented in a game. The three most important, or generic, stereotypes are warriors, wizards, and thieves, although there are numerous additional sub-classifications that one might describe as being archetypal.
Obviously, there is a close relationship between archetypes and classes, and the terms are sometimes used interchangeably, but they are not exactly the same thing. In this article I'd like to investigate the relationship between archetypes, their implementation in computer role-playing games (either as classes or derived from classless systems), and how these representations impact on game design and player satisfaction.
Archetypes may be explicitly implemented in a fantasy RPG as a limited set of static classes, from which the player must choose at the time of character creation, or the player may be left to define their own class dynamically through skill selection. (Many games use hybrid class systems which allow both class selection and class permutation.)
How well a game models the archetypes, either through classes, or through free skill selection, and how well balanced these classes are against one another often determines how well a game meets the players expectations as a fantasy RPG. A game that does a good job modeling the classes and their interrelations is a 'better' fantasy RPG than a game that does not do a good job modeling them, all other things being equal.
Developers typically implement a static class system in order to impose meaningful choice on a player. How the player plays the game (the specific strategies that they use) is determined largely by the class they select when creating their character. By imposing choice from the outset, the player is guaranteed a variety of experiences: if they play it once and then discuss the game with other players who chose a different class, they will experience that variety vicariously by hearing about the different strategies that were used and the different effects that were achieved. If they play the game through a second, third, or fourth time themselves, choosing a different class each time, then they will experience that variety directly by being forced to chose different strategies on each play-through.
Developers also choose a statically classed design in order to appeal to a broader range of players. Players who have particular preferences about the archetypes are more likely to be drawn to a game that presents them a range of choices that includes their preferences.
A statically classed game provides meaningful gameplay variety in multiple play-throughs, and appeals to a broad range of players. It's utility to developers, therefore, is obvious. Some players, however, feel this strict interpretation of archetypes to be somewhat limiting. They feel 'hemmed in' by the classes and rankle at the perception that the developers have imposed their interpretations of the archetypes on the player at the expense of the player's freedom and enjoyment. They feel pressured to role-play their characters in a certain way which does not match their interests and intentions.
Games that allow the player to generate their character dynamically in a classless, or skill-based system, are generating their gameplay variety using a different technique: instead of creating a limited set of distinct gameplay variations, they are creating a wide range of variations with minor distinctions between them.
In a dynamically classed game, no two players are going to have exactly the same experience; at the same time, the differences between those experiences are going to be less distinct than they are in a statically classed system. Each player is free to pursue their own development, either by conforming to their understanding of an archetype, or by intentionally going against the archetype in some unique combination of skills. By designing a classless system, the developers can appeal to an even broader range of players than a statically classed system, by appealing to not only archetype enthusiasts, but RP pioneers who prefer that their characters strike out in their own, unique directions.
Some players experience this freedom as liberating, a mechanic which allows them to play exactly the kind of character that they want. Other players experience this freedom as a kind of neutering, in which all players are essentially reduced to playing a single class: the Jack-of-All-Trades (JOAT). The individuality of the classes, which provides the genre with it's distinctive motifs and flavor, seems to have been sacrificed in the interest of making a certain number of other players happy.
Hybrid Class Systems
Developers have attempted to satisfy both types of players over the years by implementing various hybrids of class and classless systems. Typically, these systems allow the player to pursue a greater range of activities than a statically classed system, but at a penalty.
In Bethesda Softworks' Oblivion, for example, a player's class selection determines which skills contribute to leveling up. In this system, players may use any skill, regardless of class, but only class skills contribute to advancement. In other games, players may use other skills, but at a reduced level of proficiency, or they may have access to only a limited number of the privileges typically accorded to a skill. The intent in designing systems of this nature is to preserve the distinctiveness of the archetypes and make them appealing without curtailing the player's sense of freedom.
Class Mechanics and Gameplay
Regardless of the specific mechanic a game employs, most players expect these archetypes to be well balanced: they expect to be able to complete the objectives of the game with the same amount of effort, and with the same amount of enjoyment, regardless of the specific class that they choose.
In a statically classed design, balance can be achieved in a relatively straightforward manner: the specific strategies each class will use have been decided upon beforehand by the developers, and, by comparing things like damage output, 'hit points', 'armor class', attack speed and reach, and a number of other factors, developers can more or less determine the relative balance of the classes statistically.
In a classless system, developers are faced with an added layer of complexity, or rather, an element of uncertainty: the player's judgement; his or her ability to build a successful class from the options that are provided.
In a statically classed system, the developers can decide arbitrarily what offensive and defensive capabilities a player possesses. If a class is found to be too weak in one area, the developers may strengthen that area, or shore up another area, to compensate. Additionally, each class may be given a certain number of utility skills which provide the player with access to gameplay experiences that the developers consider to be integral to their product; for example, the ability to set or disarm traps, identify magical items, brew potions, bash locks, haggle with merchants, or intimidate NPCs. These utility functions, while generally not as important as primary offensive and defensive functions, can help to offset differences between the classes while preserving the specific character or flavor of the archetypes.
In a classless system, the developers do not have the same degree of control over the end result. Players may choose, for example, to build a class that is entirely offensive and end up making themselves too vulnerable. Or, they may build a class which has strong defensive capabilities, but which utterly lacks offensive firepower. The developers may discover that, in the interest of creating the most powerful combatant possible, a player has designed a character completely lacking in utility functions, thus closing themselves off from large areas of gameplay that the developers want the player to explore. These design challenges are not trivial and cannot be solved by tweaking numbers alone.
Rock, Paper, Scissors
In a statically classed system, the developers may assign each class a primary offensive capability, defensive capability, and utility. One class may be stronger than another in one area or another, but each class will be given enough in each category to complete the game with an appropriate amount of challenge.
In a dynamically classed system, it is possible for a player to build a class which possesses too much strength in one area or too little strength in another. Let's take for example a simple statically classed system which uses three archetypes: a warrior, a wizard, and a thief.
The warrior's primary offensive capability is their skill with a sword, which we'll rank as the second most powerful in the game. Their primary defensive capability is their armor, which we'll rank as the most powerful. Additionally, we'll give them a utility function: improving weapons and armor, which we'll rank as the least powerful.
The wizard, in turn, will receive a fireball as an offensive capability, which does the most damage, earning them the highest class ranking in offensive. The wizard will receive a shield spell, which we'll rank as the least powerful defense, as it's not very strong. Finally, we'll give them a utility skill which allows them to identify magic items, a skill which is better than the warrior's, because enchanted items can't be used until identified, but not as good as the thief's, earning them second place in utility.
The thief has the lowest damage output with her bow, but her ability to dodge attacks, while not as great as the fighter's ability to absorb damage, is better than the wizard's shield spell. To compensate, the thief has the best utility skill, the ability to pick locks and gain access to more powerful magical items than either the warrior or the wizard. These additional objects give the thief greater access to combat supplements like potions, scrolls, rings, etc.
This is a fairly simplistic and straightforward setup, and it's obvious that developers will not have much difficulty balancing the classes to provide an optimal gameplay experience. Now, let's see what happens when we turn this into a dynamically classed game.
Using the same archetypes (warrior, wizard, thief) and the same skills, we'll allow the players to create their own classes dynamically. One player chooses a character who can use a sword, bow, and cast fireballs. This character has no problem defeating every enemy in the game if they play well, though they might find that they die easily if they make any mistakes (owing to no defensive capabilities) and that they have been restricted from enjoying many gameplay experiences made available through the use of utility skills. A second player decides to create a character that can wear heavy armor, dodge, and raise magic shields. This player finds that combat is dull and monotonous, that their blows seem ineffective, and that the battles last forever. They also experience a lack of non-combat related activities just as the first player does. A third player chooses the equipment upgrade skill, the ability to identify magic items, and the ability to pick locks. While their non-combat experiences may be richly rewarding, they will quickly discover that combat is a nightmare and to be avoided at all costs. For this player, the game seems horribly unbalanced in favor of the NPC enemies.
These three examples represent three 'extremes' of dynamic class selection, the 'all offensive', 'all defensive', and 'all utility' builds; but the problem by no means disappears when we use other combinations. In fact, there are even worse combinations than these.
Let's pretend we've selected Fireball, Heavy Armor, and Picking Locks. This character will literally faceroll the game. They have the best offensive capability, the best defensive capability, and the best utility function. For this player, the game will seem far too easy and unbalanced. Enemies fall swiftly, the player walks through combat unscathed, and he collects all the best loot in the game. A second player chooses Archery, Mystic Shield, and Equipment Upgrade. To this player, this seems like an unusual and interesting combination. Unfortunately, they will discover that enemies are hard to kill, damage them easily, and that they have been prevented from acquiring the best loot in the game. For this player, the game will seem far too hard and equally unbalanced. These two builds represent the 'best of the best' and the 'worst of the worst'.
We've already seen five skill combinations that are sub-optimal in this classless design; in fact, there are only three combinations that actually work very well: the original class archetypes that the dynamically classed system is based on are the only skill combinations that provide an optimum amount of challenge and reward for most players.
Balancing Classless Classes
The only way to 'solve' this problem is to make each skill provide the player with the exact same degree of benefit. In other words: each skill has to be balanced to every other skill; the skills, in this design, take the place of the classes in the static design.
That doesn't mean that each has to provide the exact same offensive, defensive, and utility capabilities, only that each must provide the player with some benefit that prevents the player from creating a character that is ineffectual in any one area: offensive, defensive, or utility.
The result is a collection of skills that each provide some mix of offensive, defensive, and utility functions so that even if all the skills selected by the player are utility functions, they will still receive offensive and defensive capabilities from pursuing them that are more or less equal to players who choose all offensive or defensive capabilities. In other words, if you're a designer and you want to design a dynamically classed game, you have to make a player who chooses Equipment Upgrade, Identify, and Lockpicking as powerful as a player who chooses Fireball, Heavy Armor, and Lockpicking. If you don't, then the player who chooses the first skill set will be at a disadvantage compared to the latter.
You may feel inclined to object to this kind of design. A player who chooses Equipment Upgrade, Identify, and Lockpicking should not be as powerful on the field of battle as a player who chooses Fireball, Heavy Armor, and Lockpicking. Your intuitions about the archetypes make this character build seem unlikely, to say the least. You can accept the class builds in the simple, statically classed system outlined earlier (warrior, wizard, thief) because they match your intuitions about the archetypes; when those archetypes are dissolved in a classless system, certain combinations of skills no longer make sense, from a gameplay perspective. Your first instinct, then, may be to return to a class-based system, either authoritatively, through the implementation of a rigid game mechanic, or rhetorically, by persuading people to voluntarily recreate the archetypes through their skill selections.
What we end up with is a stand off: players who are in favor of freedom don't want that freedom curtailed by a statically classed system, or by a combat mechanic that seems unbalanced because they chose to emphasize utility skills over offensive and defensive capabilities; on the other side of the spectrum we have players who don't want those utility functions bolstered to the point that they can serve as replacements for archetypal offensive and defensive capabilities. These players are already intelligently choosing offensive and defensive skills that provide them with the ability to master the battlefield; to bolster these skills with utility skills that perform the same function not only makes the game too easy, it is offensive to reason: why should a blacksmith be as powerful on the battlefield as a warrior, the veteran of many battles? By buffing utility skills and turning them into combat skills, the amount of meaningful choice is reduced. If any combination of skills produces the same results, there is no longer any point in choosing. Choosing skills becomes a cosmetic activity, allowing you to role-play different types of characters that are all essentially the same.
Archetype and Expectation
So why doesn't this rule apply to statically-classed games like our three class example? Why don't players feel that they are being short-changed by their selection? After all, no matter what class the player chooses, they will all have more or less the same chance of succeeding at the game. Why is a thief allowed to be as effective in combat as a warrior or a wizard? The answer, I believe, is two part.
First, because the developers have intentionally designed the classes in such a way that it is impossible for the player to choose a class that is too weak or too strong--each offensive, defensive, and utility capability has been balanced against equivalent skills in the other classes, so while one class may logically be superior to another in one area, equally logical considerations allow them to be superior in others. Players can easily rationalize the differences between the classes in a way that supports the logic of the mechanics.
Second, because these classes match our intuitions about the archetypes. They represent the 'poles' of character design, and are different enough and distinctive enough that the player doesn't feel compelled to draw comparisons between them. Players accept their differences and their equivalence as a formal convention of the genre. Their experiences are different enough when playing them that they don't experience the decisions involved in balancing the classes as having the effect of leveling the classes.
In a dynamically classed system, the developers are not so fortunate. Without clearly demarcated archetypes, the player's attention is naturally drawn to comparisons between the skills. Players are naturally curious to see if one particular constellation of skills is superior to another. If they are, then the skills seem unbalanced. If they aren't, then the resulting characters seem illogical. The innumerable arguments over the trifecta of crafting (Smithing, Enchantment, and Alchemy), the 'nurfing' of Destruction, and the 'self-gimping' of investing in thieving skills (Lockpicking, Pickpocketing, and Speech) in the Skyrim forums are eloquent examples of the kinds of problems that may arise.
Crafting, Destruction, and Stealth: Real World Examples
Crafting in Skyrim comes under considerable fire because it allows the player to supercede the development of combat skills. A master craftsman is as effective in combat as a master swordsman. While illogical, the developers clearly intended this design to be a method of ensuring gameplay balance in an arena of unpredictable builds. This represents the developers attempt to equalize players who chose a utility-heavy character build.
Destruction, by contrast, comes under fire for being too weak. A player who wants to play an archetypal elemental wizard is disappointed by the mismatch between their expectation and the actual result of their skill selections. It appears that here, perhaps, the developers have reduced the efficacy of the skill in an attempt to prevent players from becoming heavily armored walking tanks with flamethrowers facerolling everything before them.
Thieving skills, by contrast, in spite of developer efforts to bolster them with combat-related perks, still fail to restore equilibrium with combat-only builds, and players often feel penalized for pursuing them. What we have then is a group of utility skills that are overpowered (Smithing, Enchantment, Alchemy), a group of utility skills that are underpowered (Lockpicking, Pickpocketing, Speech), and a combat skill that is underpowered (Destruction). Could the developers have resolved these issues?
Let's take a look at what happens if we buff Destruction and restore it to its archetypal place as the most powerful offensive capability. For players who choose to play archetypal wizards, the result is positive: their offensive capabilities now suit their weaker defensive capabilities. But what about a player who chooses to build a character that does not match the archetype but instead uses Destruction and Heavy Armor? Suddenly, we are back at the old problem of combining two powerful skills that we encountered in our simpler model and the player becomes overpowered.
Now let's look at what happens if we 'nurf' crafting. A player who uses crafting in conjunction with combat or offensive magic is pleased because their non-crafting skills no longer seem redundant in comparison to their gear. The game feels more balanced. But what about the player who chooses to RP a utility character? All of a sudden, they can't play the game the way they want to play it because they are being penalized for choosing utility skills. In fact, they are now in the same boat as the stealth-oriented players who choose to focus on Lockpicking, Pickpocketing and Speech.
What happens if we buff the stealth skills (Lockpicking, Pickpocketing, and Speech) and give them capabilities that put them on par with the other skills? For example, the ability at low levels to acquire access to powerful weapons and armor, or to remove powerful weapons and armor from opponents. Will players now complain that using these skills in conjunction with other offensive and defensive skills seems too powerful? Have we now made a variation on an overpowered character that uses stealth instead of crafting?
Dimensions of Gameplay
The challenge presented by designing a dynamically classed system is, as I said at the start of this essay, not a trivial one. Developers are forced to choose between allowing players to create unevenly balanced characters or characters that break fantasy archetype conventions, two of the most important factors in determining how well players like your product. You can't make destruction magic the most powerful offensive capability in the game without making the game unbalanced for players who don't choose it. If it isn't more powerful than the alternatives, it doesn't match your intuitions about the archetype. In other words, balance is pitted against expectation.
The problem is not entirely insoluble, and the solution is to remove the conflict between the player's expectations and the need for balance by removing the emphasis from the single gameplay dimension of combat. Instead of trying to make utility functions serve primary offensive and defensive capabilities, the developers need to take the challenge to the next level and create a broader range of in-game activities so that pursuing utility functions provides its own rewards.
The developers could, for example, provide utility skills with non-combat privileges that are desirable to players in their own right; for example, the ability to use persuasion to avoid combat or to acquire deeper and more meaningful relationships with NPCs, or the ability to use lockpicking to gain access to tunnels that circumvent enemies or to gain access to 'secret areas' that are inaccessible to other players. If the enjoyment derived from pursuing these gameplay experiences exceeds the frustration players experience in combat, many players will gladly accept the trade offs.
The key is making these other non-combat activities enjoyable activities in their own right and of equal importance to the game. If your main narrative requires a steady stream of combat to complete, and does not provide meaningful choice in non-combat solutions, then you will not achieve this objective. The players have to feel that, in spite of their choices, and the sacrifices required of them, that their experience was every bit as enjoyable, exciting, and engaging as it is to the player who choses to pursue a more traditional, combat-oriented path.
If developers fail to seize this opportunity and instead choose to restrict their attempts to correct these imbalances to the arena of combat alone, they are going to continue to encounter the same complaints and objections from players.
The problem isn't that a classless system is doomed to fail in one area or another but that it represents a more complex model of the world and must take into account a wider range of variables and address a greater range of needs. By deepening the representation of the fantasy archetypes, by providing thieves with richer thieving experiences, wizards with subtler magic experiences, and warriors with more demanding combat experiences the developers can appeal to a wider range of players without sacrificing balance or discrediting the archetypes. The solution is not to reduce gameplay to a perfectly balanced plane of combat bonuses and penalties, but to expand gameplay to new dimensions of activity where these tradeoffs are no longer required and no longer make sense. If you're going to give players the option of creating new archetypes, you need to give those archetypes a world to live in.