A methodical approach to systemic games
This paper aims to present systems and systemic thinking as tools to improve the process of designing games. By comparing what systems are, how they work and how emergence can be implemented to the standards axioms of game design, the following conclusions can be drawn. First of all, systemic design in simulations can increase user's learning potential. Emergence in the Theory of Emergent Learning and in improvisational theatre seem to create a more personalised and unique experience where the user is empowered in regard to the outcomes of the activity. These consequences of systemic design and emergence are then analysed in the context of a game’s design to gauge their contribution to the overall play experience.
Finally, these conclusions were turned into axioms to form the Systemic Game Framework or SGD. A framework that suggests methods to conceptualise systems, balance their behaviours and implement them into games.
Chapter 1 - Systemic Thinking
Holistic thinking Systemic Thinking
Chapter 2 - Systems
Definition of a system Different types of systems
Chapter 3 - Systemic game design
Games as simulation Learning as a requirement Simulation as learning tools
Chapter 4 - Emergence in games
What is emergence Emergence use cases Emergence in games
Chapter 5 - User-centred game design
Learning process Player Engagement Player Agency
Chapter 6 - Proposed framework
Why a new framework? Systemic Game Design Framework Systems should be created following the same structure. Loops should be balanced between positive and negative loops. Systems should be balanced among each other.
Planning for emergence to allow for a more personalised experience.
Conclusion Figures & References
With all the innovations in game development, gaming platforms, online services or even new generations of players, the requirements for fun and accessible games have become increasingly more complex than the ones from 10 years ago. In consequence, the strain on developers has never been higher. One way to relieve those strains is to improve pre-production, and production workflows to alleviate as much as possible the hassles of developing and releasing a game. This is usually done by implementing what is commonly called systems, whether it is the structure of a company, the framework used on a specific project or even the systems that the project is made out of. Systems have always been a part of the designing process of any project and in this case, a game. However, in the early days of game development, there was no source control, game design theories, agile frameworks and all the other systems that are commonly used nowadays. As computers got more powerful, more and more complex systems were implemented in games. From a simple artificial intelligence to play against to emergent social environments, the potential for new, complex and unique experiences has exponentially increased over the past decades. In an effort to improve the design workflow of systems in games, this paper compares the core elements of both systems and game design to suggest a new approach to designing systems for games.
Chapter 1 and 2 establish a basis to understand systems and explores what a system is made out of as well as its variants. Chapter 3 and 4 presents the uses of systems and emergence in various activities including games. Chapter 5 explores the requirements for a fun and enjoyable game. Finally, chapter 6 aims to compile the previous chapters into a framework to design and implement systems in game design.
Chapter 1 - Systemic Thinking
Systemic thinking represents the theories and philosophies behind the creation and function of systems. The origin of systemic thinking can be traced back to the Classical Greece period (5th century BC) with the word system originating from the ancient Greek “sústēma”, meaning a whole made of several parts or members. However, most of the concepts discussed in this paper have emerged from Bertalanffy’s General System Theory or GST from 1950. Since then, it has been widely accepted that holism, considered the opposite of reductionism, is inherent to systemic thinking. They both suggest a method of analysing systems and are necessary to properly define systemic thinking.
Reductionism is the belief that any problem can be broken down (or “reduced") into smaller components interacting with each other. It strongly represents the idea of causality, the idea that one process can be the cause of another one. It is a process that comes naturally to most people in order to understand the world in a rational way. For these reasons, it has long been used as a method to analyse systems in science, mathematics and religions. For example, if we remove the wheels off a wheelbarrow, it would not move as well. Using the reductionist approach would result in the conclusion that the wheels are an essential component to the wheelbarrow’s efficiency.
However, this idea of causality imposes certain limits. It requires that any phenomenon is the result of the sum of its reduced components. In other words, the behaviour of the whole can be logically explained by only looking at its parts. It makes it impossible to analyse larger, and complex systems, which are known to have emergent properties that aren’t the sum of its component (a more in-depth definition is given in the next chapter, Systems).
René Descartes who fully adopted this view wrote Discourse on Method in 1637, in which he breaks down into steps the method he follows to observe and study a problem and/or phenomenon. The second most important step according to him is “to divide each problem I examined into as many parts as was feasible, and as was requisite for its better solution.” (Descartes, 1637). His and Isaac Newton’s work later resulted in the creation of a unified scientific method still used today.
Holism, as opposed to Reductionism, is the idea that systems should be viewed as wholes, not a simple group of smaller entities. They are still made out of parts; However, these cannot exist independently from the whole or at least cannot be understood without it. It tends to emphasise on studying a system by looking at its resulting behaviours rather than understanding what caused them. Here again, thinking in holistic terms can come naturally for most people. For example, when thinking about a dog, one usually does not think of the parts that makes it a dog (organs, bones, fur etc …), but rather of the dog’s personality, behaviours and any other phenomenon that results from the system's individual parts interacting with each other.
Holistic thinking supports the idea that the whole can be greater than the sum of its part, meaning that not all behaviours can be associated to a limited set of interactions between specific parts. Since the focus is not to understand the causality of a behaviours but rather to observe it, it becomes difficult to draw definitive answers. It does however make it easier to understand more complex systems, which was impossible with the reductionist approach.
The term holism is generally attributed to Jan Smuts, who pioneered the idea in his book Holism and Evolution published in 1926. In his writing he describes holism as “the tendency in nature to form wholes that are greater than the sum of the parts through creative evolution” (Smuts, 1973). This tendency is more commonly called emergency and is further defined later in this paper.
Both Reductionism and Holism enable the analysis of a system, but they can only be used on specific types of systems and aren’t compatible with each other. The idea behind systemic thinking is to have one unified method to analyse all kinds of systems and therefore shares a lot of similarities with both ideas.
Systemic thinking revolves around understanding how every part of a system interact with each other. Similarly, to reductionism, the parts must be defined in the first place, but the key aspect is to identify the “loops” of interaction instead of individual parts. A loop can be defined as a chain of interactions originating from one part that involves multiple parts that in turn will interact with the starting part again. For example, Part A affects Part B, Part B affect Part C which in turn affect Part A again.
As A. Laszlo and S. Krippner puts it, “The principal heuristic innovation of the systems approach is what may be called ‘reduction to dynamics’ as contrasted with ‘reduction to components’ …” (Laszlo and Krippner, 1997). Because this “reduction to dynamics” can be used on all kinds of system, it represents the main strength in systemic thinking.
In conclusion, three different methods to analyse a system or phenomena were defined.
Reductionism, which focuses on breaking down a system into individual parts.
Holism, which focuses on observing behaviour patterns of a system as a whole.
Systemic thinking, which focuses on defining the interactions between individual parts.
The notions system (a whole made of several parts) and complex system (systems that have emergent properties) were also introduced. The next chapter will go more in depth in the definition of these two types of systems alongside other ones as well as their main differences.
Chapter 2 - Systems
Systems can be identified pretty much anywhere and can take many different shapes or form. Whether its mechanical (a car and its parts), biological (flowers and pollinating bees) or even social (public services being the results of a ruling government), systems exist in many different forms in both natural and human made worlds. Systems are a very abstract concept, for that reason it is important to keep its definition relatively abstract too.
Definition of a system
Although systems can be vastly different and have very different behaviours, they can all be broken down following the same method, systemic thinking. They are defined as “A set of parts that together form loops of interaction between them to create a persistent “whole”. The whole has its own properties and behaviours belonging to the group but not to any single part within it.” (Sellers, 2017). There are a lot of way of breaking down a system to its components, however Mike Sellers’ definition exposes the three main components of a system in a simple manner: parts, interaction between parts or loops and wholes.
Parts, or objects are the smallest building blocks of a system. Each part is characterised by its own attributes or state and its behaviours. A part’s attributes are all the characteristics held by that object at any given time. In games, they represent game objects and their variables at a given timetfor. For example, a weapon might have damage, accuracy and fire rate values. These are considered to be the weapon’s attributes whether or not they are modified throughout the gameplay. Each part in a system has to do something, this is defined as the part’s behaviour. A weapon’s behaviour could be the actions of shooting, reloading and aiming down sight.
Loops as defined by Mike Sellers are all the possible interaction a part can have with other parts. These interactions emerge from each part’s behaviour working together in way that forms a loop. It is usually described using the following model: part A affects part B which affect C in turn affecting A again. Using the firearm’s example again, it might be able to shoot an enemy forming two loops. First, the player’s stock in ammo can be reduced to the point where the weapon cannot shoot anymore, we called this a negative or balancing loop as it limits the behaviour. Secondly, the enemy might be eliminated granting the player some experience for this effort in turn enhancing the player’s shooting ability, this loop is defined as a positive or reinforcing one as it might enhance the ability for this to happen again. This component helps differentiate two types of systems simple and complex. Loops' ability to speed up or down interactions between parts, allow those behaviours to change over time, we can then say that these modified behaviours emerge from the system. In that case, the system becomes complex and fosters emergence as opposed to a simple or complicated system where every interaction is linear, predictable and repeatable.
Finally, wholes are the resulting behaviours of interacting parts. They are a way of representing a group of interacting parts into one entity that has its own set of attributes and behaviours. Wholes can also interact with other systems in which case they would become parts of another bigger system. For example, a weapon attached to a player character forms a whole, and that character in an environment with enemies is another bigger whole with different behaviours.
This breakdown of a system does emphasise complex systems compared to other ones. Another method to look at systems is defined in Rules of Play (Salen and Zimmerman, 2003) where a system is broken down into four components: objects(parts), the object’s attributes, internal relationships (each part’s behaviour) and environment (which is the context that surrounds the system). The addition of the environment component and the exclusion of the loop component make this definition more neutral from a system’s point of view.
Different types of systems
As systems can be vastly different, there was a need to categorise them. Simple, complicated and complex systems form three of the four different type of systems. Welsh management consultant Dave Snowden created a sensemaking framework to aid decision-making, the Cynefin framework (Snowden, 2010). Although its primary use isn’t related to this study, it is heavily based on systems theory and its layout is very useful to the understanding of the various types of systems.
The Cynefin framework has 5 components, each describing a type of problem one would have, and for each component, the steps to take to overcome that problem. The idea is that the framework would give an insight on the steps to take to solve a problem depending on its type. This is interesting because each problem is described as a system, and the suggested approach can both help us as game designer to design systems, but also the player on how to approach a specific system.
Figure 1: Cynefin Framework, Dave Snowden (CC BY 3.0)
The first component is the domain of the “Obvious” or simple systems. It represents a system that is stable and answers to clear rules. The relationship between cause and effect is clear making it compatible with the reductionist approach. The suggested steps to approach this type of system are “sense”, “categorise” and respond following the rules. In other words, it is just a matter of understanding and then acting. An example of this type of system could a game’s settings interface, let’s say a user wants to change the audio volume, it is obvious that modifying the volume slider will do just that.
The second component represent “Complicated” systems. In these types of systems, the relationship between cause and effect requires some analysis or expertise (previous knowledge). The reductionist approach still supports this type of system. A complicated system can be dealt with by assessing the facts, analysing and then applying the appropriate behaviour. Here again, an example of such a system could be a game’s user interface. The first time around the user might be confused, not knowing its layout or how it works, but by analysing its changes and responding accordingly, the player will in turn learn how to use it.
The third component represent a “Complex” system. In complex systems, cause and effect can only be deduced in retrospect and there is no absolute right answer. As a side note, this is due to the nature of complex system being emergent, there are too many varying possibilities to deduce one single answer. This is where the reductionist approach has its limit and the idea of holism makes more sense. It is dealt with by probing, sensing and then responding. As an example, Conway’s Game of Life (Gardner, 1970) is a grid base simulation where each cell can either be dead or alive. There are three rules stating when a cell dies, when one is born and when one survives. This system's outcome is totally dependent on where the starting cells are placed, and it would be very difficult to imagine in what state the simulation looks at any given time for any given starting cell. However it turns out that by arranging living cells a certain way, various patterns of surviving and ever-changing cells emerge. By keeping track of these behaviours, we are keeping track of the potential outcomes the system has, but no one outcome is the right one.
The fourth component is a “Chaotic” system with which, cause, and effect are totally unclear. Events in this domain are “too confusing to wait for a knowledge-based answer” (Lambe, 2007) therefore the process of dealing with it, is act, sense and then respond. In other words, the only way to progress is to act and then react. Here again, the holistic approach is preferred to reductionism. Chaotic systems are less common in games due to their nature to cause uncontrollable randomness; However, one could argue that a multiplayer game opposing groups of players could represent a chaotic system. “For example, the game environment may be oscillatory or even chaotic, and the evolution of strategies in these circumstances will be very different” -(Akiyama and Kaneko, 2000).
Finally, the last component represents “Disorder”, or basically any system that do not belong the four others. D.Snowden defines disorder as the state someone is in before using the framework.
The four types of system can be further categorised; Simple and complicated systems as ordered and on the other hand complex and chaotic as unordered. This differentiation is important because we can’t apply the same thinking process to both. Ordered systems can easily be broken down to parts following a Reductionists thinking. Unordered systems are however, as T. Stewart puts it “impervious to a reductionist, take-it-apart-and-see-how-it-works approach, because your very actions change the situation in unpredictable ways." (Stewart, 2019). We should therefore have a holistic approach to understand the latter type of systems. We we have established here, is that what makes a system type different from another is the nature of its parts’ interactions. A system could have a thousand parts, it doesn’t mean it will have a complex behaviour.
In conclusion, systems have been defined as an entity made out of three components:
Parts, the smallest building block of a system, their attributes and their behaviour.
Loops (positive and negative), the interactions between different parts of a system.
Whole, the behaviour resulting from the parts’ interactions.
With that in mind, four types of systems were identified via the Cynefin framework:
Simple systems, in which cause, and effect relations are obvious.
Complicated systems, where cause and effect relations require previous experience.
Complex systems, where cause and effect relations can be predicted but not exact.
Chaotic systems, in which cause and effect relations is unpredictable or even random.
The following chapter will further explain how these different types of system are implemented in games, and what consequences they have on the playing experience.
Chapter 3 - Systemic game design
It has been established that systems could take many forms, and that is also true when it comes to games. Games, regardless of the genre, have a panoply of systems that enables the users to interact and enjoy it. We can differentiate two main types of systems that a game can have, software systems and gameplay systems.
Software systems are the technical systems that enable the user experience making sure the game runs properly. Video and audio output interfaces, keyboard or controller input interfaces, the operating system the game's settings are all software systems. These systems have very little if no direct influence over gameplay, they simply enable a game to be functional. They are usually designed by software engineers or programmers.
Gameplay systems on the other hand are the ones responsible of the player’s experience, they are the game's mechanics. They include systems such as a movement system, combat systems but also economy systems and and any other systems that directly affects the gameplay. They are often called the “nouns” and “verbs” of a game. They are typically the systems that a game designer would be working on.
In the context of systemic game design, gameplay systems are the ones that are being referred to. They are the systems this paper intends to improve, therefore for the rest of this study the terms systems and gameplay systems can be interchanged.
Games as simulation
Certain genres of games make extensive uses of systems. Most notably simulations games, which almost entirely rely on the player interacting with individual parts in order for anything interesting to happen. They are usually considered to be toned down versions of real simulations that would be used in education and research. This chapter aims to explore how simulation systems can be used games to improve their design. First, definitions of a simulation and game should be established and compared.
A simulation is generally understood to be an imitation of a reality (or world). In games, this translates to imitating key elements of a reality in the way that the player can recognise these elements from that said reality. Ellington, Addinall and Percival defined a simulation as “an operating representation of central features of reality.” (Ellington, Addinall and Percival, 1982). This definition is a very interesting and accurate one, from which the authors identify two main components. The first one is that a simulation must represent an actual situation that contains central features of reality. One that regardless of its origins, whether it was imagined or inspired from the past, needs to be anchored in "our" reality in some ways. That will enable the user to understand the situation and apply previous experience from said reality to the simulation. The second component is the statement that simulations aren’t any type of representation, it must be an operating one. The authors describe it as an “on-going process” (Ellington, Addinall and Percival, 1982). This essentially means that simulations cannot be static, they should involve some sort of activity with a theoretical start and end. In other words, photographs, maps and diagrams are not classed as simulations in their most simple form.
Furthermore, W.Robinett when writing about the creation process of a simulation states that “Given a phenomena to simulate, the problem is to decide what are its parts, how these parts can be represented with numerical values, and what the relationships are that let these parts affect one another” (Robinet, 1983). This is interesting in multiple ways, first is that simulations are made up of systems. Just like how you would create parts interacting with each to make a system, in a simulation the idea is to implement individual elements and the interactions that make them part of a system. Secondly, it confirms the idea of a simulation being an abstraction of reality. A conversion from a real phenomenon to simulated numerical values, from "our" reality, to "a" reality.
Finally, given this definition of a simulation, we can argue that all games are simulations. K. Salen and E. Zimmerman state that “Chess and Tic Tac Toe for example can be framed as “representations of territorial conflict in which simulated units war for control of a stylised battlefield.” (Salen and Zimmerman, 2003). In a way, Chess and Tic Tac Toe are simulations, be it, simple ones but still replicate behaviours we would find in real wars.
However, as a closing note, we established that all games can be considered a simulation, that doesn’t mean that all simulation can be considered a game. To turn a simulation into game we need to add the formal structural elements of a game: rules, goals, competitions and most importantly fun. As M. Prensky puts it “The worst simulation games are merely a set of learning points with the simulation part designed only as a sneaky way to get the player to each of them. The best keep pulling you to continue to the end in spite of yourself.” (Prensky, 2001). In other words, a bad simulation is closer to a learning experience than to a fun one. Which brings us to our next topic, learning in games.
Learning as a consequence
Games as much as any other activity represent an experience for the person involved in it. And as for any activity, one usually uses its past knowledge and set of skills to perform said activity. This learning process is often necessary for the players to even understand a game. It involve learning the rules, controls and generally how to play the game. As Hostetter puts it “Game players learn cognitive skills through problem-solving strategies such as observation, hypothesis, and trial and error by trying to figure out the rules of the game.” (Hostetter, 2006). In other words, by playing a game, the user assimilates cognitive skills that can be applied outside of this context. P. Greenfield explains in Mind and Media (Greenfield, 1984) that even for a game as simple as Pac-Man there are a lot of skills involved. She mentions ones such as eye-hand coordination, fast inductive skills and parallel processing. However, games teach players some skills doesn't mean they're a good way to do so. And as a matter of fact it is quite difficult to quantify learning objectively. Despite this we can easily imagine how engaging learning is on a game compared to reading a book.
In that regard, we can consider games as a learning tool but also, and maybe more so, as a practice tool. They not only facilitate the learning of skills, but it also inherently eases the practice process thanks to the level of control we have over them. But as we will see in the last part of this chapter, learning in games is not limited to abstract cognitive and motor skills. Games and simulations combined can be designed to teach specific topics and train to perform certain activity in a totally virtual environment.
Simulation as learning tools
In a previous chapter, we established that one of the requirements of a simulation is that it must be anchored in a reality whether it is fictional or not. This not only enables the player to use his past experiences to better understand what challenges they’re facing but also allows them to learn new things applicable outside this simulated reality. One theory to better understand how a player might acquire knowledge through simulations is Piaget’s theory of Constructivist learning.
The constructivist learning approach proposes that people create their own understanding of what is real as opposed to passively receiving information to digest. It assumes that after each and any new experience, a person tries to understand what happened, abstracts it and memorises it. We tend to refer to these abstractions as mental models. If a similar event happens in the future, that person will either use a multitude of mental models to try and understand this particular event and potentially improve them or create new ones. This subject is however discussed in more details in a later chapter.
Given that a simulation is heavily anchored in a reality, we could say that simulations could help the creation of mental models usable in said reality. Users could theoretically use their mental model from this reality and those from the simulation interchangeably. This does depend on how close to reality those simulated systems are. The closer the simulation’s systems are to the one found in that reality, the easier it will be for them to link the simulated phenomena to the real one. In other words, it enables them to deal with real problems as if they were happening in an authentic situation with no real consequences. As J.P. Gee puts it, “If learners are put into a situation that feels like the real thing, but with risks and dangers greatly mitigated, they can learn well and still feel a sense of authenticity and accomplishment.” (Gee, 2005).
In the case that the phenomenon has never been encountered in the past, the simulation format has a big advantage. It enables the users to try again by simply restarting the simulation whenever something doesn’t go as expected. This allows them to progressively get better, and potentially become better than other learners who did not use simulations. J.P. Gee (Gee, 2005) refers to this as the “Cycle of Expertise” when talking about how expertise is formed through repeated cycles of practicing skills until they become automatic. By doing so, users continuously improve their mental models of that specific event until they fully understand or until the limits of the simulation are reached.
To conclude, in order to narrow down the types of systems to study, two game systems have been identified:
Software systems, systems that enables gameplay systems to be interacted with.
Gameplay systems, systems that combined form the gameplay experience.
It has then been established that simulations inherently encourage learning and that learning is an essential part of games in general. Games can teach players some cognitive and motor skills but combined with simulations can teach specific topics by creating a realistic environment where mental models can be challenged and improved.
Chapter 4 - Emergence in games
We previously identified one type of system to have the potential for emergence, complex systems. We more precisely pinned down this feature to loops, a chain reaction of part interactions that will in turn affect itself. Overtime, these interactions can change in nature, resulting to an unpredictable change of behaviour, this phenomenon is referred as emergence. To better understand how loops work and are designed, the term emergence needs to be properly defined.
What is emergence
The concept of emergence refers to the properties or behaviours of a system that aren’t predictable by observing the system’s parts individually. As stated earlier, it is idea that the whole can be greater than the sum of its part. For example, let’s imagine a school of fish. Each individual fish behaves similarly and independently, they try to avoid collision with each other, and move in the general direction of their nearest neighbours. These simple rules lead to what looks like a perfectly coordinated group of fish swimming together. An important aspect here, is that no one fish leads the others, each fish is independent, and has the same impact on the group as the others.
This last concept is important, each part of an emergent system should be independent from the others, and its behaviour context-dependant. These two conditions lead to local properties (the part’s) forming unpredictable global properties (relating to the system as whole). Using the previous example again, if a predator is approaching the school of fish. Few fish will see the predator before the others and will change direction in an effort to avoid the danger. This will cause each of the fish’s neighbours to follow them, in turn creating multiple smaller groups rather than one big one. This shows how individual parts, behaving independently can lead to a global and unpredictable change to the whole.
K. Salen’s and E. Zimmerman’s definition of emergence sums up pretty well these concepts: “Emergence is above all a product of coupled, context-dependent interactions. Technically these interactions, and the resulting system, are nonlinear: The behaviour of the overall system cannot be obtained by summing the behaviours of its constituent parts.” (Salen and Zimmerman, 2003)
Emergence use cases
Just like systems, emergent behaviours can take many forms and be applied and different contexts. Before exploring the use of emergence in games, let’s have a look at other use cases outside of this context. This should allow us to identify some similarities in the benefits of emergence between each context.
Firstly, emergence is often at the centre of alternative pedagogies. E. Kays and R. Sims suggest a new approach to teaching using emergence, the Theory of Emergent Learning. They believe learning should be a learner-centred process rather that structured around the teacher. In order to achieve this, they emphasise on creating an environment with a set of simple rules in which both the environment and the student establish complexity by their individual interactions. This means having a “a bottom-up approach, where the complexity, creativity and flexibility of the human is given an opportunity to flourish and for knowledge and learning to consequently emerge. “ (Kays and Sims, 2006).
In other words, they want to design the learning process as a complex system in which learning is one of the emergent results. In practice, it involves giving a group of students the chance to all work together on the same learning exercise. Each student having their own set of knowledge, learning can happen on different levels, whether its students gathering knowledge from other students and passing it through or students learning together by figuring things out together, as a group. In any case, it is important to note that in this scenario, the students are at the centre of the learning experience.
Secondly, emergence theory is also used in improvisational theatre where the goal is for an interesting story to emerge out of two improvising actors. In order to achieve that, the actor generally have a set of rules to follow. Iwo Swartjes outlines fours main principle required to have an interesting story that emerge out of two improvising actors (Swartjes and Vromen, 2007). Each principle dictates how the actors should behave. In a way, we are making the actors be parts of a system that follows simple rules. Together, all the actors interacting with each other form the whole resulting in an emerging story.
An interesting effect of improvisational theatre is the role of the audience. Spectators often play a very important part; their reactions are the only feedback actors have on their performance. In a way they are the context in which the improvised story emerges. In other words, the story is unique to that event and to that specific audience. Moreover, spectators can easily be integrated as part of the scene. Whether it is from a direct question to the audience, or from a random event (such as a phone ringing, someone coughing etc…). The actors can easily use these events and implement it in the improvised scene.
In both use cases, the concept of emergence seems to bring more than just a new alternative to standard methods. It is used to create an experience that is more personalised, active and social to the users. In the Theory of Emergent Learning, everything relies on the learner’s ability to learn at their pace from other students, enabling them to be more creative and flexible than if they were to learn from a teacher. Moreover, in improvisational theatre, emergence seems to create a uniqueness to the experience and empowers the spectator. One downside these implementations of emergence have is the lack of control over the result. We can foster for emergence in learning and in story making, but we cannot predict its efficiency, quality or even its usability.
Emergence in games
Emergence in games, and more specifically in gameplay systems can be fostered in a lot of different ways. Penny Sweetser in her book Emergence in Games suggests various ways this can be achieved:“The key to creating emergent gameplay is to define a simple, general set of elements and rules that can give rise to a wide variety of interesting, challenging behaviours and interactions in varying situations.“(Sweetser, 2008). In that context, the simpler the rules, the easier they will be understood by the players. The rules are the only element that will stay constant throughout an emergent experience. Every part, challenge and interaction should be able to change depending on the context. She more importantly identifies four areas that hold potential for developing emergence in games: games worlds, agents, narrative, social systems.
The game world represents the environment players are playing in, and the objects they can interact with. The environment on its own is inert and does not allow any interactions with the player, all the interactivity of an environment originates from its objects. Usually games will have multiple environment populated with objects that have different interactions resulting in vastly different levels. Some objects will only interact with the player, others with the environment. Therefore, by smartly positioning these objects and anticipating the player’s actions, not only the level looks different, but will play differently. This enables players to come up with solutions and strategies of their own in order to progress. For example, an explosive barrel is an object that can be used in various different ways in a level. Placed next to enemies, its radius of explosion can clear multiple with very little cost for the player. An explosive barrel could also be placed next to a destructible wall, unveiling a path to a secret area.
Agents are a type of object that populate environments. They are different to other objects in that they are dynamic, and do not require any interaction from the player to function. More specifically, agents are able to sense and react to the environment and its objects. That means that not only agents can change the state of the world, they can also dynamically react to the player’s actions. They are a crucial component in creating emergence as they add a new level of complexity to the game world. Agents include any type of playable and non-playable character whether they are governed by an AI system or indirectly by the players.
The narrative is the story that gives a context to the player’s actions. There are two distinct types of narratives, linear, the player gets told a story and has little impact on its delivery and non-linear, where the player can actively change the course of the story. Non-linear narratives can even be divided in two sub-categories, branching, and emergent narratives. The first category, branching stories, contains “blocks” of narrative that can be interchanged with each other depending on the player’s actions. The second category, emergent narratives, doesn’t have any “blocks”, and the story isn’t pre-planned. Instead, the player, by interacting with the game and the environment becomes the creator of the story. In principle, anything can happen as long as the game’s mechanics allow it. For example, games such as The Sims or Sim City have emergent narratives that the players create as they play.
Social emergence in games relates to any unintended behaviour observed in multiplayer games. Usually noticed in massively multiplayer games involving more than a dozen of players, it goes hands in hands with virtual economies that change based on the supply and demand of the players. Most games will see their communities form complex social structures thanks to social mechanics such as chats, guilds, forums, and competitive ladders. It is a very difficult aspect of games to design with set ideas in mind as it would defeat purpose of emergence. Designers tend to implement social tools for player to exchange and communicate but also create. Some games will even allow players to directly modify the mechanics via mods and custom game modes.
As a closing note, looking back at the two first use cases of emergence, we can see they share a lot of similarities with emergence in games. The Theory of Emergent Learning can be considered as social emergence, and improvisational theatre as narrative emergence. In both cases, emergence seems to empower the users, learners or players by giving them some control over the experience. However, as stated earlier, in those use cases, emergence seemed to lack control over the results. This is somewhat compensated in games by the fact that designers have way more control of the environment around the emergent system. Things can be corrected without the player noticing it and limits can be imposed to avoid absurd results. Overall, games seem to be a very good platform to create emergent experiences.
Chapter 5 - Game centered design
Throughout the previous chapters, it has been suggested that systemic design and emergence seems to foster certain types of experiences for the users. It most notably seems to encourage the learning process, engagement and overall user agency. These three consequences of systemic design are also well researched game design components. To further understand how exactly these are enhanced, a better explanation of each should be given. Moreover, these three processes also correspond to the three components of well-being as defined by Schueller and Seligman in the 2010 Journal of Positive Psychology, pleasure, engagement and meaning. Including these will also allow for a better understanding of what makes a game fun and enjoyable.
When making games, designers usually want to create new unique experiences that no other game can offer. But in order to achieve that, it is important to understand how a player reacts to these new experiences. And for any players, playing a new game usually involve a learning curve. A process to bring the player to a level of skill where he can understand and enjoy the game without external help. This process usually involves two types of learning, knowledge based learning and experiential learning.
Knowledge based learning is the process of learning from facts and unformatted information. Benjamin Bloom published in 1956 “Taxonomy of Educational Objectives: The Classification of Educational Goals”, a taxonomy that goes in depth to explain how knowledge based learning happens. It is still today a standard reference in terms of educational teaching and skill evaluation. In games, this type of learning happens through instruction manuals, informative tutorials and any learning that happens outside the game. This type of learning usually happens at the early stages of play, when the player knows very little about the game and has to assimilate the basics.
Experiential learning on the other hand, happens when a player uses previous experiences to better understand future ones. Contrary to knowledge based learning, it is a process that cannot be “forced” but rather induced. David Kolb and his 1984 publication of the Experiential Learning Theory shows how non linear this process can be. It is a process that requires more involvement from the learner but can results in greater skills and faster. In games, it translates to any learned skills or piece of knowledge that the player isn’t obviously made aware of by the game. This includes skills that aren’t specific to a game, such as cognitive and motor skills.
Although knowledge based learning is designed to be in pretty much any game, it usually represent a small portion of the learning curve compared to experiential learning. Games tend to be designed to have an easy hands on experience, but require some skills and attention to be mastered. As Nolan Bushnell puts it: “All the best games are easy to learn and difficult to master. They should reward the first quarter and the hundredth.” (Bushnell’s Law, Nolan Bushnell).
In a previous chapter, the notion of Constructivism was introduced in the context of a simulation as a learning tool. It proposes that people create their own mental models from any new experience, and can apply them to future ones to better understand them. It is however not limited to simulations, players do carry over their knowledge of previously played games to another. In this context, mental models include the player’s understanding of all perceivable aspect of a game such as in-game variables (health, inventory etc ...), prognostic on their performance (via the feedback provided by the game), the goals within the game and their predictions of what will happen based on their past actions. Given how complex a mental model can be, it becomes obvious why knowledge based learning is not the best design option to bring a player to a level mastery, its too difficult to implement.
Although knowledge based learning is very important at the early stages, in an effort to analyse the learning process throughout the entire play experience, only experiential learning will be studied in details here. The average player will acquire more skills and knowledge from his play experience than from a tutorial or an instruction manual which are usually used when the player is missing very specific elements of the game. In that regards, we’ll used D.Kolb Experiential Learning Cycle (Kolb, 1984) as basis to study learning in games. It is a 4 step process with no start or end, meaning the learner can enter the process at any step.
Figure 2: David Kolb, Experiential Learning experiences as a source of learning and development.
Concrete experience consists in the learner experiencing a new situation or a new instance of an already known type of situation. For experiential learning to happen the experience needs to bring new elements, unknown to the learner.
Reflective observation is when the learner then reflects on what he experienced. Similarly, to the Comprehending step of Bloom’s taxonomy, it consists on organising and comparing the memories.
Abstract conceptualisation happens when the learner makes sense of what happened during the experience, an interprets the events to draw some theories. The theories the learner draws are the core element of the knowledge acquired through experiential learning.
Finally, active experimentation is when the learner puts into practice the knowledge he acquired in the previous steps.
The step of experimenting with the acquired knowledge is key in the learning process. By experimenting, the player is testing his mental model about the game, his understanding of the mechanics. During that step, it is important for the game to be able to provide feedback to the player whether the player is right or not. Whether it is via sound cues, visual effects or any other way, the games needs to establish a language to talk back to the player. If the player doesn’t receive feedback, he will never be able to understand the game. Now say a player does something right, and knows it from the game’s feedback, they’ll feel a sense of accomplishment, whatever actions they were experimenting is now validated and added to their mental model. In other words, “The mental model therefore arises out of a combination of the player’s emotions attention plans and goals.” (Sellers, 2017).
Learning represents the first component of well-being, pleasure. Without learning, a game becomes monotone, and promotes grinding instead of progress. This is why the mental model formed during the play is one of the most important element of the experience. Every other emotions will arises from it. It is therefore necessary for a game to be thought of as learning tool, a tool from which making a mental model of is facilitated. “The more easily this model is to build within the player’s mind, and the more consistent the player’s understanding is of the world model defined by the game designer, the more engaging the game will be.” (Sellers, 2017).
Player engagement can be defined as a player’s willingness to invest his time and cognitive skills into a game. It can also be seen as the player’s motivation, though some argue motivation represents the reasons why people begin to play, where engagement would relate to what makes the player continue playing. Although it has been previously established that engagement originates from learning, it is only a part that makes up for a good play experience. The player also need to be able to challenge his knowledge and skills. If we take the example of a step by step tutorial, players will more often that not get bored really easily, since its sole purpose is to block the player from experimenting so that he can learn in a limited environment with no ambiguities.
In that regard, most designer will try to create a balance between learning and challenging to put the player in a state of “Flow”. Mihaly Csikszentmihalyi in “Flow: The Psychology of Optimal Experience.” (Csikszentmihalyi, 2016) defines flow as being a state of operation in which a person is fully immersed in a feeling of focus, involvement and enjoyment. The theory suggests that any activity can enable the state of flow to be reach as long as the 8 axioms of flow are respected. All 8 axioms proposed by Csikszentmihalyi describe how an activity should be to enable flow. He sums up his theory in a graph where skills and challenge are the two axis. The graph shows 3 zones, anxiety when the task is too hard, boredom when not engaging enough and the flow chanel, which is according to him the place to be for an optimal experience.
Figure 3: Flow Channel Wave, by Jesse Schell (From “The Art of Game Design”).
This theory of flow is however not specific to games, it is actually applicable to pretty much any activity. In an effort to have a model a bit more related to games Penelope Sweetser and Peta Wyeth published GameFlow: A model for evaluating player enjoyment in games in 2005. It is a model based on Csikszentmihalyi’s in which the 8 components of flow have been turned into criterias to gauge player enjoyment in games. The following are the 8 components of gameflow as stated in P.Sweetser and P.Wyeth’s publication:
Concentration: Games should require concentration and the player should be able to focus on the game.
Challenge: Games should be challenging and match the player’s skill level. Player skills: Games must support the player’s skill development and mastery.
Control: Player should feel a sense of control over their actions in the game.
Clear Goals: Games should provide the player with clear goals at appropriate times.
Feedback: Players must receive appropriate feedback at appropriate times.
Immersion: Players should experience deep but effortless involvement in the game.
Social Interaction: Games should support and create opportunities for social interactions.
As pointed out by J.P.Gee “Good games adjust challenges and give feedback in such a way that different players feel the game is challenging but doable and that their effort is paying off.” (Gee, 2005). Although considered to be a state void of emotion, flow enables players to perform at their best on a prolonged period of time which in turn will be emotionally rewarding. In that regards, the player engagement and enjoyment represent the second component of well-being, engagement.
Finally, player agency is the player’s ability to interact meaningfully with the game’s environment, and make sensible decisions based on the situations he is in. This ability isn’t fully dependent on the individual, it also refers to how well the game introduces the player to those situations. Csikszentmihalyi when discussing the challenges of an activity says “It is not only the “real” challenges presented by the situation that count, but those that the person is aware of. It is not skills we actually have that determine how we feel, but the ones we think we have. “ (Csikszentmihalyi, 2016). During a play experience, the goals the player is working towards, are either set by the player or by the game.
A goal set by the game, or by someone other than the player is said to have extrinsic value to the player. This means the player doesn’t fully know what the actions taken to reach that goal are before doing them. This usually happens when a player is learning how to play via a tutorial. The game sets small goals, with few actions to enable the player to form a mental model. However, this is rarely pleasurable for the player, and too much of it can turn the game into a “grinding” experience.
On the other hand, a goal set by the player, and the actions taken to achieve that goal have intrinsic value to the player, meaning the player is playing for themselves. These autotelic goals, ones set by the player himself, are the proof that the player has a mental model sufficiently developed to be challenged and therefore takes meaningful decision on his gameplay. In many instances, games that give the player more freedom of movement will make the player have an autotelic experience, on the condition that his mental model is advanced enough to support it.
Player agency represent the third and last component of well-being, meaning. Players that are able to make decision based on their knowledge, challenge their mental models, and potentially learn new things on the fly are more susceptible to enjoy themself than players that simply follow strict rules set by the game. Agency empowers players, and make them feel they have control over their experience.
From the previous chapter, it has been suggested that systemic design and emergence encourage three particular processes that are very important in a game’s design.
Learning: Gives the player the means to understand, and engage with the game.
Engaging: Gives the player the means to control and enjoy the game.
Agency: Enables the player to reach the level of mastery
These aspect of game design are also homologous to the three component of well-being as defined by Schueller and Seligman. This relation proves how important the design of these components are to the overall experience.
Chapter 6 - Proposed framework
Frameworks allow designers to conceptualise new experiences by following a structure or certain base rules. There are multiple game related frameworks out there, but very few are focused on systemic games. Among the most popular, the MDA Framework by Hunicke, Leblanc and Zubek (2004) is probably the reference in terms of breaking down any game’s design into individual components.
Why a new framework?
There are currently no framework to specifically design games based on systems. The existing frameworks tend to be very abstract and aim to be applicable to as many games as possible. The aforementioned MDA framework is usable on pretty any game possible, making a very accessible framework but the fact that it is so abstract makes it open to interpretation for many designers. This means that a lot of other frameworks are based off the same source, ones like the 6-11 framework (Dillon, 2010) and the SSM framework (Grip, 2017) are two of them.The goal here is to create a framework specifically aimed at systemic games, which can be used directly by designers to analyse their systems and balance their gameplay.
Systemic Game Design Framework
The previous chapter have all been establishing specific requirements of systemic game design. These can be reduced to 4 axioms to guide designers in their implementation of systems in their games.
Systems should be created following the same structure.
Loops should be balanced between positive and negative loops.
Systems should be balanced among each other.
Planning for emergence to allow for a more personalised experience.
Figure 4: Systemic Game Design Framework
Systems should be created following the same structure
By creating each gameplay system following the same rules, systems become easier to abstract and understand for the player. Using the MDA framework and Mike Seller’s definition of a system we can easily break down every systems of a game to 3 core components.
Parts (Mechanics) are the base components of the game. They are entities characterised by a set of attributes and behaviours. The parts are what dictate the game’s rule and the actions the player. They can take under the form of data structures and algorithm.
Loops (Dynamics) represent the behaviours emerging from the interactions between the parts and the player. These behaviours can also emerge from parts interacting with other parts.
Wholes (Aesthetics) represent the system’s purpose to the experience as a whole. It refers to the reasons why the player would use such a system, and how it is interacting with other systems.
Figure 5: Systemic Game Design Framework - System Conception
Loops should be balanced between positive and negative loops
Once modelled, a system’s loops should be balanced from another point of view. Every loops can be considered either positive or negative.
Positive or reinforcing loops enhance or amplify changes in a system. They can be used to both make a player progress or regress. As long as it involves changing the system’s state making it unstable.
Negative or balancing loops tend to dampen or buffer changes in a system. They are usually used to slow down the player’s progression or regression and promotes stability over change.
By sorting every system’s dynamics in a cross matrix, every behaviour can identified and balanced. The cross matrix should be read from left to right, meaning for each row, we assume it is the row’s system interacting on the column’s. This allows for the identification of unidirectional interactions, when the interaction from system A to B is different than the one from B to A. It is also possible to have multiple interactions between two system depending its environment. A cross matrix should be made for each case to highlights every system’s behaviours in every possible environment.
Figure 7: Systemic Game Design Framework - Loop Balancing
Systems should be balanced among each other
Each system can then be categorised in the Cynefin framework depending on how it behaves. Using GameFlow on each category, the systems they represent can be further defined. The elements of GameFlow not only serves as a way to categorise systems but also as guidelines to design them. Each element has been given an importance based on the type of system it relates to. It is important to note here that social interaction has been removed as it will be relies more on the game as a whole rather than its individual systems.
Simple systems tend to be related to player engaging with the game. They represent the simplest form of interaction between the player and the game where the relation between cause and effect is obvious. Simple systems are in all games in some shape or form, and are necessary to enable the game to be played. These systems should not be a challenge for the players to use and should not require any advanced skills. In addition, these systems should give the players full control of the outcome by presenting them with clear goals and feedback for their actions. Finally, immersion is not a requirement for this type of system.
Complicated systems tend to be the second most abundant type of systems in games. The relationship between cause and effect requires analysis making them the systems responsible for the player learning. These systems should not be challenging at all to the player, and should offer sufficient control for him to focus on assimilating. They should support the players’ skills development as much as possible by giving them feedback when possible. Clear goals here should be prominent compared to immersion since the aim is for the player to learn.
Complex systems tend to represent the majority of systems in most games and should be balanced with complicated systems to enable “Flow”. They are responsible for challenging the player. It is important to note here, that in the context of this framework what is considered a complex system does not inherently have emergent properties. This allow for a simpler comparison between challenging systems and and learning ones. These systems should challenge the player and enable them to further develop their skills by giving unclear goals, less control over the outcome and less feedback. These system should allow players to fully immerse themselves and express their play style.
Chaotic system usually give the player a level of agency other systems of the game don’t allow. These systems will for example allow the player to challenge the game by allowing the game to be played in different ways than intended by the designers. For these reasons, they are not implemented in every game and can be considered as optional. These system usually requires the players to have an extensive knowledge of the game, although not initially designed to challenge the player. Since it is being used for other reasons than the intended ones, control, goals and feedback are not the priority either. These will be part of the player’s autotelic experience.
Figure 6: SGD Framework - System Balancing
Planning for emergence to allow for a more personalised experience
Emergence tends to be a aspect of game difficult to design and implement. It requires loops’ behaviours to have consequences beyond the boundaries of their respective systems. These dynamics should also be context-dependent meaning the overall system cannot be obtained by summing the behaviours the behaviours of its constituent parts, its environment should also be taken in account. There are four areas in which emergence tends to be implemented: Game Worlds, Agents, Narrative and Social Systems.
Games worlds allow for the easiest implementation of emergence. It involves creating systems that associate the inert environment with the objects placed in it, in turn making it more dynamic. It is important to make sure that for each challenge presented to the player, there are multiple solutions. These dynamics can be interchangeable, combinable or concurrent and rival (in which case the player can usually only use one or the other).
Agents are a way to further implement emergence in game worlds. Agents are dynamic game worlds objects too but they do not require any interaction from the player to function. This means these agents are able to change the state of the game world and dynamically react to the player’s action. This allow for the game worlds to be ever changing independently from the player’s action.
Emergent narratives allow for the player to alter the course of a story through his action. There are two main types of emergent narratives, non-linear and emergent. Non-linear narrative involves creating “blocks” of the story that can be interchange. It tends to require more attention than other types of narrative for it still heavily relies on scripting parts of the game. On the other hand, emergent narratives are more “modular” and involve giving the player the tools to build a story of his own. In this case, any phenomena that the game’s mechanics allow becomes a “block” of the story which combined will result in a player made narrative.
Finally, Social System relies on giving the players some tools to interact with each other. The more potential interactions in a game, the more it has the potential for social emergence. These interactions can be materialised by simple communication and exchange tools to ones that allow the players to create and modify the game itself.