The conception of games as complex systems with emergent properties is so prevalent in the discourse of both game design and Game Studies that it would be impossible to cite its origins. Descriptions of emergence can be found in a diverse array of contexts, from books on “popular science” (Johnson 2001) to “game design theory” (Juul 2002; Salen and Zimmerman 2004). So what, precisely, do we mean by “emergence?” Emergence is a phenomenon that falls under the general rubric of “complex systems” or “complexity theory,” a set of ideas that span an unwieldy array of fields and disciplines and as a result has become a fulcrum for interdisciplinary research. The Santa Fe Institute, one of the preeminent centers for the study of complex systems in the United States, encompasses fields as diverse as social science, economics, mathematics, game theory (a branch of applied mathematics and economics unrelated to “game studies”), ecology, evolution, neuroscience, intelligent systems and network infrastructures. (REF: SFI web site) The Human Complex Systems group at University of California Los Angeles embraces every permutation of its theme, from economics to urban planning and computer-generated “synthetic cultures” to multiplayer online games. (REF: HCS web site)
The term “emergence” describes how complex, often decentralized, systems self-organize in ways that cannot be predicted by their underlying structures or rule sets, nor by the individual behavior of agents within the system. (Bar-Yam 1997)Anthills, freeways, neural networks, stock markets, terrorist cells, cities, the internet and computer games are among the examples used to describe emergence (Johnson 2001). These disparate systems share in common a display collective behaviors and even collective “intelligences” that arise out of, and yet, transcend the actions of the individual parts or elements. According to Steven Johnson, author of Emergence: The Connected Lives of Ants, Brains, Cities and Software, complex systems exhibit emergence because they
…solve problems by drawing on masses of relatively (simple) elements, rather than a single, intelligent ‘executive branch.’ They are bottom-up systems, not top-down. They get their smarts from below. In more technical language, they are complex adaptive systems that display emergent behavior. In these systems, agents residing on one scale start producing behavior that lies one scale above them: ants create colonies; urbanities create neighborhoods; simple pattern-recognition software learns how to recommend new books. The movement from low-level rules to higher-level sophistication is what we call emergence. (Johnson 2001)
(REF: Page #)
It is not insignificant that one of the key characters Johnson features in Emergence is Will Wright, designer of the games SimCity (Wright 1989), The Sims (Wright 2000) and The Sims Online (Wright, Trottier, and Chalmers 2002). Even in the context of this otherwise “serious” book, games make regular appearances. That games produce emergence is a de facto assumption throughout the game studies field. Pioneering media scholar Janet Murray has described one of the properties of computational media as being “procedural,” or rule-based. Rule-based systems have a greater tendency towards emergence because they have a larger possibility space with affordances for more varied outcomes. While the rule system itself does not have to be complex in a procedural system, as illustrated by “Conway’s Game of Life,” (a classic A-life simulation) the traditional Japanese game of Go, or even chess, simple rules systems can produce complex, emergent outcomes. Using examples of board games, sports, most action games and all strategy games, “Ludologist” Jesper Juul argues that emergence is “the primordial game structure, where a game is specified as a small number of rules that yield large numbers of game variations, that the players must design strategies for dealing with.” “Progression” he describes as “the historically newer structure” in which we find “cinematic storytelling ambitions” in this otherwise indigenously procedural and hence emergent medium (Juul 2002; Juul 2004). In Rules of Play, Salen and Zimmerman look in-depth at notions of games as complex systems and emergence as an outcome of the interaction of rules. (Salen and Zimmerman 2004) In my 2002 paper on “emergent authorship,” I describe a new model for storytelling in which players themselves contribute to narratives in games such as The Sims,Ultima Online and EverQuest. (REF: Pearce 2002) Cindy Poremba’s Master’s thesis provided a further analysis of the player as co-creator within the context of these emergent story systems. (REF: Poremba thesis) These ideas build on Henry Jenkins’ notion of “textual poaching,” in which fan cultures, such as Star Trek fans, aka “Trekkies,” develop their own emergent narratives from the kit of parts provided by the television series. (REF: Jenkins 1992)
So what, precisely, is emergence, and how might it be studied? In his essay for the book Virtual Worlds: Synthetic Universes, Digital Life and Complexity, Yaneer Bar-Yam, President of the New England Complex Systems Institute, defines emergence as a set of “collective behaviors” in which all the parts are “interdependent,” arguing that the more distinct and specialized the individual interdependent behaviors, the more complex the collective behavior likely to arise (Bar-Yam 1999). Bar-Yam describes emergence as
…what parts of a system do together that they would not do by themselves; collective behavior.
…what a system does by virtue of its relationship to its environment that it would not do by itself.
…the act of process of becoming an emergent system.
According to (1) emergence refers to understanding how collective properties arise from the properties of the parts. More generally, it refers to how behavior at a larger scale of the system arises from the detailed structure, behavior and relationships at a finer scale. In the extreme, it is about how macroscopic behavior arises from microscopic behavior.
In discussing methodology, Bar-Yam suggests a holistic approach to observing the relationship between the parts and the system as a whole:
…emergent properties cannot be studied by physically taking a system apart and looking at the parts (reductionism). They can, however, be studied by looking at each of the parts in the context of the system as a whole. This is the nature of emergence and an indication of how it can be studied and understood.
To describe this process, Bar-Yam invokes the metaphor of “[seeing] the forest and the trees at the same time... We see the ways the trees and the forest are related to each other” (Bar-Yam 2000b). Sociologist C. Wright Mills has drawn upon the same metaphor to describe the essential character of what he calls “the sociological imagination” (Mills 1959).
This apt metaphor illustrates the key challenge of studying emergence in large-scale social systems. This type of research necessitates a methodology that enables one to observe and analyze phenomena at different scales simultaneously. In other words, it must enable us to look at the behavior of individual units in a complex system, their relationship to each other, and the overarching patterns of the system as whole, all at the same time. It is also crucial to be able to observe the system dynamics in-progress as well as their outcomes. Capturing its evidence exclusively after-the-fact, either through surveys or forensic evidence, such as artifacts, will not allow for a complete understanding of patterns of emergence. In addition, we are faced with the problem of observing the relationship between the play community and the play environment.
As Bar-Yam points out, “One of the problems in thinking about the concepts of complex systems is that we often assign properties to a system that are actually properties of a relationship between the system and its environment.” This is particularly significant to the research described here, where the relationships between players, as well as the players’ relationship to the environment of the virtual world, are central:
When parts of a system are related to each other, we talk about them as a network, when a system is related to parts of a larger system, we talk about its ecosystem.
We thus arrive at the concept described earlier of framing massively multiplayer player worlds that fall along the “fixed synthetic” and “co-created” worlds spectrum as “ecosystems of play” in which “networks” of players engage in various emergent behaviors. This is where the distinctions between different types of worlds becomes important: each ecosystem provides particular characteristics and affordances which effect the emergent behavior of networks within it. As we shall see, the group exhibits patterns of emergence that transcend any particular world, but these are made explicit through interactions unique to the affordances of each play ecosystem.
One of the critical properties of complex systems is feedback. In cybernetics, feedback is defined as a phenomenon in which some portion of the output of a system is passed through the input. This can be used to describe machines that utilize feedback systems, the classic example being a thermostat on a heater. The thermostat continually reads the temperature and makes adjustments accordingly.
Within networked social systems, feedback can be a powerful engine for large-scale social emergence, and the accelerated forms of emergence seen in these systems are a direct result of the designed affordances of the software. Examples of this on the Internet include iTunes, MySpace and YouTube, each of which has grown exponentially since its inception through feedback. This process, epitomized by YouTube, can be described as follows: the more people who watch, the more people who upload videos; the more people who upload videos, the more people who watch. Networks are particularly good at processing feedback since feedback can move quickly through the system and be distributed to a large number of inputs. This research concerns the ways in which both the social context of play and the design of the game software itself facilitate this feedback process.
The qualities of properties of play are critical. Play can be viewed as a particular type of engine for emergence by virtue of its feedback dynamics. Play is inherently spontaneous and experimental, and therefore, players will find themselves responding to social feedback in a very different way than they might in other contexts. The common types of emergence seen within virtual multiplayer games and virtual worlds illustrate this point. As we’ve seen, they include online weddings, game-wide protests, social organizations such as guilds or social groups, various types of social and fashion trends, and extra-virtual phenomena such as fan sites and selling of virtual characters, items or currency on eBay.
The “play frame” sets the stage for many of these phenomena, but the virtual environments themselves also have particular properties that lend themselves to emergence:
Discrete: Virtual worlds are (mostly) closed systems, discrete synthetic environments that possess and maintain a consistent set of internal rules. Within that closed system, we can observe classic properties of emergence, such as feedback, and multi-generational patterns. (Bar-Yam, 1997; Johnson, 2002) In addition, they also have a variety transactions with worlds outside themselves, which can both influence in-world emergence and produce extra-virtual forms of emergence.
Open-Ended: Both social virtual worlds and game worlds are open-ended. Unlike many traditional games, including most single-player digital games, they do not have and end state that can be considered “winning.”
Persistent: They are “persistent state worlds.” This means that whatever is done in them remains and is cumulative over time. This includes both the individual player identity and the world as a whole. This is a vital component to create the feedback needed to produce emergence. First person shooter games, for instance, while they do exhibit their own forms of emergence, do not possess persistence of this type within the world itself.
Synchronous & Asynchronous: Because the game is persistent and remains “on” at all times, players can inhabit and construct the world asynchronously from one another, thus adding another vital ingredient for feedback and emergence.
Long-Term: Engagement in multiplayer games and virtual worlds is long-term and emergent behaviors happen over time. Typical console games, as an example, allow for anywhere from ten to forty hours of total play time. While this provides for some emergence, the significantly higher time commitment of MMOGs (twenty hours a week on average and sometimes significantly more) (Seay, Jerome, Sang Lee, and Kraut 2004; Yee 2001) (REF: Castronova?) provides for a much greater array of emergent outcomes. It is possible for players to maintain involvement in the game for as long as it is operating, although the “churn rate” (rate of subscription turnover) is more typically around 18 months. (Appelcline 2000-2006; Yee 2001) (REF: Koster?) Churn can also produce emergent behavior, such as a mass-exodus to a new game, which is common among players of medieval fantasy games.
Accelerated: Social phenomena in MMOGs tends to happen at an accelerated rate, in spite of the fact that tasks often take significantly longer to perform tasks than in the physical world. Basic tasks such as communication take longer, yet players often report losing track of time and the sense that “time flies.” At the same time, there appears to be a phenomenon of time compression in which social processes that would ordinarily take much longer are perceived and observed to occur at a highly accelerated rate. Friendships and romantic relationships appear to develop more quickly, and the growth and decline of communities seems to progress much faster than would be the case in “real world” settlements.
Networked: As mentioned earlier, MMOGs and MMOWs are by definition populated. The more people, the larger the possibility space for emergence.
Diverse: As Bar-Yam points out, the more specialized the units in a complex system, the more complex the system, and the more opportunities for emergent behaviors. In more homogenous systems, behavior is relatively uniform, so emergence is less likely to happen as behaviors are less likely to diverge from their initial purpose. (REF Bar-Yam new article?) Surowiecki, author of The Wisdom of Crowds, points out that collective intelligence is much more likely to occur in groups that are diverse than those that have uniform skills and abilities. (REF: Surowiecki)
Based on the preceding definition of emergence and the characteristics above, the following criteria were used to select a study subject the study of which would provide us with deeper insights and understandings about the nature and process of emergence in online games:
Events Over Time. The study had to be somewhat longitudinal (in “Internet years”) in order to detect emergent behavior over time; eighteen months was selected as the time frame for practical reasons. This timeframe is also line with the “churn” figure described above, and also parallels timeframes for field studies in qualitative anthropological and sociological research, one year being the time-frame for typical fieldwork in anthropology.
Scale. The group studied had to be sufficiently large to exhibit emergent behavior patterns, yet small enough for a single researcher to feasibly study at multiple levels of magnification. The main focus of this investigation, The Gathering of Uru, comprised between 450 and 160 players during the course of the study. They were a subset of two larger meta-groups, the Uru Diaspora comprising an estimated 10,000 players, and the inhabitants of There.com, whose exact number was unknown, but was likely to be in the hundreds of thousands.
Components vs. System. By definition, emergent phenomena transcend the life cycle of any one of the elements within the complex system. Therefore, the emergent phenomena studied had to demonstrate recognizable patterns across a diverse sampling of individual participants.
System vs. Environment. Emergent phenomena happen when a system comes into contact with a specific environment or “ecosystem.” Earlier, we described the game software as a “play ecosystem.” Emergent behavior arises out of tripartite interactions between a) the individual components of the system, b) the system as a whole, b) the ecosystem(s), or environment(s), that the system inhabits. (Bar-Yam 2000a)In the case of the Uru Diaspora, the individuals and the system actually traversed through different gaming environments, giving us a glimpse into how it adapted in an emergent fashion to each “play ecosystem.”
Relationship to Game’s Intent. As we’ve stated, we are looking for patterns of behavior that fall outside of the formal structure as intended by its designers, and which exhibit bottom-up process. This means larger patterns that occur as a result of individual agents in the system acting independently, or interdependently, but not through any central control mechanism.
Method. The study had to utilize a method that would enable a multi-scaled method that would allow for observation of the “forest and the trees at the same time,” in other words, it had to be possible to observe the three components, system, parts and ecosystem, concurrently.
A methodological conundrum confronts us at this point. What tools and methods shall we use to observe the emergent phenomena we have defined here? There are a number of different established methods in game studies. Quantitative methods, such as surveys, and in-game data-mining can provide us with very useful information. The former are excellent at understanding the scope of individual’s attitudes about their gameplay experience. They are excellent at getting at the larger patterns of behavior and attitudes displayed by individuals, but less effective at getting at larger patterns of interaction between individuals. Large-scale surveys help us to understand that people are spending an average of 20 hours a week in online games, but not specifically what they are doing, who they are spending time with, and specifically how they interact in social contexts with the ecosystem. Data mining, such as capturing chatlogs in a fixed location, is an excellent method for discourse analysis in specific contexts, although it does not give us the attitudinal data of surveys, nor does it measure larger cultural patterns across multiple locations.