EMERGENCE
Emergence describes the phenomenon of systems that consist of relative simple, interacting parts, creating rich, unforeseen patterns of behavior after being set into motion. Within the science of complexity that studies emergent systems, games are a popular and intuitive illustration. Classic board games such as chess, checkers, or Go create nearly endless variety in possible game states; all the possible configurations of pieces on the board that are the result of playing the game from its starting conditions. Series of successive moves create patterns of behavior that are the building blocks for strategic play. Anyone familiar with these games will analyze a game state for patterns that can be understood as lures, traps, defensive positions, offensive potential, and so on. To play them successfully, players must be able to read the game state in this way. This is both the beauty and the difficulty of mastering an emergent game, and requires considerable experience with playing the game. At the same time, for emergent games, “the whole is more than the sum of its parts.” The sheer number of possible game states and the endless variety in subtle gameplay patterns stands in stark contrast with the number of rules that define these games: all the rules of either chess, checkers, or Go can easily be printed on a single sheet of paper. These classic games are the epitome of elegance and effectiveness in complex systems designed for emergence.
Video games share this aptitude for emergence with board games. Although video games offer new opportunities to include much more content in the game, the richest games with the highest replay value tend to rely on emergent techniques to create unending variation in their gameplay (Juul, 2002). Players often enjoy the freedom that emergence brings in games, even if that freedom leads to unforeseen player tactics such as rocket-jumping in Quake (id Software, 1996) or grenade climbing in Deus Ex (Ion Storm, 2000). Video game designers do well to embrace emergence as a design philosophy and design goal (Smith, 2001).
For those that study and design games, emergence is a serious challenge. By its very nature, emergence in games can only be appreciated after the game has been built and set into motion. Simply looking at rules reveals surprisingly little about the quality of the gameplay. Because of this, designing a classic board game with enjoyable, emergent gameplay is probably one of the hardest design tasks anyone can set for him or herself, one that sometimes can seem to involve more luck than skill, in order to come with just the right mixture of rules. A common design exercise given to new students of game design is to take a simple, published board game and change its rules (for example, see the Up the River (1988) exercise described in Fullerton, 2008). More often than not, this exercise illustrates how fragile the balance of a game’s components really is. It is incredibly easy to break a functional game, while it is incredibly difficult to improve one. Starting from a poorly-designed and badly-balanced game does not make it much easier. Even in that case, most rule changes will not improve the game. It takes skill, experience, and a lot of hard work to find exactly those changes that move the game’s dynamic behavior in the desired direction.
Effectively designing games of emergence requires specialized knowledge and an open attitude that embraces the inherent uncertainty of the venture. Likewise, the study of emergent games requires similar knowledge, and a true appreciation of the open-ended nature of game systems is paramount. This essay explores the knowledge required to fully appreciate games of emergence. It studies the philosophy behind their design, and seeks to resolve the paradox of deliberately engineered, emergent gameplay.
Complex Systems
From the science of complexity, we know two important things: (1) emergence exists somewhere on the border of chaos and order, and (2) a number of structural features of the complex system give rise to dynamic, emergent behavior (for example, the number of parts, the number of connections between the parts, and the presence of feedback loops within the system).
The behaviors of systems can be classified into four rough categories: they can be ordered, periodic, emergent, or chaotic (Holland, 1998; Salen and Zimmerman, 2004). The boundaries between these four categories are fuzzy and together they create a something of gradual scale with ordered behavior on one end and chaotic behavior on the other. Examples of all types of behavior can be found in games. Scripted level design is ordered behavior; the players’ movement around the Monopoly (Parker Brothers, 1934) board creates a periodic system. The distinct gameplay phases in a game of Civilization (MicroProse, 1991) (early expansion, consolidation, direct conflict, overseas colonization, space race, and so on) are the result of the game system’s emergent behavior, while the random numbers generated by rolling dice are the result of a chaotic interplay of forces as the dice bounce on the table. Games typically employ mechanisms that push the system’s overall behavior toward one of the two ends: random factors typically push a game toward chaotic behavior, while a predesigned progression of fixed levels pushes a game toward more ordered behavior. As emergence is to be found between the two extremes, the balance between chaotic and ordered mechanisms is vital. Individual games each have their own balance: some lean more toward predictable progression, while others are more emergent. The categories of games of progression and games of emergence (Juul, 2002, 2005) describe both ends of these gradual scales. Many games, however, combine elements of both categories. For example, the physics simulation in Half-Life (Valve, 1998) creates highly emergent gameplay, while its level structure and story creates a linear progression through the game.
Games do not need to rely only on random number generators to create emergent gameplay; a game can be made to have unforeseen outcomes without resorting to random mechanisms at all. For example, chess, checkers, and Go do not use dice, yet they are games that display highly emergent behavior. If a system is sufficiently complex, that is, if it consists of enough interconnected parts, it can already display sufficiently emergent behavior. Studies by the mathematician Stephan Wolfram reveal that systems that consist of parts that are all easily described in isolation as simple state machines, create emergent behavior when the parts are connected so as to allow long-range communication and maintain a sufficient level of activity (Wolfram, 2002). In this case, “level of activity” means that individual parts must change its state frequently, and long-range communication means that these state changes trigger a cascade of subsequent state changes that propagate through the system over long distances or over long periods of time. For these effects to occur, it is not the number of parts (nor the number of states they can be in), but the number of connections between the parts that is the most important factor: emergent behavior is more likely to occur in systems with a high number of connections.
An important structural feature that has been identified by scientists investigating complex systems and game designers alike are feedback loops in the (game) system (LeBlanc, 1999; Fromm, 2005; Adams and Dormans, 2012). A feedback loop is created when activity created by the state change of a part of the system is propagated through the system in such way that it ultimately feeds back to the original part to create new state changes with more activity as a result. Feedback loops are common in games. For example, most real-time strategy games involve some sort of resource harvesting. In these cases, resources can be used to buy or build new worker units that in turn produce more resources. This creates a positive feedback loop: an effect that strengthens itself and quickly spirals out of control as more resources lead to more workers, which in turn lead to more resources, and so on. Negative feedback loops, where the feedback effects act against the initial change that triggered them, are also frequent. For example, in Civilization, growing cities require more and more food and money to keep the citizens alive and content. In this case, negative feedback creates a type of friction on a city’s growth that causes it to stabilize on a particular size based on the available food, technology, and gameplay choices made by the player. Emergent games often delicately balance multiple positive and negative feedback loops interacting within its mechanics.
Creating interconnected feedback mechanisms is an important aspect of the design of a game’s core mechanics (Adams and Dormans, 2012), which requires a detailed understanding of the game’s structural features and mechanics. However, the real goal is to create a compelling gameplay experience. To this end, a designer must also keep an eye on the larger patterns in the game’s emergent behavior. As was already mentioned, emergent games typically go through a number of distinct gameplay phases. These phases are the result of the game mechanics’ complex interactions, but frequently they also depend on more ordered and predesigned aspects such as restrictions imposed by the level design or pre-scripted narrative triggers. For instance, on the one hand, in a real-time strategy game, close proximity to vital resources and little hostile activity will likely to lead to sessions where players can build and consolidate fairly easily before starting to explore the map and attack the enemy. On the other hand, a level with scattered resources and more active foes will create sessions where building and consolidation phases are much shorter and will alternate with offensive and exploration phases more frequently. Game mechanics, level design, and story-telling are the main materials with which the game designer composes the game’s overall pattern of behavior.
A relatively simple example of such a “composition” of two phases can be found in Tetris (Pajitnov, 1984). Initially, the player has plenty of time to place falling tetrominoes efficiently and clear lines fast enough to prevent the play space from getting cluttered with blocks. However, a positive feedback mechanism drives the game to its conclusion: as tetrominoes block up the game place the player has less and less time to place the tetrominoes efficiently. This makes the player’s task more and more difficult. As s/he clears more and more lines, the game speeds up. Eventually the game goes too fast and the positive feedback quickly spins out of control, ending the game mercifully quickly. It might be argued that there is also a third phase in between these two phases, where the player struggles to keep up with the game but is not quite losing yet. During this time, the player might get lucky and receive the particular tetrominoes that are needed to clear lines in quick succession. Sometimes the player, with a combination of skill and extra effort, might even push the game back to the initial phase. Tetris always progresses through these two (or three) gameplay phases as the simple and entirely ordered mechanism that causes the game to speed up will always eventually push the game into the last, losing phase.
In the case of Tetris, the progression through the different phases is always the same (dismissing cases where players are able to push the game back from a losing state to a stable state, either by luck, effort, or by deliberately performing poorly initially). In games where the mechanisms that cause shifts between phases are more complex, the progression of phases will take on different shapes. For example, the mechanisms that cause shifts in the phases of Civilization depend much more on the decisions of individual player and his/her AI opponents. At the same time, the dramatic shifts between phases of expansion and conflict do contribute greatly to the gameplay experience. In order to understand and harness emergent gameplay, game designers and game analysts alike should study the way different structural aspects of game mechanics create gameplay phases, and what mechanisms can be used to create dramatic shifts between gameplay phases at exactly the right moment. Whereas in other (non-emergent) media, designers have complete control over the artifacts they create, emergent games require a different approach altogether.
Embracing Emergence
To fully appreciate emergence in games, designers and analysts need to embrace the open-endedness and uncertainty that it brings; in order to capitalize on the opportunities emergent games offer, designers and analysts need to adopt an approach to design that breaks away from the classic authorial perspective. Games of emergence are not media expressions in the same way books, films, or paintings are. They are machines that produce narrative, cinematic, or picturesque expressions. To paraphrase Espen Aarseth: they are a form of “ergodic” literature, which means that nontrivial effort is required by players in order to traverse the game and to produce their own unique experiences (Aarseth, 1997). This means that the “author” of a game does not have full control over what experiences are generated by the game; this responsibility is shared with the player and the game itself. Although it might be argued that the author produces the game and therefore has some level of control over the events it might generate (Murray, 1991), in the case of emergent games, the number of possible events is far too large to be controlled in a strict sense. Unexpected and unforeseen events are always possible; in fact, they are to be expected. In the end, the author cannot predict all possible combinations and variations produced by players and the game system.
The distinction between “textons” and “scriptons” made by Espen Aarseth (1997) provides a useful starting to explore this issue in more detail. Textons are the pieces of text included in the game (the models, scripted dialog fragments, cut-scenes, designed levels, and so on), while the scriptons are the expressions that are the result of playing. Obviously, the number of possible combinations of textons leads to a vast number of scriptons. For instance, there are only seven different tetrominoes in Tetris, yet the number of different ways they can be placed in the game space is far beyond count. The game designer has full control over textons but far less over the resulting scriptons. This does not mean any combination of textons is a scripton that could be produced by the system. For example, the game of chess has all the pieces that could represent the actors of a medieval romance (knights, kings, queens, bishops, castles, and two different colors to denote two different families), but the rules of the game only allow them to be combined in such way that they always form an abstract representation of battle. Only looking at textons and scriptons does not give a full account of what type of messages can emerge from a game.
The mechanics of a game define what operations are possible to combine different textons into scriptons. It is through these operations that the designer does execute some indirect control over the meaningful behavior of the system as whole. It might be argued that for certain games, and games of emergence in particular, these operations on elements of the system are more important than the textons. To use Ian Bogost’s words: games are all about the “unit operations” within the system (2006), or to quote Brenda Brathwaite: “the mechanic is the message” (2009).
One of the most intriguing aspects of games of emergence is how these games can become so much more than their designers originally intended. Games of emergence can act as catalysts to explore new ideas, and are increasingly used to do so. It is because of this that Harvey Smith (2001) is enthusiastic about grenade climbing in Deus Ex. Grenade climbing is strategy that leverages the mechanics of the game’s proximity mines. Players can stick these mines onto walls, where they arm themselves in five seconds. However, due to the implementation of the game’s physics, players are able to jump on top of these mines, allowing them to place another proximity mine a little higher, to create a ladder of some sort. This allowed players to take shortcuts the designers never intended and reach locations in the game that would otherwise be unreachable. On the one hand, it might seem that the players simply came up with a gameplay strategy the game designers did not foresee, but on the other hand, it might be said that the consistent, yet emergent behavior of the game inspired players to use all their creativity and explore ideas outside the box. Management simulation games and policymaking games could benefit tremendously from this effect; and it also means that the game designer does not have to know all the answers to a problem beforehand. A game can be a vehicle to explore a problem and produce answers in an efficient and safe way. In fact, the success of the nineteenth-century simulation war game Kriegsspiel (Von Reisswitz, 1824) largely lies in the way it prepared German officers to the eventualities of the upcoming battle: they would have run through several scenarios before a single gun was fired (Gray, 2008). The game did not contain any of the answers to the tactical challenge offered by the battle, but it offered players a machine with which they could find those answers themselves.
However, most entertainment games seem to refrain from tapping into this vast potential (as do most “serious” games). For example, in the case of games with procedurally-generated content, arguably the most emergent games of all, there is a huge variety of scriptons based on very few textons and some very sophisticated procedure, yet gameplay seems to vary far less. Classic examples of this type game are the experimental and independent games Rogue (Artificial Intelligence Design, 1983) and Dwarf Fortress (Adams, 2006), but also triple-A games such as Diablo (Blizzard Entertainment, 1996) make use of these techniques. In these cases, the gap between the textons (individual dungeon tiles, and creatures) and the scriptons (a dungeon-crawling adventure) is huge and bridged by an intermediate form (generated dungeons and quests). However, at the same time, while the details of dungeons and quests that are generated might vary indefinitely in these games, their type does not vary as much. As it becomes clear from Ernest Adams’s satirical “Letter from a Dungeon” (2000), they all grow alike very quickly. In other words, a system’s ability to generate an infinite number of scriptons does not necessarily lead to an equal number of meaningful interpretations or significant sessions of play. This effect can be found in other emergent games, too: Civilization generates a huge number of possible game states and supports a large number of different strategies, yet an ideological interpretation of the game reveals that its message is always the same and supports a Western, technocratic world view (Kline et al., 2003, Galloway 2006). Even games that claim to offer a scale of moral choices, such as Black & White (Lionhead Studios, 2001), Fable (Big Blue Box Studios and Lionhead Studios, 2004), BioShock (Irrational Games, 2007), or inFamous (Sucker Punch Productions, 2009), almost, without exception, resort to ridiculous stereotypes of what it means to be good or evil, as was eloquently pointed out in the Zero Punctuation review of inFamous (Croshaw, 2009). These games reveal nothing about good or evil that the designers have not put in beforehand. All these games successfully create an open world for the player to explore, but the openness of the exploration and the action mechanics seems to be in stark contrast with the less emergent design of the games’ narrative structures. On the level of physics and action mechanics, they clearly fall into Jesper Juul’s category of games of emergence on one level, while on the level of narrative they are games of progression.
Conclusion
The question remains whether emergent games often are restricted in the number of significant gameplay results because designers subconsciously cling to a high level of control out of habit (after all, for a long time our culture celebrated, and still celebrates, the individual, creative genius of the creator of any form of art), or because they lack experience to create games that are truly emergent and that can be used to explore a vast array of possible significant strategies? It is safe to say that complex, emergent systems as a media form are relatively unexplored. Although games have been around for a long time, the ubiquitous presence and procedural power of video games is far more recent. Likewise, the science of complexity is a young field and most of its results have still to find their way into the common parlance and knowledge of game designers and analysts. Yet, as long as both are willing to embrace the designer’s new role of indirect composer of gameplay phases (instead of direct author of media texts) and study the structures of games that contribute to the emergence of stable gameplay phases as well as contribute to dramatic shifts from phase to phase, the future of emergent games is bright, no matter how much research and experimentation there is left to do.
References
Aarseth, E. J. (1997). Cybertext: Perspectives on ergodic literature. Baltimore: Johns Hopkins University Press.
Adams, E. W. (2000) Letter from a dungeon. Retrieved August 8, 2012, from www.designersnotebook.com/Columns/027_Letter_From_a_Dungeon/027_letter_from_a_dungeon.htm.
Adams, E. W., and Dormans, J. (2012). Game mechanics: Advanced game design. Berkeley, CA: New Riders Games.
Bogost, I. (2006). Unit operations: An approach to videogame criticism. Cambridge, MA: MIT Press.
Brathwaite, B. (2009). My games: The mechanic is the message. Retrieved August 8, 2012, from http://mechanicmessage.wordpress.com/.
Croshaw, B. (2009). Zero punctuation: inFamous. Retrieved August 8, 2012, from www.escapistmagazine.com/videos/view/zero-punctuation/767-inFamous.
Fromm, J. (2005). Types and forms of emergence. Retrieved August 8, 2012, from http://arxiv.org/abs/nlin/0506028.
Fullerton, T. (2008). Game design workshop: A playcentric approach to creating innovative games, 2nd edn, Burlington, MA: Morgan Kaufman.
Galloway, A. R. (2006). Gaming: Essays of algorithmic culture. Minneapolis: University of Minnesota Press.
Gray, B. (2008). What is wargaming? Retrieved August 8, 2012, from www.hmgs.org/history.htm.
Holland, J. H. (1998). Emergence: From chaos to order. Oxford: Oxford University Press.
Juul, J. (2002). The open and the closed: Games of emergence and games of progression. In F. Mäyrä (Ed.), Proceedings of computer games and digital conference (pp. 323–329). Tampere: Tampere University Press.
Juul, J. (2005). Half-real: Video games between real rules and fictional worlds. Cambridge, MA: The MIT Press.
Kline, S., Dyer-Witheford, N., and De Peuter, G. (2003). Digital play: The interaction of technology, culture, and marketing. Montreal: McGill-Queen’s University Press.
LeBlanc, M. (1999). Formal design tools: Feedback systems and the dramatic structure of completion. Retrieved August 8, 2012, from http://algorithmancy.8kindsoffun.com/cgdc99.ppt.
Murray, J. H. (1991). Anatomy of a new medium: Literary and pedagogic uses of advanced linguistics computer structures. Computers and the Humanities, 25, 1–14.
Salen, K., and Zimmerman, E. (2004). Rules of play: Game design fundamentals. Cambridge: MIT Press.
Smith, H. (2001). The future of game design: Moving beyond Deus Ex and other dated paradigms. Retrieved August 8, 2012, from www.itu.dk/stud/speciale/worlddomination/files/pelle/The%20Future%20of%20Game%20Design.doc.
Wolfram, S. (2002). A new kind of science. Champaign, IL: Wolfram Media Inc.