A Formal Approach to Game Development and Research
Introduction
All game elements are created within the framework of one design methodology or another. Whether it's creating a physical prototype, a software interface architecture, building an argument, or conducting a series of controlled experiments, design methodologies guide the creative thought process and help ensure quality work.
In particular, iterative, qualitative and quantitative analyzes support the designer in two important ways. They help analyze the end result to clarify its implementation, and analyze the implementation to clarify the result. By approaching a problem from both perspectives, a designer can consider a wide range of possibilities and interdependencies.
This is especially important when working with computer and video games, where the interactions between coded subsystems create complex, dynamic (and often unpredictable) behavior. Designers and researchers must carefully consider interdependencies before making changes, and scientists must acknowledge them before drawing conclusions about the nature of the experience.
In this article we present the MDA (Mechanics, Dynamics, and Aesthetics) method developed and taught as part of the Game Design and Tuning Workshop at the Game Developers Conference, San Jose 2001-2004 [LeBlanc, 2004a]. MDA is a formal approach to understanding games that attempts to bridge the gap between game design and development, game criticism, and technical game studies. We believe that this methodology will clarify and strengthen the iterative processes of both developers and academics and researchers, making it easier for all parties to decompose, explore, and design a wide range of game designs and game structures.
Towards a comprehensive system
Game design and authoring occur at different levels, and the field of game research and development involves people from a variety of creative and scientific backgrounds. While it is often necessary to focus on one area, everyone, regardless of discipline, will at some point have to consider issues that extend beyond that: the underlying mechanics of game systems, overall design goals, or desired gameplay outcomes.
AI developers and researchers are no exception. Seemingly unimportant decisions about data, presentation, algorithms, tools, vocabulary, and methodology will trickle down to shape the final gameplay experience. Likewise, all the desired user experience must be built into the code somewhere. As games continue to generate increasingly complex behavior among agents, objects, and systems, AI and game design are merging into one.
System coherence occurs when conflicting constraints are satisfied and each part of the game can relate to each other as a whole. Decomposing, understanding, and creating this coherence requires movement between all levels of abstraction—a fluid movement from systems and code to content and game experience and back again.
We propose the MDA method as a tool to help designers, researchers, and scientists perform this translation.
Games are created by designers/development teams and consumed by players. They are bought, used and eventually thrown away (discarded) like most other consumables.
The difference between games and other entertainment products (such as books, music, films and plays) is that their consumption is relatively unpredictable. The series of events that occur during the gameplay and their outcome are unknown at the time of completion of work on the product. The MDA method formalizes the consumption of games, breaking them down into individual components:
… and the creation of their design analogues:
Mechanics: describes specific components of the game, at the level of data presentation and algorithms.
Dynamics: describes the behavior of mechanics during execution, affecting the player's actions and the results of each other's actions over time.
Aesthetics: describes the desired emotional reactions evoked in a player when interacting with a game system.
The core of this concept is the idea that games are more like interactive products than media. By this we mean that the content of the game is its behavior, not the media that flows from it towards the player.
Thinking of games as interactive products helps to think of them as systems that shape behavior through interaction. This facilitates clearer design choices and analysis at all levels of study and development.
MDA in detail
MDA as a lens
Each component of the MDA system can be thought of as a “lens” or “perspective” on the game – separate, but causally related. [LeBlanc, 2004b].
From a designer's point of view, mechanics give rise to dynamic behavior of the system, which in turn leads to certain aesthetic experiences. From a player's perspective, aesthetics set the tone, which translates into observable dynamics and, ultimately, workable mechanics.
When working with games, it's helpful to consider both the designer's and the player's perspectives. This helps us see how even small changes in one layer can cascade into others. Additionally, thinking about the player contributes to the development of experience-oriented design (as opposed to function-oriented design).
Therefore, we will begin our study with a discussion of Aesthetics, move on to Dynamics, and end with fundamental Mechanics.
What makes a game “fun”? How do we recognize a particular form of entertainment when we see it? Talking about games and gameplay is difficult because the vocabulary we use is relatively limited.
When describing the aesthetics of a game, we want to move away from words like “fan” and “gameplay” and move on to more specific vocabulary. This includes, but is not limited to, the taxonomy listed here:
Sensory pleasure. Game as a feeling of pleasure
Fantasy . Game as fiction
Story (Narrative). Game as drama
Challenge. Game like an obstacle course
Community (Fellowship). Game as a social structure
Discovery. The game is like uncharted territory
Expression. Game as self-knowledge.
Submission. Game as entertainment/distraction
For example, consider the games Charades, Quake, The Sims and Final Fantasy. While each of these are “fun” on their own, it's more informative to consider the aesthetic components that create the respective player experience:
Charades: Community, Self-Expression, Challenge
Quake: Challenge, Sensory pleasure, Competition, Fantasy.
The Sims: Discovery, Fantasy, Self-expression, History
Final Fantasy: Fantasy, History, Self-expression, Discovery, Challenge, Immersion.
Here we see that each game has multiple aesthetic goals, to varying degrees. In “Charades” comes to the fore communication (community)not trial; in Quake trial – the main element of gameplay. And although it doesn't exist Great Unified Theory games or a formula that details the combination and relationship of elements that lead to “fun”, this taxonomy helps us describe games by shedding light on how and why different games attract different players or the same players at different times.
Aesthetic models
Using an aesthetic vocabulary as a compass, we can define gameplay patterns. These models will help us describe the dynamics and mechanics of gameplay.
For example, Charades and Quake are competitive games. They are successful when different teams or players in these games emotionally interested in victory over each other. To do this, it is necessary that the players have opponents (in Charades teams compete, in Quake the player competes with computer opponents) and that all sides want to win.
It's easy to see that maintaining competitiveness and clear feedback about who is winning is very important in competitive gaming. If the player doesn't see a clear victory condition or feels like they can't win, the game becomes much less interesting.
Dynamic models
Dynamics works to create aesthetic impressions. For example, Call created by things like time pressure and opponent play. Communication may be encouraged by sharing information between certain session participants (the team) or by providing victory conditions that are more difficult to achieve alone (such as capturing an enemy base).
Self-expression arises from dynamics that encourage individual users to leave their mark: systems for purchasing, building or earning in-game items, designing, building and modifying levels or worlds, and creating personalized, unique characters. Story arises due to dynamics that promote tension build-up, release, and resolution.
As with aesthetics, we want to keep our discussion of dynamics as specific as possible. By developing models that predict and describe gameplay dynamics, we can avoid some common design pitfalls.
For example, a model of two six-sided dice will help us determine the average time it would take a player to get around the board in a game of Monopoly, given the probabilities of different rolls.
Similarly, we can identify feedback systems in gameplay to determine how certain states or changes affect the overall state of the gameplay. In the game of Monopoly, as the leader or leaders become increasingly wealthy, they can punish players with increasing efficiency. Poor players are getting poorer.
As the gap widens, few (and sometimes only one) of the players are truly interested in the game. Dramatic tension and independence are lost.
Using our understanding of aesthetics and dynamics, we can imagine how to fix Monopoly, either by rewarding lagging players to keep them at a reasonable distance from the leaders, or by making progress more difficult for wealthy players. Of course, this may affect the game's ability to recreate the reality of monopoly practices – but the reality is not always “fun”.
Mechanics
Mechanics are the various actions, behaviors, and control mechanisms provided to the player in a game context. Together with the content of the game (levels, assets, etc.), the mechanics support the overall dynamics of the gameplay.
For example, the mechanics of card games include shuffling, dealing, and betting, from which dynamics such as bluffing can arise. Shooter mechanics include weapons, ammo, and spawn points, which sometimes lead to things like camping and sniping. Golf mechanics include balls, clubs, sand traps and water hazards that sometimes cause clubs to break or sink.
Setting up game mechanics helps us fine-tune its overall dynamics. Consider our Monopoly example. Mechanics that could help lagging players could include bonuses or subsidies for less successful players, as well as penalties or taxes for wealthy players, perhaps calculated upon crossing the Start square, leaving prison, or having monopolies above a certain value. By applying such changes to the core rules of the game, we could keep lagging players interested and competitive for longer.
Another solution to the lack of tension when playing Monopoly for long periods of time would be to add mechanics that encourage time pressure and speed up the game. Perhaps depleting resources over time with a constant rate tax (to get people to spend faster), doubling all monopoly payouts (to get players to differentiate quickly), or randomizing all properties below a certain value threshold.
Tuning
Obviously, the last stage of Monopoly analysis involves testing and tuning the game. By iterating on fines, tax rates, or reward and penalty thresholds, we can refine the Monopoly gameplay until it becomes balanced.
When tuning, aesthetic vocabulary and MDA models help us articulate design goals, discuss the game's shortcomings, and measure our progress through the tuning process. If our taxes in Monopoly require complex calculations, we can take away the sense of investment from the player, making it difficult for them to keep track of their money and therefore their overall progress or competitive position.
Likewise, our dynamic models help us identify where problems might be coming from. Using the D6 model, we can evaluate proposed changes to board size or layout and determine how much the changes would increase or decrease the length of the game.
MDA in game production
Now let's look at the possibility of developing or improving the AI component of the game. It is often tempting to idealize AI components as “black box” mechanisms that, in theory, can be implemented into a wide variety of projects with relative ease. But, as the method implies, game components cannot be assessed in a vacuum, apart from their impact on the behavior of the system and the player's experience
First pass
Let's look at an example of a game about a nanny [Hunicke, 2004]. Your manager has decided that it would be useful to prototype a simple game AI for playing tag. Your player will be a nanny who needs to find and put one baby to bed. The demo will be designed to showcase simple emotional characters (such as a baby) in games aimed at children aged 3 to 7 years.
What are the aesthetic goals of this design? Study And self-expressionare probably more important than trial. Therefore, the dynamics here are optimized not for “winning” or “competition”, but for the child to express emotions such as surprise, fear and anticipation.
Hide spots can be manually marked, and the paths between them hard-coded; much of the game logic will be devoted to maneuvering the child and creating child-like reactions. Game mechanics include talking to the child (“I see you!” or “Boo!”), chasing the child (using an avatar or mouse), sneaking, tagging, and so on.
Second pass
Now let's look at a version of the same design, created to work with the Nickelodeon “Rugrats” franchise and aimed at girls 7-12 years old. From an aesthetic point of view, the game should seem more complex – perhaps there is some story (requiring several “levels”, each of which introduces a new part of the plot and associated tasks).
In terms of dynamics, the player can now track and interact with multiple characters at the same time. We can add time pressure mechanics (like getting everyone to bed before 9pm), include a “mess factor” or track characters' emotions (dirty diapers cause crying, crying loses points), and so on.
For this design, static paths will no longer be suitable, and it will probably be nice if they choose their own secluded places. Will each baby have individual characteristics, abilities or problems? If so, how will they show these differences to the player? How will they track their internal state, reason about the world, other kids and the player? What tasks and actions will the player need to perform?
Third pass
Finally, we can imagine this same game as a full-fledged military-strategic simulator – like Splinter Cell or Thief. Now our target audience is men 14-35 years old.
Aesthetic goals are now expanded to include the element fantasy (role-playing as a military elite hunting spies or an outcast looking for prey), and trialmay probably border on by immersion. In addition to the exciting plot (story), full of intrigue and suspense, the player will expect coordinated activity from opponents – but probably much less emotional expression. For that matter, agents should express fear and disgust at the mere hint of his presence.
Dynamics may include the opportunity to earn or acquire powerful weapons and spy equipment, as well as to develop tactics and techniques for covert movement, deception, evasion and escape. Mechanics include extensive technology and skill trees, a variety of enemy unit types, levels or zones with variable ranges of mobility, visibility, field of view, and so on.
Agents in this space, in addition to coordinating movement and attacks, must operate on a wide range of sensory data. Reasoning about a player's position and intentions should create complexity but contribute to overall success. Will the enemies be able to overcome obstacles and navigate difficult terrain, or will you “cheat”? Will sound propagation be “realistic” or will simple distance-based metrics suffice?
Summing up
Here we see that simple changes to a game's aesthetic requirements entail mechanical changes to its AI on many levels, sometimes requiring the development of entirely new systems for navigation, reasoning, and strategic problem-solving.
Conversely, we see that there are no “AI mechanics” as such – intelligence or consistency comes from the interaction of AI logic with gameplay logic. Using the MDA method, we can reason explicitly about aesthetic goals, identify the dynamics that support those goals, and then expand the range of our mechanics accordingly.
Conclusions
MDA supports a formal, iterative approach to design and configuration. It allows us to reason explicitly about specific design goals and anticipate how changes will affect each aspect of the method and the resulting designs/implementations.
By moving between the three levels of abstraction of MDA, we can conceptualize the dynamic behavior of game systems. Understanding games as dynamic systems helps us develop methods for iterative design and improvement – this allows us to control undesirable outcomes and tune desired behavior.
Additionally, by understanding how formal gameplay decisions impact the final user experience, we can better decompose those experiences and use them for new developments, research, and criticism accordingly.
Authors of the article: Robin Hanicke (game designer Journey); Mark LeBlanc; Robert Zubek (author of the book “Elements of Game Design. How to Create Games You Can't Stop Playing”)
