The Specialization Trap

A trap that many ARPG designs fall into is overspecialization in the face of relatively permanent character advancement decisions. Path of Exile is a great illustration of the pitfalls of optimal play revolving around specializing your character using permanent decisions over a long span of play time.

PoE has an extremely large and intricate skill tree which allows significant specialization: down to the weapon type level--swords, axes, etc and element (fire, lightning, etc) level. You'll need to look for specific weapons and items that suit your specialization choices.

This is damaging to the ARPG model. ARPGs are a combination of various treadmills that complement one another: loot, character level, ability levels, and perhaps a few more (Din's Curse had "reputation" which would grant you a relatively viable rare weapon upon leveling up). An emphasis on specialization limits the character construction decisions (once you've picked the specialization, you jump pump points into that indefinitely) and loot decisions, cutting off significant branches of the decision tree. The more equipment-related specialization, the fewer pieces of loot a character will find that can possibly be upgrades. A character can easily go from having a 20% chance of finding a relevant piece of loot because he specializes in one-handed weapons, to a less than 5% chance because he has been forced to specialize further, perhaps into one-handed swords, in order to continue getting damage and/or accuracy bonuses from new passive skills.

ARPG players are familiar and fond of systems that involve an extreme amount of character choice permanence. Diablo II's attribute and skill points were non-refundable, and the pattern of severely punishing the uninformed by making character decisions irreversible does not enter the consideration of the majority of ARPG-players who grew up playing Diablo and Diablo II. Titan Quest and Din's Curse made a step forward in this regard by allowing point-by-point respecs at escalating prices. Torchlight, the most casual-friendly ARPG I have played, doesn't have respecs built into the game at all--the only way to respec is to use a console command to spawn a respec potion.

Overspecialization leads to minute-by-minute play being less interesting. If you need to invest a high percentage of your points into a very small number of abilities to make them effective, then you're naturally going to be filtering your ability choices by which abilities you have spent points on. Abilities available are further filtered by what loot you have available to you. In the majority of ARPGs you end up in a situation where you have at most two or three useful abilities--sometimes you'll only have one. When you have no more than two or three tools to work with, it's harder for combat to remain engaging.

Spamming a single ability and watching everything die can only remain fun for a little while unless you are specifically looking for a relaxing experience. Unfortunately, the theorycrafting involved in making an adequate character often is beyond the interests of a relaxed player, so they wouldn't get to the point where they could effectively do one ability spam. The mechanical systems thus lead to patterns of play which are not appealing to the kinds of players that have the capacity to use that style of play.

Permanent point-investment schemes also lead to perverse incentives for building characters. You really shouldn't spend points on skills beyond what you specifically need to get to that highest level skill that you want to specialize in. When optimal play is to not participate in the character advancement and be underpowered for tens of hours so that you can be somewhat above average later, the game clearly suffers.

The reasoning and evidence above would indicate that specialization-focused design is poor design.

So why do players seem to like it so much? A game like Diablo II is a complexly layered system of rewards that vary in intensity and frequency in such a way as to draw us in and addict us. Players have trouble separating the fun and not fun mechanics of a game when they are layered as they are in Diablo II. The rewards systems are strong and interleaved into all other mechanics of the game, so it's hard to separate skill tree manipulation and loot sifting from the enjoyable feeling of character progress. Those activities do contribute to character progress, but their design isn't trivial to separate from that positive feeling.

Good ARPG design is much more than simply causing the player to have the positive feeling of character progress, it's optimizing that feeling to happen in as intense and frequent a way as possible without it being diluted by overexposure. It's a difficult balancing and timing act, and different players have higher engagement at different points along the spectrum between constant rewards and rare rewards, large rewards and small rewards.

In games that feature overspecialization, you'll notice that combat tends to reduce to the repeated use of a couple of skills at most. This trivializes combat and turns it into a chore. When combat is a chore, in order to enjoy the game you must enjoy the minmaxing of character construction, which in ARPGs is done via loot sifting and planning/spending skill and attribute points. So the games' fanbases naturally require their members to enjoy that minmaxing and not mind somewhat boring combat.

ARPGs can have exciting combat and enjoyable character advancement. Diablo III is proof. I hope to discuss Diablo III's success in a future article.

Depth Perception: Casual and Hardcore Views of Depth

A game’s depth can be defined as the amount of “stuff” that the player can learn about the game.

A shallow game has few things to learn. Think of when you first learned algebra. Because you didn’t understand the concepts of algebra, you would engage in a guessing game to find the value of x in simple expressions like “x - 10 = 3.5”. Not long thereafter, your teacher taught you that there was a method you could employ that could obviate the guessing: simply “undo” the operation on the left side and carry its effect over to the right side (“x – 10 + 10 = 3.5 + 10”). Now the guessing game isn’t a game anymore, it’s a rote task. If someone gave you 50 problems of similar complexity to solve, you would previously have had an interesting (debatably) task of excitedly guessing what x might be. Now that you know how to do it, you just mechanically run through the problems subtracting or adding numbers to both sides and receiving an result. There’s nothing left to learn there. So clearly simple algebra isn’t a deep game.

The scary part comes when you realize that you can think of World of Warcraft as algebra. You’ve got a few abilities that you can use and you’ve got enemies to kill. The first few times you fight, you’ll try different methods—essentially trying to guess the answer to “how can I make my relative DPS greater than theirs?” But once you’ve figured out a reasonably effective method, you no longer have to think. You only have to press 1-3-4-2-4-… until the enemy is dead or you are dead. The monotony is only broken when you move to a radically different kind of enemy (usually it’s a difference in magnitude of enemy DPS, so your remaining health will be the only thing to change between the old “easy” enemies and the new “harder” enemies) or when you level up and gain new abilities which may or may not actually impact this process.

Playing WoW through the course of one level doesn’t have much depth. The little bump in depth you get out of advancing doesn’t help for long, either. It’s just like learning that you can undo division with multiplication on the next homework after you mastered undoing addition with subtraction. Regardless of that bump in complexity, you still must solve those same problems (kill those same enemies) 50 times before the teacher will give you a passing grade and let you move on to the next-most-difficult homework (move on to the next level).

There’s a crucial distinction here: Depth is not the number of problems you could conceivably solve, it’s the number of conceivable techniques you could viably use to solve problems.

Players perceive these two different phenomena in MMOs and attribute both of them to the concept of depth.

In easy games like WoW, depth is artificially created through making enemies always too easy. Players are never forced to explore much of the game’s strategy space—even in the kiddie pool there can be a “deep end.”

This definition of depth only serves the “serious fun” or competitive style of gamer. To a casual player, such depth is just annoying. A casual player can’t spend the time to explore the strategy-space—he just wants to move inexorably forward and be tickled by something interesting occsionally as he receives a stream of ego-boosting rewards at a regular clip.

To the casual player, depth becomes the sheer amount of content. The casual player does not care about the advancement and evolution of strategies. They don’t want to have to think that much; games are distractions and ways to pleasantly pass time. Casual players only care that there is more to do—that the pleasant rollercoaster ride continues until he wants it to stop.

So we’re left with two definitions of depth that are useful when designing games for different kinds of players:

• Having more to do.
  • Character advancement.
  • Game as meditation.
  • Relaxed play.
  • “I must lay 300 more bricks to finish this wall.”
• Having more to learn.
  • Personal advancement.
  • Game as intellectual exercise.
  • Serious play.
  • “I must learn how to effectively approach enemy positions in Go to improve my early-game.”

Visualizing the Interactions of Depth

I see these two kinds of depth as two faces of a shape that represents the amount of fun that can be had in a game. Casual players look down one face of the shape, their depth is the shapes’ width, while hardcore players look down another face, their depth is the shapes’ height. The two are manifestations of different ways people enjoy games.

This means that games can be seen as two-dimensional objects in “depth-space”. The x-axis represents the amount of content completed, and on the y-axis is strategic understanding. Each shape represents the depth profile of the game—the width of the shape at a certain value of y indicates the amount of content that can be completed with that level of understanding; the height of the shape at a certain value of x represents the amount of knowledge that the player can gain having completed a certain amount of content in the game.

A game like WoW stretches a long way on the x-axis, but only gets high on the y-axis later. Chess, on the other hand, has a significantly taller shape, but there is not much width at any one point along the y-axis. This is because you have to engage with more of the strategy-space (learn more about the game) in order to unlock more content in chess. WoW is wider at all points on the y-axis because there is a lot you can accomplish with even a severely limited amount of strategic knowledge.

This visualization technique is not perfect. Aside from the general abstractness of it (a lack of metrics), it’s definitely not true that the two kinds of depth are orthogonal. Playing through more content usually increases your understanding of the strategy-space, which will gradually advance you along the y-axis naturally until you reach a saturation point that’s probably dictated by your intelligence.

These Depths Converge in the Long-Run

These two conceptions merge as the casual player plays the game more and must confront either the fact that the game is no longer stimulating or the fact that they cannot achieve their goals without digging deeper into the strategy-space. So games aimed at casuals do need to have some degree of traditional depth to sustain them long-term—though casual games are usually aimed at children and non-gamers—people who are not going to spend enough time in the game to hit the “no learning left” issue.

Suddenly, Bioware is Incompetent.

This is one of the few times I’ll jump on an issue that is a hot topic on MMO blogs. I’m doing it now because certain prominent bloggers’ opinions on this topic have been too ridiculous for me to bear. Perhaps Keen’s numbers haven’t been good recently and he wants to start a huge debate to draw links and views?

Bioware has released information on companion characters. For some unknown reason, this fact has led people to believe that Bioware cannot tie their design shoes anymore.

With very little actual information available, people are assuming that Bioware are going to mangle the design of companions and ruin their game by removing one or more “M”s from “MMORPG.”

Arguments against companions:

1. It throws off rewards because you will get loot as if you were two players.
2. Who would choose to bring someone along if they’ve already got an AI companion?
3. Everyone will always have their companion out, so the game will be balance for that, rendering non-companion strategies non-viable.
4. The game won’t be balanced for rampant companion use, so the game will fall apart as everything becomes extraordinarily easy due to companion use. This makes non-companion strategies non-viable.

All of these complaints are ways of saying that Bioware suddenly is incapable of good design. With even a modicum of thought I can come up with ways to counter every one of those arguments with game mechanics that are easy to implement. This isn’t difficult work or hard thought—this is basic stuff that any self-respecting game designer (or MMO pundit) should be able to figure out.

1. Parties get loot based on the number of PCs in the group, as done in DDO.
2. AI companions will always be inferior to players because AIs just aren’t that good at positioning and tactics. As long as NPC companions occupy group slots that players would otherwise occupy, PCs will be preferable over NPCs.
3. Pet classes already exist in MMOs. Even if the design isn’t great, you could just treat every class in SW:TOR as a pet class. This isn’t great design—but it’s the worst case scenario.

DDO has already implemented a feasible and functioning companion system and the world hasn’t exploded. Why can’t Bioware do the same?

This is yet another instance of people throwing a tantrum because they are incapable of seeing past the tip of their noses. Here, on bold display, is the common approach that everything with which I am not familiar is poison and evil. This kind of punditry is damaging to the discussion and progress of MMOs.

(You should also read Andrew’s great concise post at Of Teeth and Claws. There's also a strong post over at Player vs Developer that further shows good signs for companion systems.)

Concrete and Abstract Games

I also arrive at a design philosophy of moderate simulationism in MMOs through an analysis of features that define MMOs as concrete games.

Games can be classified along a continuum from abstract to concrete. The most abstract games, you might call them “pure”, are abstract strategy games. The most concrete game would be life itself. The gamism-simulationism dichotomy that I engage with frequently in my abstract design discussions is parallel to the abstract-concrete continuum. Gamism tends toward abstraction—rules for the game’s sake—and simulationism tends toward the concrete—rules for the metaphor’s sake. MMO design is locked in a struggle between the game and the metaphor; the debate is the center of many design discussions and also the source of many ill-conceived arguments.

Abstract Games

“Abstract” as used in “abstract games” means a lack of reference or relation to the real world—Abstract games’ mechanics do not model or have a significant intended relationship with the real world. Think of an abstract game as conceived and played for its own sake. Any similarities we see between it and the real world are not the sources of design decisions, but simply exist because people relate their world to everything they see (a classic reformulation of “if your only tool is a hammer, everything looks like a nail”).

A good example of this is the abstract strategy game of Go (also known as Weiqi or  Baduk). Legend states that Go was invented some 4000 years ago by an counselor to the then emperor of China. The emperor had a son who was a bit dull. He commissioned the counselor to invent a game that his son could play so that he might pick up some degree of mental acuity. How would you design a game if put in the counselor’s situation? It makes the most sense to try to invent a game with simple rules that requires deep thought to play well. To design the game, we’re starting from the very abstract and working our way to mechanics that can be implemented using a few stones and a board with a grid scratched into it. This is demonstrative of how abstract games are conceived.

Abstract games are games designed for the sake of gaming. They do not seek meaning through association with the real world. They give meaning to themselves and their peers. In this way board games can seem utterly trivial to a mature human being unless they genuinely enjoy games (disregarding the social aspect for the sake of discussion).

The “purest” and most extreme abstract games are abstract strategy games. These games have three crucial features:

• Perfect information is available when the state of the game is fully known by both players. (Note that the state of the metagame does not matter here.)
• No element of chance can be present in the game mechanics. This means that only the strategizing of the players affects the outcomes of individual in-game actions.
• Only two players must play the game together. If more players are added, the game becomes political and no longer is purely strategic based on the game mechanics.

We can discern the abstractness of a game by judging how far away from these three features the game strays.

Concrete Games

Concrete games are metaphors for parts of real life. Concrete games create the illusion that the laws of the real world exist in the game. Extremely concrete games are in-depth simulations that allow the player to interfere on behalf of some element in the simulation. These games cannot be played on boards or with pieces of wood or stone, they need the computational power of computers to allow them to provide a model of some part of the real world that can allow players to suspend disbelief.

Concrete games have many players, very limited information flow, and are perpetuated by chance.

If you took all chance away from a concrete game, it would most likely fall apart as a game and become quite boring.  Each player would be able to precisely forecast the result of each of their actions, so all of the complex mechanics would boil down the number of viable strategies to a mere pittance. All of the highly situation decisions made in real life that necessitate the use of rare capacities and patterns of thought would be removed from the game, because chance models the interactions of the myriad minute details that a game cannot hope to simulate effectively. Quantum physics has shown us that even the most basic fabric of reality is subject to the whims of fortune. So to simulate (and therefore to be concrete) requires chance.

In real life, information flow is choked by the limits of our perception. A game that hopes to simulate real life in any way must model this stunted flow of information that may be incomplete or outright incorrect.

The vibrancy and ever-newness of our world is due to the fact that billions of people inhabit it. We are able to communicate with millions of people every day using technologies that were invented only decades ago. The effects of millions of individual interactions between diverse peoples leads to en endless stream of situations that would be fun to model and toy with in a game. Concrete games take advantage of this reality.

MMOs are Concrete Games

MMOs are virtual worlds—in no way are they abstract games. Every individual indicator in MMOs is in favor of their classification as concrete games. As such, they benefit from designs that allow simulation to happen. In this way a game can harness the natural processes of the real world that lead to endless fun and interesting situations. To choke an MMO with abstractness is to take away the very best and most natural fun that can be had in worlds full of thousands of people, saddled by necessity with limited information, and brimming with opportunities for serendipity and chance to reap havoc.

Reducing and Translating Games

A game, in the abstract, is a set of game bits (like individual actors in the game world) and rules (what those actors can do and what the world does in response). The player uses an interface of some kind to interact with the game. This interface could be any collection of physical or virtual objects: a board and special pieces, cards, a ball, or a computer program and the graphical assets associated with it.

Consider tic-tac-toe. We recognize this game because it consists of a three-by-three grid that is filled with Xs and Os by different players in turn, one per grid slot, until Os or Xs are aligned three-in-a-row. But we can create other games that can be translated directly into tic-tac-toe. Here’s one: we take the numbers between one and nine (inclusive) and take turns picking numbers until one of us has a set of three numbers that sum to 15. The same mechanics underlie these two games—the only difference is the interface.

More complex games cannot be directly translated into other games as easily. They must be broken down into constituents that then are simple enough to be translated.

All this translation shows us that a game’s interface can be separated from its mechanics, but how do we actually describe the mechanics of games? Translating one set of abstractions into another, as we did with tic-tac-toe doesn’t seem to get us anywhere in this endeavor. Mathematics is, in fact, a language in which we can concisely and completely define game mechanics. Mechanics can be torn from the game’s interface and described as an abstract mathematical problem that the player is attempting to solve. I will spare you the exact mathematics here, but any game can be reduced to some composition of mathematical problems.

Math has had several more millennia of exposition than game design, so it can provide our young field with some useful language. Problems in mathematics (more than just “1+x=2”-style problems, but problems like sorting a list of numbers without knowing the exact contents of the list) have varying degrees of complexity. The complexity of a problem is basically its “difficulty” to solve, usually expressed in how the time required to solve the problem grows with respect to the “size” of the problem (the size is, for example, the number of numbers that are in the list we’re trying to sort). Non-trivial games can be reduced into math problems of a complexity class of higher than NP.

(Math nerds please excuse me. This isn’t going to be super-precise because I’m trying to make it understandable without taking 1,000 words.)

NP stands for “Non-deterministic Polynomial”, a polysyllabic train-wreck that alludes to how long it would take an “oracle” machine to guess the solution correctly. A problem of complexity higher than NP (I’m thinking particularly of NP-complete problems here) is intractable for a computer to solve—it would take a computer millions of years to solve even modest varieties of the problem. Computers can only verify that the solution is correct in a reasonable amount of time. Reasonable in the theory of computability is a polynomial relationship between problem size and time. It may take a computer 8 million years to solve a certain NP problem, but the computer would only take 8 seconds to check that a solution is correct.

Great games ask you to solve problems that are NP-complete or harder. It has actually been proven that tetris is an NP-complete problem. You’re doing some very heavy lifting when you play these games—your brain works hard to get as close as possible to solve the problem. The startling part is that this is fun!

Raph Koster talked about complexity theory and games in his AGDC presentation this year. I think it’s a profound truth about games that more game designers should understand and utilize.

Goal Generation in MMOs: The Problem

Goal generation is fundamentally a social endeavor. In real life, a person’s goals are largely  dictated by the people with which that person socializes. Your friends relate their goals to you and you, to stay friends with them, align your goals; or, if you don’t particularly favor the friendship, you can find other friends onto whom you can project your own goals. Similarly, successful products in market economies manipulate how consumers generate goals so that the goals of the consumer appear to align with (and cooperate with) the goals of the producer. Goal generation is central to the way we live our lives; many philosophers have dedicated themselves to defining the process of goal generation and validating the processes we use and should use to prioritize goals.

Goal generation is as crucial to games as it is to real life. Understanding how games generate goals for players can help us to see better ways to make goal generation a natural and self-perpetuating process that can lead to games with significant staying-power.

MMOs fail at goal generation, a failure that leads to a soul-sucking emptiness that has driven me from almost every MMO I have played.

The player’s obvious long-term goal in an MMO: to reach the end of whatever content is provided. Here we see the root of the theme-park model. The player is conditioned to get from the start to the end by society and prior games. Limited linear static goal design is a carry-over from single-player games—it follows directly from single-player game design where game designers and game writers create goals for the player based on the motivations that the player’s character should have. The story (and, perhaps, game mechanics) supply these motivations to the player’s character and these motivations are portrayed to the player through cutscenes, dialog, and character behavior. In great single-player games, the motivations of the player’s character are so well-portrayed that the player’s own motivations in the game align with the character’s. This is rarely the case in MMOs.

Character motivation in MMOs is a thin veneer at best—it’s usually completely absent. Because of the broken symbiosis of character advancement and storytelling, character motivation is relegated to a minor role if any. The PC is not considered a unique element of the world that pushes the story forward. MMO design treats each PC the same as every other PC (although sometimes only so far as the PC is a certain race or class). Content is static, simple, and manually designed. Any motivation that the player concocts in an attempt to roleplay is a handicap against character advancement because the only power the player has over his character’s motivation is manifested in avoiding certain pieces of content. The choice isn’t between ways to effect the world—the choice is deciding whether to participate. This choice can be valid, but it represents very few of the choices a hero would feasibly make.

In a single-player game, static linear content makes sense. The player can assume the role of a character who changes the world, and those changes can be relayed back to the player through story events. Limiting the goals of the player’s character works within the framework of the character’s motivations.

In an MMO, using static linear content does not make sense. Designs can use this approach and most theme-park games do, but these designs need to work around the fundamental disconnect between static content and a world that should be changing as players grow their characters. The player’s character moves through physical locations as she advances, progressing towards the eng-game, but those locations are not actually changed. Physical space in MMOs is used to act like the progression of time and events in a single-player game.

Goal generation in theme-park MMOs places the player on a treadmill. This must happen in order to have a world that does not change due to characters’ actions. Goal generation in theme-park MMOs will always be reduced to a grind because it does not demonstrate actual progress. The player maneuvers his character through content to reach whatever advancement goals she might have, but she will ultimately be inhabiting the exact same immutable and unchanging world at every second. When the scenery moves but you’re actually still in the exact same place, the feeling of progress changes to disillusionment. The facade is clear; only our innocence protected us from this understanding when we first entered, wide-eyed, into MMORPG worlds. We can never get our innocence back, regardless of how a game like Aion makes it tantalizing. No matter how fancy a treadmill may be, running on it will never get you to a new destination.

What is the solution? Clearly we must explore dynamic world design. My articles on accountability and simulationism provide clues at where I’d like to go here. I will explore my ideas for dynamic, self-renewing goal generation systems in a future post.

Accountability System for Dynamic World MMOGs

Let’s think about how society holds people accountable for their actions and why. We’re not going to be able to directly model this in a game, but understanding how morality and social structures maintain themselves in real life can provide us with a basis for the mechanical social structure in a game world.

What (most likely) separates human beings from other animals is that we can calculate consciously the affects our actions will have on others. We can conceive of systems to judge actions on a scale beyond our own self-interest. These conceptions are institutionalized through religion and law. Societies dynamically generate and modify moral codes that their members tacitly understand. When there are enough people in the society, institutions arise to enforce moral codes and perpetuate, beyond the scope of the family, certain moral standards.

Accountability and moral agency are tightly bound. If we can be held accountable for our actions, we have the capacity to make moral decisions and be held to moral standards. We want the morality of actions to make a difference in what a player chooses to do when he interacts with others.

Societies in online games take on much of the morality of the real-world society in which the game is played. The effect of morality is deadened significantly in online games, though, leading to plenty of negative, uncooperative behavior that leads to undue pain and suffering.

Active morality promotes social order and cooperation, which, in general, leads to a much better social experience and a more enjoyable gameplay experience. In order for morality to matter in the game world, though, there needs to be accountability. Actions taken against (or with) other players need to have consequences, positive or negative, in order for the morality of those actions to matter to the player. Consequences cannot be enforced by players without knowledge of the precedent action.

Accountability in MMO dynamic worlds is deadened by several factors. Dynamic world games have to mitigate these factors significantly to make accountability work.

1. There’s no way to teach the morals of the game’s society to newbies without them experiencing it. Newbies are given a character that should have a solid understanding of the moral framework, but it’s impossible for the newbie to have that level of understanding.
2. There’s no ultimate punishment in the game world because death has little meaning. This has far-reaching consequences and cannot be overlooked.
3. The cost of abandoning a character and creating a new one isn’t prohibitive. Griefers only lose time when they are forced to make a new character, and those griefers usually have more free time than those who are being griefed.
4. Players don’t play every waking hour, so decisions that may need to be made have to work around when a player is online. The player’s perceptions and actions are limited by how long they spend online—there’s no analog for this in real life.
5. Survival is not a motivator for belonging to social groups. I-game groups are much more ephemeral and the bonds between players are much looser because there aren’t many pressing needs, and no need is as pressing as survival.
6. Membership churns in social organizations. As a result of weaker bonds between players in social groups, a player can jump from group to group at little penalty.

An accountability framework for a dynamic world needs to have several parts:

1. Event capturing and fact collection. The game engine can provide plenty of information about what characters have done. This information can be factually perfect—which is actually better than the general imperfection of information in real life. Collecting data can be difficult in games that have a lot of ambiguous actions available to players, but if players’ actions are relatively well defined, event capturing can lead to a significant increase in accountability.
2. Player aggregation and contextualization of information. Trusted players should be able to take the bare-bones fact sheets generated by the game engine and write histories around them that can be read by other players within the game.
3. Limited information availability. People don’t immediately know all of one anothers’ deeds and misdeeds. A game shouldn’t model this directly, but some gradual information spread depending on player contact and faction contact is necessary.
4. Account-wide activity memory. Because we cannot assume any degree of roleplaying, the moral actor should not be considered the character, but the player. All characters played on the same account should be connected. Activity histories should appear the same for all characters on one account (though there should be some reorganization of the view to allow the activity of the current character to be at the top or highlighted).
5. Give in-game social groups facilities for communicating at all times within the game. This means much more than guild chat. Social groups should have message boards accessible from within the game. There should be something like a wiki that the group can put up to hold important data about itself that its members need to know for the social order to be maintained. Organizational tools are crucial for the promotion of moral and cooperative behavior.
6. Allow social group membership to have a dramatic effect on how a player plays the game. More than simply giving the player a tag beneath their name and a chat channel, guilds should open up a whole new world of coordinated and uncoordinated group activity. When a player joins a new group, she should notice a difference in the way she plays the game every single time she logs on. In this way, group membership can have significance and membership churn can be reduced. Survival can be replaced with raw utility as a motivator for social behavior.

MMOs should make playing with others as smooth and rewarding as possible. In a game world where players can have significant effects on each others’ gameplay, it’s critical that social groups have the tools to invent and maintain morality frameworks to ease cooperation and promote social stability. Without tools to aid tracking and managing accountability, dynamic worlds will continue to breed a disproportionate number of psychopathic characters—such games will always be alluring but, ultimately, socially unstable, exploitable, and brutish.

The Character-Player Gap

In tabletop games, it’s important to maintain the gap between the player and the player’s character. These games hinge on the players roleplaying—the conflict resolution mechanisms are a big aid in the process. The quality of a game of D&D often depends largely on if the players can maintain a separation between themselves and their characters. When the players roleplay their characters effectively and don’t metagame too heavily, suspension of disbelief can take hold and lead to intense immersion. If the gap closes between the player and character, the game becomes a poor simulation with little soul.

MMO design, though it finds its roots in tabletop RPGs, cannot work off the assumption that there is a gap between the player and character. Social competition and cooperation force the gap shut.

How the Gap Closed

In order to compete most effectively against other players (i.e. take the path of least resistance at every possible point and maximize gain) the character is discarded. Roleplaying is significantly less efficient than playing as yourself, so a player motivated by competition will not bother doing it.

To cooperate most effectively in games, players use tools like voice chat that will erode the sense of talking to the character. The player’s interaction with the game world is no longer tightly bottlnecked through the game, it spreads out onto forums, vent channels, IRC, and IM clients. It becomes significantly more difficult to maintain a character-player gap under this diversity of contact—and most of this contact is necessary to cooperate most effectively.

If there is a not a social contract based around roleplaying, people will not roleplay. One or two people failing to roleplaying consistently leads to exponential numbers of players thinking it’s OK not to roleplay. So games cannot expect roleplay where there are a significant number of new players who are not familiar with the concept.

What Does It Mean?

Design games that don’t rely on their being any separation between player and character. Don’t limit the player’s ability to talk to other players because they are not on a certain character. At its most superficial level, persist guilds and friends lists across an entire account (perhaps on a server-per-server basis).

We can only safely assume that a player is roleplaying when he’s interacting with NPCs. MMOs should have a wider variety of NPC interaction choices. NPCs need to have lives that mean something to players—this alone will lead to a significant change in the meaningfulness of an MMO’s world.

The closing of the character-player gap doesn’t necessitate gutting systems of all immersion and stripping them down to their most addictive mechanics. The player character primarily needs more means of effecting the world that are affected by the character and not the player. We simply cannot rely on the player to roleplay in any form. We must fashion gameplay around this fact, not, as WoW and its descendants have done, in spite of it.

(When I present the design for the MMO I’d like to make, you’ll see more clearly how I intend to do what I say here.)

The Abstraction of Character

Perhaps my post about the abstraction of character progress was a bit premature. I defined the progress of characters without defining characters. This post will fill that gap. I aim to present vocabulary and set of patterns of defining and analyzing character abstractions as they exist in RPGs. (This may also be applicable to other genres, as well.)

A pragmatic view of a character in a game world would consist of two facets:

• State: A physical manifestation of the character in the game world, this implies a description of the character.
• Behaviors: A set of abilities that the character uses to interact with the game world.

More simply stated: a character’s meaning consists of what it is and what it can do. Characters allow interesting gameplay because their state affects what they can do and what they can do can affect their state.

State

What is the character?

Typically we describe a character (in terms of game mechanics) using a set of attributes. Attributes are the “ability scores” that are ubiquitous in modern RPGs: strength, dexterity, agility, constitution, intelligence, leadership, wisdom, charsima, etc.. Attributes only exist because they effect the abilities of a character; they represent the capacities of the physical being of the character.

Attributes are either intrinsic or derived. Intrinsic attributes belong to the character’s very being—you can’t break down their capacities further than intrinsic attributes. Intrinsic attributes are the individual variables in the various equations that dictate the effectiveness of abilities. Derived attributes are constructed through the mathematical combination of intrinsic attributes.

Although it doesn’t matter in most MMORPGs, attributes can be used in either their base or modified form. The base form is simply the attribute’s value when no novel effects apply to the character. The modified form arises when some effect happens to the character that changes the attribute. Permanent attribute damage or permanent attribute growth, as happens when a character levels can alter the base attribute value.

Behaviors and Effects

What can the character do?

Usually characters are quite limited in their actions. They can move about the world at varying speeds, but can only touch the world through killing creatures or combining ingredients to make items. On a smaller scale, characters are limited to a certain set of relevant abilities in each kind of activity they can undertake. In crafting, this is usually restricted to each type of item they can craft—an ability is to execute a recipe. In combat, a character has a number of abilities that center on changing the attributes of itself or other characters involved in the combat. The effectiveness of all of these abilities is dictated by individual abilities’ mechanics and the attributes of the character executing the ability.

A Schematic for Describing RPG Character Systems

Here’s my attempt to define a generic way to describe any RPG system. This is not set in stone: it may be incomplete, so please let me know if you can think of systems that this does not cover or anything else I’ve omitted. I will edit this schematic based on input from comments.

Define a set of attributes. Each attribute has

• A name.
• If the attribute has mutually exclusive values (like eye color).
• If the attribute can have multiple values simultaneously (like heritage).
• If the attribute is to be represented as a number:
• Its resolution. (The smallest increment possible that the attribute will recognize or allow.)
• (Optional) Its range. (The smallest and highest numbers possible.)

Define a set of abilities such that each ability has

• A name.
• A list of effects, such that each effect has
• (Optional) a name.
• A magnitude, either as a flat number or as an equation in n-dimensions that may refer to any set of attributes or external conditions. (The amount of damage dealt, hp healed, strength regained, etc. may rely on character level, strength, the color of the character’s hair, the phase of the moon, etc.)
• Either:
• An attribute effected.
• Something created that has its own unique state, which may consist of a different attribute system entirely and different abilities that should be defined on their own terms.
• A duration. (Instantaneous, permanent, or expressed in some unit of time.)
• A set of targeting rules. What will this effect affect?
• A set of triggering conditions. The effect could happen immediately or be reliant on some other condition, such as a certain amount of an attribute, the character’s knowledge of non-euclidian geometry, etc..
• A set of keywords that describe the effect. (A fireball would have the “fire” and “area of effect” keywords, for instance. These exist to allow effects to be modified by one another.
• A list of costs, such that each cost has:
• A duration. (If an effect needs to be sustained and the cost isn’t paid up front or if, for instance, the character debuffs itself in exchange for some other benefit.)
• A magnitude, either as a flat number or as an equation in n-dimensions that may refer to any set of attributes or external conditions.
• An attribute or object in which the cost is paid.
• A list of triggers that could cause the cost to be paid, modified, or diminished.

Use-based Skill Gain, Revisited

My post on the character progression has informed my previous analysis on use-based skill gain. Using the tools I provided in that article, I’d like to re-examine use-based skill gain in more detail and draw some distinctions that I missed last week.

Ryzom vs. Darkfall

We are only concerned with systems that grant XP based on the use of specific abilities within a skill-based system. These systems are called “use-based” because the amount of XP you get towards an ability or ability group depends on how much that ability or an ability within that ability group is performed by the character.

Ryzom and Darkfall illustrate two distinct approaches to use-based skill gain. Here’s how their systems answer the three “scope of experience” questions:

• What will grant experience?
• Darkfall: Performing any action to completion.
• Ryzom: Completing a quest or killing an enemy.
• What will an experience point apply to?
• Darkfall: The ability group to which the action belongs that granted it.
• Ryzom: XP is distributed proportionally to ability groups depending on the abilities used to kill a monster. XP is scaled based on the difference between ability group level and enemy level.
• What can an experience point be spent on?
• Darkfall: Only on increasing the level of the skill or skill group to which the experience point applies.
• Ryzom: XP in an ability group goes towards leveling that ability group. Once the group is leveled, XP is converted into points used to buy new abilities and effects.

Runaway Positive Feedback

Notice how narrow and direct Darkfall’s system is. You get XP that is allocated to what you just did. You can’t specialize in any way beyond what you’ve done because of the narrowness of experience’s scope. If you want to perform well (before you’ve capped all your abilities—the lack of cap really kills specialization at growth’s end), you are stuck using the abilities that you’ve used most. So if you are pushing yourself by trying the most challenging content you have a chance at completing, you are stuck using the same set of abilities to the exclusion of almost all else. The positive feedback loop created in this way does not meet its end until you reach the ability cap. Then, and only then, does an optimal build begin bringing its other abilities up.

The pendulum swings between the necessity of complete specialization during character growth to the complete lack of specialization at growth’s end. I see this as the worst of both worlds. To play optimally, you’re trapped into using only what you’ve spent the most time using—until you’ve maxed everything out, at which point it doesn’t matter.

By broadening what XP can be spent on, a system allows specialization beyond exactly what the character is using. Ryzom’s system benefits significantly from letting players choose what their experience will improve within an ability group, instead of forcing the growth to occur in specific points. The positive feedback loop of growth in a certain area spurring further growth does not disappear, but it is significantly moderated.

Exploitation 

A system is more exploitable if the rewarded action is one swing of a sword. If the completed action is broader—like, say, killing an enemy in Ryzom’s system—then the problem largely goes away. Ryzom’s system proportions XP in a use-based fashion, but doesn’t focus all growth only on counting ability use.

The optimal strategy for becoming a master swordsman should not be:

• Find a friend who can debuff attack power and buff defense.
• Have him debuff your attack power.
• Have him buff your defense.
• Find a mob that has a lot of health.
• Debuff the mob’s attack power.
• Buff the mob’s defense.
• Swing at the mob a million time until it dies.

If that is feasibly the most effective way to gain skill levels, the system is broken. It’s trivially easy to level up that skill. In Ryzom’s system, you can’t gain XP beyond some fixed cap for a given monster, regardless of how many times you swing your sword at it. It’s clearly easier to exploit a use-centric use-based system than it is to exploit a system that rewards only broader accomplishments, like killing an enemy.

Limited Reward Potential

Darkfall has such as strict conception of use that characters cannot be rewarded for any activity above the use of an ability. Why should a system be so restrictive? It cuts out much of the benefit in the experience point abstraction. Ryzom reaps the rewards of using an experience point system while still maintaining the customizability and flavor of a skill-based advancement system.

Use-centric Advancement Systems are Inferior

It’s important that we draw the distinction between use-centric systems, like Darkfall, and other use-based systems. Use-based systems filter XP towards limited ability groups based on what abilities were performed to cause the XP gain. Darkfall’s system represents a pure version of this where use is the center of advancement: all character progress is based directly on use and little else. Systems work better when, like Ryzom’s, use has a more abstract incorporation into how character growth happens.

Your Job is to Make Tasty Sandwiches!

I walked into a Subway sandwich store for the first time earlier today. After ordering, I was confronted by a lot of options that weren’t presented in an efficient way. There were approximately four steps where I had to make some choice as to what to put on my sandwich—four distinct phases where different ingredients were either added or withheld from the sandwich.

When the sandwich was completed, it has a total of 4 ingredients in it (bread included).

I ordered a sandwich that had a distinctive name that suggested almost every ingredient that would be present. “Sweet onion chicken teriyaki” leads me to expect the sandwich has onions, teriyaki sauce, and chicken on it, probably with lettuce thrown in because that’s how sandwiches are usually made. Instead, after ordering the sandwich, I was confronted with four different decisions, each with more than four options. It was not clear at all what would go on this sandwich or what should. Why is that? Because I am not a professional chef. I don’t go into the store expecting to be saddled with making myself a good sandwich. I went to the store so that I could buy a good sandwich that they had thoughtfully designed and put together.

This is a great metaphor for a significant problem application designers face: giving the user as few decisions as possible while allowing them to effectively and easily use the application to accomplish their goals. Subway failed at this basic design problem by saddling me with far too many choices, almost invalidating its primary reason for existing (to sell me tasty sandwiches as easily and quickly as possible so as to make a profit).

Aion makes an innocent but devastating mistake in an effort to simply make a tasty sandwich. They’ve streamlined their game into a candy-coated, artificially enhanced stream of purpose-built monotony. Dungeons and Dragons 4th Edition falls into this trap as well.

When you log into Aion, you enter a world that is clearly made so that you can run around and effectively kill monsters. Everything’s arranged nicely for your viewing and killing pleasure. The ducks are lined up and a well-maintained rifle is put in your hand for an enjoyable afternoon of shooting—but the ducks are wooden and the rifle doesn’t shoot anything, a wooden duck flips backwards whenever you fire. The gameplay is directed; it’s so directed that, after 10 hours, it felt completely empty to me. My path through the game was perfectly clear. It was so well lit and nicely paved that I felt that I might try a different, dustier, less-traveled path. But there was no other path.

Dungeons and Dragons 4th Edition (4e from here on) streamlined the hell out of the tabletop roleplaying—or, perhaps more aptly, rollplaying—experience. Its predecessor, 3.5e, was asymmetric, arcane, and had a serious case of power creep. Wizards of the Coast apparently had enough of that and decided to finely tune the 4e rules to cut off all those rough edges and remove the arcane and less-trodden paths, to simplify the complex, and to run out of town all the rules but those that governed situations that could be vaguely described as combat. Suddenly the sprawling 3.5e is replaced by a very consistent, smooth 4e that has all the corners neatly rounded and all the danger areas surround in safety fences. 4e is a game system that is very obviously a game—it has clear boundaries that become quite obvious when you try to run a 4e campaign. All those fun utility items that abounded in 3.5e are gone. The unique and flavorful mechanics of each class are replaced by abilities that are minimal variations on a consistent framework. The abilities and, by extension, the classes, feel like repackaged and rebranded copies of the same few ideas.

Streamlining games can solve mechanics problems, but it ultimately rips the soul out of a game. When all of the interesting detail is stripped and all the excitement is paved over, it doesn’t matter if the game is perfectly balanced: it simply will not have enough flavor to be fun. You might leave the game feeling like it was well-balanced and complete, but you won’t find yourself excited.

Moral Choice Beyond Good and Evil

While browsing through the Alganon website for publicly available data on the game, I was angered by the insistence on using the horrible good-evil dichotomy that has become the standard system of moral choice in games.

There are two factions in the game: The Asharr and the Kujix. Guess what they represent? Just pick the easiest possible dichotomy that has been used the most in videogames: Good vs. Evil. Alganon doesn’t stop there, though! It folds all of of the positive “good-aligned” traits—light, nature, heart, mind, order, obedience, and protection—into Asharr and similarly it folds the opposites into Kujix. They went all-in on this cliché. This kind of design laziness borders on the obscene. Reading about Alganon’s weak backstory brought memories of a certain AGDC session flooding back into my thoughts.

Mot and I attended a group session at AGDC ‘09 entitled “The Jesus-Hitler Problem” where we had a series of small-group discussions about how to make moral choice in games less banal and ludicrous. Everyone agreed that the good-evil dichotomy is weak, overplayed, and should be relegated to the trash heap, but few people had much to say about how to replace it. Some suggested avoiding the question all together and divorcing moral choice from game mechanics. Some suggesting having some kind of faction system in games to represent a players alignment with the wants of different important groups in the story.

It’s definitely time we ditch the good vs. evil dichotomy in games.  Both sides are stupid. No one actually ever fits into either of the sides accurately. They’re caricatures that have been dulled by overuse. (And I find it ridiculous that people don’t have a problem how MMOs imply that the moral caliber of one’s being has to do with one’s race.) We should keep moral choice as a mechanic, though, because leaving it out doesn’t encourage players to try different paths. The stakes become very low if moral choice doesn’t actually have an effect on the game world—we almost shouldn’t bother with moral choice at all in that case. Faction systems are a better idea, but don’t fit a wide range of genres.

My suggestion is that we keep moral choice, but change its gamut radically. Moral choice shouldn’t run from perfect good to perfect evil separated by a vanilla neutrality of uselessness. For moral choice to be effecting and memorable, players have to be forced to choose between two equally appealing (or equally disastrous) options. There should be a solid case for either choice being good or evil. 

Here are some dichotomies that arise in moral dilemmas; one’s beliefs on a dichotomy need not be either one or the other, there can be some degree of dithering:

• Idealism vs. Pragmatism (Hope vs. Reason)
• Material vs. Spiritual
• Mercy vs. Justice
• Need vs. Deserve
• Impulse vs. Reason (Heart vs. Mind)
• The Many vs. The Few
• Authority vs. Equality
• Self vs. Others
• Present vs. Future
• Certainty vs. Opportunity (Fate vs. Free Will)
• Intent vs. Consequence
• Unity vs. Diversity

If we profile NPCs through using their positions on these dichotomies, we can construct almost lifelike belief systems. Once we have belief systems, we can present the player with options that will either appeal to or disgust NPC groups that with which the player interacts. The tests of morality can occur relatively frequently, probing at the player’s conceptions of each of these dichotomies. Different factions react in different ways depending on how the player has behaved earlier in the game.

Through expanding the moral quandaries and removing the pretense of good vs. evil, we can create arresting moral decisions, and then have those decisions have deep-rooted effects on the way the game progresses. Such a system will be significantly more engaging, replayable, and thought provoking.

On Balance, Part 2: Ten Key Tips for Balancing

On Monday, I wrote about the fundamentals of balance. Here’s the second and final part of that discussion.

Here are ten important take-aways from Sirlin’s series on balance, but with an MMO bent. These tips cover a wide range, from ways to find imbalance to ways to quantify and fix those imbalances. I added specific MMO-related applications of some of Sirlin’s points about balancing fighting games.

Finding Imbalance through Tier Lists

• Ask players to organize classes, character types, and abilities within each class into five tiers based on their power or usefulness in different scenarios. The highest tier should be “God”-level—these are dominant strategies—and the lowest tier should be trash-level—these are strictly dominated strategies. You want to clear out those two extremes and ensure that the other tiers are as close together as possible.
• Tier lists can be applied to classes or abilities in MMOs, but with a caveat. Characters in typical fighting games are mutually exclusive in the context of play—a player can not be more than one character at once. But in an MMO, a character can have multiple abilities at once and those abilities can be at multiple power levels.
• In a class-based game, it’s easy to apply the tiering system: for each role, tier the classes in their effectiveness. It’s not a disaster if a class is God-tier in one role, as long as it isn’t God-tier in too many, making it basically a dominant strategy to play that class (ala Channellers in Shadowbane). You can use tiering to roughly balance the utility of each class in different situations. It’s a good idea to make the design intentions of classes available to the player so that they don’t make a terrible decision when they choose a class that gimps them at doing what they love.

Avoid Imbalance through Preventative Design

• Imbalances are avoided by the use of counters. Sometimes these counters are passive: Elemental damage is countered by elemental resistance; physical damage is countered by dodging, blocking, and armor. Sometimes these counters are active, like using a shield bash to interrupt a healer casting a life-saving heal spell.
• Design counters and counters to counters. But don’t turn your game into rock-paper-scissors.  Iterative deletion of dominated strategies can be used to determine where bedrock is hit. Sometimes counters can be generally weak, but they can exist just to counterbalance a possibly exploitable mechanic.
• Don’t become fixated on balancing at a micro-level. In a class-based game, you’re balancing class against class, not ability A against ability B. Keep this in mind—sometimes combos of a class’ abilities can make it overpowered and you’d miss that if you were focusing on micro-level ability balance.

Balance Towards Fun

• Abilities have to be powerful. Balancing games isn’t about lining numbers up so that they sum to zero, it’s about making the game as fun as possible for as long as possible.
• Maximize the time where both sides have a fighting chance. Always give each side a reason to fight further—there should always be something to lose worth protecting and something to gain worth taking. Be careful of runaway negative and positive feedback loops.
• Always pull the bottom up to meet the top. This can be difficult in MMOs, but you should work very hard to avoid nerfing classes or abilities. It’s better to have a gap filled a little too much than to leave a void.
• There should be gamist reasons why every mechanic is present in the game. There should be simulationist reasons why mechanics work as they do. The challenge is to pick the mechanics that are balanced for simulationist reasons—actually real-world systems that balance one another aren’t easy to find.

Mechanic Assessment: Use-based Skill Gain

(I made a more recent post which basically replaces this one while doing a much more thorough analysis. Please read it instead of or in addition to this one.)

From my experience with use-based skill gain systems in Oblivion, Morrowind, and Darkfall, I’ve noticed that such systems are inferior. They should be avoided in favor of other approaches to skill-based advancement (as in Fallen Earth) and class/skill hybrid systems (like the one in Final Fantasy Tactics).

First, three gamist reasons why use-based skill gain is an inferior character advancement system:

The character can only be rewarded for a much narrower set of tasks. And that set of tasks is doing whatever action the player wants to level. It’s not clear how quest (or whatever kind of achievement system you want) rewards can fit into this framework. Use-based skill gain cuts out an important part of the incentive structure. In an RPG like Dungeons and Dragons, Dogs in the Vineyard, or Mousegard, combat isn’t an end, it’s a means towards surviving a precarious scenario. There are other means, such as parley, avoidance, and escape that serve that purpose just as well. With use-based skill gain, it becomes difficult to reward the player character for accomplishing anything except easily quantifiable combat and crafting tasks. This leads to gameplay focusing on direct combat and crafting, which narrows significantly the effective and beneficial conflict resolution methods.

Use-based skill gain leads to runaway positive feedback loops that restrict character growth and ability diversity. I kill using ability A so ability A becomes more powerful so I use ability A to kill stuff. This loop generates a second-order effect on skill growth. If you only have a few abilities that are strong enough to use against mobs that drop worthwhile loot, you’re going to use those abilities frequently leading to them becoming more effective and the farming being more efficient and worthwhile. In this way, characters are stuck using the same abilities because only certain abilities are day-to-day useful. But all the abilities are on a similar scale. The Illusion and Mysticism schools of magic were like this in Oblivion: they had some nifty effects, but they were largely composed of utility spells that you would never justifiably use enough to keep the skill level competitive with your melee skills or destruction magic.

Use-based skill gain encourages and rewards exploitation, macroing, and cheating. Some skills necessarily will be used less than others. By factors of hundreds. This forces designers to balance skill advancement against use. This problem cannot be solved. Designers need to measure skill-use frequencies and balance that frequency against how difficult advancement should be. But if a player wants to level a skill, he’s going to find ways to use it more than is reasonable, throwing these calculations off and leading to imbalance. If the player wants to level his buffing abilities, he is going to cast buff spells on everyone he sees if he’s nice, but more likely he’ll cast a buff on himself, then dispell it, and repeat those two actions until he has the desired skill level. Players will always seek to find safe ways to level skills, trivializing the advancement system—developers will always be behind in preventing this kind of behavior. Exploiting and macroing becomes the only way for an honest player to keep up. Darkfall’s EU server has fallen victim to this problem. Exploitation is always the most effective way of increasing skills and it breaks the balance of skill gain.

And one simulationist reason:

Use-based skill gain doesn’t make sense from an immersion/metaphor perspective either. People do not go out and put their life in direct danger to advance from novice to super-novice at using a sword. They spend years practicing with the weapon for several hours every day. The time spent practicing far outstrips that time spent in actual life-threatening struggle. When you’re engaged in combat where life is in the balance, the amount of skill you have when combat begins determines if you live or die. You’re focused on survival, not on dinging 34 on your sword skill. Certainly you will learn from direct combat, but not even a tenth the amount you learned from training since you were the age of 10.

Use-based skill gain should be avoided for primarily these four reasons. As a mechanic, I thought it was a great idea before I played games that implemented it. Now I don’t see a reason to go with use-based skill gain over a different skill-based advancement system, such as purchasing skill levels with XP or some other broader resource.

On Balance, Part 1: Strategy and Depth

Everyone at all interested in game design should read Sirlin’s series of posts on balance. He brings up just about every important facet of balance relevant to games in general. He uses mainly examples from fighting games to illustrate his points—they work great as examples of the concepts he discusses.

In this post, I’m going to summarize many of concepts Sirlin illustrates (you should definitely read his articles if you have the time). Balance is a critical concept in MMOs and certainly merits a post on this blog.

Strategy

A strategy is a planned set of actions. Every game has a strategy-space that consists of every possible strategy the game permits.

Some strategies are better than others. The best strategies dominate all other strategies. A strategy is dominant if it is always the best strategy to choose regardless of the state of the game—an expert in the game would always choose that strategy regardless of his opponent’s decisions and the expert would always win.

A strictly dominant strategy is always the best to choose. A strictly dominated strategy is never the best to choose.

There are different layers of strategy. In theme-park MMOs, there are character growth strategies specific to each class (the specific talent builds for a feral druid, a prot warrior) as well as in-combat strategies (a DPS rotation for a hunter or an aggro control plan for a tank). The order in which you complete quests is also a strategy. Each of these layers have different goals and optimizing one may necessitate choosing suboptimal strategies in others. MMOs are fun because there are a lot of strategies that allow us to succeed—success is almost guaranteed—so we almost always feel like we’re being smart players by choosing good strategies, even if our strategies are far from optimal.

Darkfall is a unique example of the strategy paradigm of character growth being turned on its head in an MMO. Any character, as of October ‘09, can max out every skill in the game. Every character is expected to be able to effective perform every role as the game is played now. Most MMOs necessitate character growth strategies that are essentially time-independent, because there is usually some limit to the character’s growth that is set by its class or a skill cap—character growth strategies focus on the character optimizing for a desired role when it reaches the limit of its growth. In Darkfall, all growth strategies are rendered moot in the long run. Every character is exactly the same given a several month span of time. The strategies that matter for Darkfall in its current for are those that optimize the instantaneous power of a character at all points in its life. This problem is significantly harder to solve, but it tends to be less interesting than endgame minmaxing. (There probably can only be one optimal growth strategy in Darkfall.)

Depth

Depth is the result of there being enough viable strategies for the number of possible strategic permutations to outstrip the player’s capacity to experiment with many of them in a reasonable amount of time. This results in an evolving metagame where certain combinations become popular.

Depth is, at its root, the result of asking a player to solve variations on a problem that is very difficult to solve. It’s not obvious to the player at any level what strategy will net him the best results, so players will try out many different strategies against many other strategies in search of optimal solutions.

For instance, assessing mid-game positioning in chess can be a monumental task as the number of possible moves grows significantly as the board opens up. Calculating the value of positioning vs material is a monumental task. Usually there is not enough time to come to a definite valuation, if one is possible. Chess has a lot of depth in that you can play it many times without seeing identical mid-game positions, so not only is piece valuation and position valuation a hard problem, it’s rarely the same between games. Different players put different weight on different ways of evaluating board position and material; the metagame of chess has been evolving for at least four hundred years and most players still struggle to gain a grasp on evaluating board positions and possible moves.

MMOs currently lack depth. The problems that a player must face when he sits down to play are severely limited in difficulty. Modern MMOs are mostly built to tickle players with rewards and those rewards are their primary motivation for continued play. If game systems had enough depth to rival the reward addiction, MMOs would be able to get over the Kosterian Curve of rapid adoption followed by devastating desertion.

Accountability is the Currency of Dynamic Worlds

Theme-park MMOs are consequence-free zones—unless you get into the kind of behavior that is against the TOS, but even then the worst punishment is being kicked off the game.  It’s not a problem that theme-parks don’t have serious consequences for player character actions within the game mechanics because no character has an impact on the world. Regardless of what you do (barring some very rare GM-run events) the same mobs will continue to spawn in the same places and the same quests will be done by different people without interruption.

As soon as you give the player the ability to take actions that have far-reaching impact on the experiences of other players, you need to instill a conception of consequence in the player’s mind or face a blight of sociopaths actively ruining the game.

The basic building blocks of dynamic worlds are the actions of the players as they interact with one another and the environment. These actions have meaning in that they change the behaviors or capabilities the environment—both the simulated world and the players that inhabit it. Players need to have feedback from the environment as they interact so that they can learn the rules of interaction and the extent of their own capabilities.

Feedback can be supplied in two ways: in that the players sees what effect his actions have on the world, and in that the player sees how he should feel about that effect.

All games give feedback in the first form. You push on a crate and it moves in the direction you pushed. Simple feedback like this teaches you how to interact with your environment and helps you construct a mental image of tools you can use in further problem-solving endeavors. These rules tend to be too simple in MMOs and this feedback is too minimal, but this feedback’s existence provides the underpinnings for the second kind of feedback.

The second kind of feedback is less common in MMOs. Usually single-player games have NPCs that will react to the player’s actions by interacting with the player differently. The way NPCs react to the player suggests how the player should feel about what they are doing in the world. If the player is behaving badly (in a particular social context), NPCs react with shock, horror, and derision towards the player—the player is supposed to feel this about his actions and adjust them. Because NPCs in MMOs are generally worthless cardboard cutout quest-givers, their reactions have no importance to the player—even though through executing the quest-givers will, the player has interacted with the world in the only way possible in the game, the player doesn’t care about the NPC and skips through quest text. NPCs are just tools used to move forward, to get to the endgame and do the real business the game brags about, be it raiding or PvP.

When the player interacts with other players as his main means of playing the game, either through direct interaction or through effecting a cohabitated world, the tools required to show the player how he should feel about his actions are altered beyond recognition. No longer are NPCs the central focus of the game—players have to make moral judgments about other players. The quality of those judgments has an impact on how much fun each player has.

Through making moral judgments, players establish de facto tribal societies. Once in the context of a society, players behave in regimented, sensible ways while relating to others in their society. The player who acts out will be stripped of his status within the society and will not be able to take advantage of the facilities that society provides, so players are incentivized to conform and contribute. The relations built through this socialization keep players hooked into the game world and happy. PvP is contextualized into society versus society warfare, not meaningless and random killing.

Accountability is at the center of the social and moral systems that form the backbone of player-driven, dynamic worlds. Developers have to provide tools to allow players to hold one another accountable for their actions. Developers need to build tools to track the behaviors of players and reveal important details to other players in appropriate places, building a framework for players to establish crucial trust relationships. By giving players the power to avoid untrustworthy or uncooperative agents, developers can give their players a world where actions have meaningful consequence without the world falling apart into a chaotic mass of criminality and complexity.

Accountability is the social currency of dynamic world MMOs. In order for a player-driven dynamic world to succeed, mechanics must facilitate accountability.

Design Goal: Kill the Boring

When you’re playing a character in a huge 3D world, you want to go to wherever the interesting stuff is happening so that you can have fun. Yes, it’s a trivially true statement, but it lies at the center of many debates in MMO design.

The obvious design decision is to make players walk or ride or float on clouds or fly with their wings (sometimes) through every foot of the world. They have to do this manually, because if they aren’t logged in and staring blankly at their computer screens, bad things must happen.

Bad things must happen, right? Obviously this is the case because no MMORPG allows you to AFK travel. Hell, few MMORPGs allow you to do much aside from various guises of waiting around  (watching progress bars, walking on a safe road, riding on a gryphon, watching your character auto-attack an enemy to death). But what bad things would happen if you let your players skip the boring parts of their character’s lives?

If we’re going to follow the industry-wide pattern of streamlining MMO gameplay, we should take the obvious but oft-overlooked step of allowing the players to be offline or AFK during as many boring or trivial tasks as possible.

EVE Online lets you queue up manufacturing tasks and log off or do something else while they complete. This is an important step towards automation, but a small one. EVE’s mechanics are based around largely non-interactive fundamental activities: combat involves a lot of waiting and gathering is almost entirely non-interactive. In most combat and gathering activities, less than one-third of the player’s time is spent interacting with the game—and those interactions may not actually involve making interesting decisions.

MMOs will be more fun to play if as many boring parts as possible are stream-lined out of gameplay. Keep travel time appropriate so the world can be big, keep production time appropriate so items can have meaning, but don’t make the player pay for this by making them stare at the screen needlessly.

When you’re addicted to an activity, the most intense moments are usually when you desperately want to fulfill the addiction, not when you’re sating your urge. By that logic, it superficially seems that having a lot of dead time your MMO is justifiable. But if we extend the logic further, we have no answer to the question “why must players be logged at all while their character does mundane and boring tasks?” The more time a player spends bored or disinterested in a game, the more boring the game appears. But if being bored in-game is optional (and actually sub-optimal), players won’t stay in your game to be bored—they’ll stick around to do what’s fun and they’ll only remember that your game is fun.

What would you prefer? Four hours spent in-game where only one is spent having fun, or two hours spent in-game where one is spent having fun? MMOs these days lean towards the 1-in-4 ratio.

So why don’t we ditch the boring stuff?

This suggestion is not one that can be implemented with little change to the theme-park model. It requires a rethinking of how MMORPGs engage players. But if the average player is going to put in 1,000 hours of play time, I’d rather have those hours in concentrated bursts of fun lasting around one hour, not in five-hour slogs.

Defining Moderate Simulationism

Simulationism does not mean blind reproduction of real-world processes in a game. I have a more nuanced and objective definition of simulationism. Using this conception, we can better understand what I mean by moderate simulationism.

My approach to moderate simulationism is based on a managerial technique employed to some success among Japanese automakers and subsequently many businesses worldwide: The Five Whys.

Take a look at the high-level mechanics in a game and ask “why”—“why does this work?” or “why did this happen?” After drilling down through consecutive “why”s, you reach a point where you have no other answer but “because the devs say so.” This is the point at which there is no further game-level reasoning: you must step out of the rules themselves to answer this “why.” I call this jump to metagame reasoning a “bottom”. It’s the end of whys that are useful to the player when they’re playing the game. Each time you have a successful answer to a “why” question in a chain, the why-level for the mechanic increases by one.

Gamist systems tend towards metagame reasons taking control almost instantly. The rules in a pure gamist system exist primarily because of metagame reasons—all of the rules do, high and low abstraction layers alike. The perfect simulationist game would allow this chain of whys to proceed all the way until you reach the point you would say “well… I don’t know!” in real life.

In virtual world design (which is the primary concern of just about every single MMO) pure gamist systems fail because the game rules become too arbitrary for the players to suspend disbelief and become immersed in the world. The immersion loss is too great to justify the marginal balance improvements a pure gamist system ensures. Pure gamist virtual worlds do exist, though, in the form modern MMORPGs. There are no pure simulationist MMOs because pure simulationist systems are impossible to implement. Limitations on computing power doom pure simulation. Even if the computing power limitations disappeared, thorough simulations routinely become too complex to program and too complex to play. We don’t even have the means to interact with a computer accurately enough for a simulation of such depth to be feasibly usable.

What, then, is moderate simulationism?

Start with a living virtual world. It doesn’t have to be an accurate simulation of some subset of the real world, but it should be a virtual world that sustains its own operation but changes as different agents interact with it. These agents should be AIs to start. Once you have a self-altering world, you define a set of interaction points at which the player can touch the game world—these interaction points manifest in the player character and its capabilities in RPGs. The way that the player character can interact with the world depends on the parameters of the simulation and the natural conditions that arose as the world evolved under the influence of AI agents. The why-levels of most of the key aspects of the game world will now be significantly deeper than they are in modern theme-park (and even further, sandbox) MMOs, but there is still a bedrock of world design principles that are safely below the why-level of the real world.

(Props go to Tales of the Rampant Coyote’s article on making magic less mundane which helped me to better understand the core of simulationism.)

A Path to the MMO Revolution

Bloggers prattle on about mainstream, big money, big staff, mass-market MMOs. Only there is the stagnation so brutal and the gameplay so monotonous, yet the money is so huge.

But MMOs will not be redefined in this grand marketplace of titans. MMOs will be reiterated—progress has slowed. Now it is no more than a telephone game between big development houses. What developers misinterpret from WoW will guide the changes in the AAA MMO industry, not the visions of luminaries.

When there is (at least the perception of) big money available but a large and risky initial investment is required, market participants will aim to minimize all other risks. World of Warcraft has shown the market the revenue potential of MMOs, but, in the process, it has kicked risk-aversion into a high gear. Investors see WoW’s success and think that only WoW can be so successful, so they should reiterate WoW to capture equivalent numbers. The more deviation from the WoW formula, the more risk. This has slowed MMO evolution to a crawl. WoW is so big, so polished, and so old that other MMOs who aim to duplicate WoW’s success in a high-investment, high-risk environment will inevitably fail.

If you think it’s feasible to compete directly against a game that has 300,000 man-hours  invested (high quality man-hours by some of the best MMO designers and programmers in the industry) and has already captured the market, I have a bridge in Alaska to sell you. It ends abruptly in the Behring Strait. Maybe you’ll be on Deadliest Catch.

How can new devs combat the million-ton behemoth? By being more agile and keeping their eyes open. It also helps to stay as far away from the towering King-Kong-standing-on-King-Kong-in-a-Hitler-disguise as possible.

Agility means being small and quick. Through limiting the scope of projects, experimentation can take root. Developers can rapidly iterate on projects with small scopes, trying new features and honing old ones in the fraction of the time required by a towering monolithic AAA title. Project complexity rises exponentially with respect to the number of modules that must interact. Keep the number of modules small and you can make fairly radical changes without worry about the entire game turning into a flaming pile of debris. This small-and-sharp model for project design can be applied to each module in a game as well. When developers recognize the functional boundaries key modules, they can manage the complexity of module interoperation more efficiently.

Keeping your eyes open ensures that you don’t fall into the traps into which your predecessors tumbled headlong. Communicate with your community. Don’t promise what you can’t deliver. If you find you’re promising something you can’t deliver, let people know. Keep everyone informed and keep them excited. Pay attention to the design of other MMOs. Play other people’s games—and, more importantly, play your own game. Don’t add features that aren’t important. Make sure you’ve got one guy who has wrapped his head around the design trajectory of the game and the goals of the design. Make sure you’ve got someone who has their head fully wrapped around the lore. Make sure you’ve got a programmer that understands the architecture of the modules inside-and-out and can direct the general construction of the game. Etcetera.

Staying away from the Space King Kong Nazi Ape means do not try to directly compete with WoW. Find new mechanics and perfect them. Make a different game. Don’t be afraid to genre-bend, but don’t let genre-related marketing drive design. Don’t make a game that is a reflection of a game already considered fun; Make a fun game.

The games that act on the prerogatives set forward in this post will comprise the MMO revolution. The generation of MMOs that starts the revolution will be smaller games than we’re used to in the MMO field. They will be released early and patched often. They will evolve into fun games elaboratively. There will be no Great Leap Forward, but instead a series of rapid, modest steps. It’s easier to walk up a mountain than it is to leap to the top of it.

Make the path of least resistance more fun.

If you let players design their content, they will not design it to be the most fun. They will design the content so that they can accomplish their goals as easily as possible. This dooms open PvP games set entirely in traditional purely persistent shared worlds.

My philosophy of design is to make the path of least resistance into the path of most fun. There doesn’t always need to be an easy way out of every situation that yields positive results for the player, but if the player is optimizing for the easy way out, which players usually do, they should find their optimization leading them towards engaging with game systems in intended ways designed to be fun.

In a typical open PvP MMO, players can pick and choose their engagements so that they very rarely have to fight a superior enemy. This means that players hunting alone are going to get ganked by roving bands of players who risk nothing in the engagement. Rarely will there be a scenario where players on each side of an engagement assess the situation as being even enough that unpredictability of the outcome leads to a willingness to take the risk on an even-strength enemy. The result of this kind of behavior: most engagements are routine for one side and frustrating for the other.

Discontent stirs regarding PvP as pageantry, as well. Players who seek alternate worlds where consequences can be devastating and rewards significant abhor scenarios and “battlegrounds” preened and molded particularly for the delight of theme-park goers.

There must be some middle-ground between meaningless risk-reward scenarios and largely uneventful, often-frustrating open world PvP.

The most obvious solution is to restrict the actions that players can take in the game world. This doesn’t mean adding even more hyper-gamist rules to an already rule-riddled mess. It means limiting the parts of the world in which the game allows direct player interaction so that each interactive simulation can be very good, very fun, and very convincing. Only interactively simulate the fun stuff—allow the player to only participate in the simulation of what players want to do in their character’s shoes. The game world and AI can handle the mundane stuff. If the player needs to control the mundane stuff, let her give the AI goals and then go and do something fun with another character or *gasp* let the player play another game instead of getting burnt out from playing your game too much.

My primary issue with modern MMOs is that they encourage players to put four hours or more into the game per day, and out of those four hours less than one hour is spent doing what’s fun. It’d be better for both players and devs if games were made that can be lots of fun for 30 minutes a night out of 45 minutes played. If I’m going to get 1,000 hours of total game time out of your target audience, I’d prefer to get those 1,000 hours at one hour per day instead of five hours per day—especially when that one hour will not lag behind the five-hour session in terms of time spent having fun.

(In a later post I’ll get into the specifics of how to do what I suggest here. This is a basic precept on which I’m designing my own MMO. I intend to make the design completely public—I don’t care if someone steals my ideas. I don’t have the time and resources to make the MMO I’d want to. I’d rather play the game of my dreams than have it sit in a word doc on my desktop.)