The concept of "serious games" in academic and research discourse has evolved significantly since its origination in Clark C. Abt's seminal work, "Serious Games," yet the current academic narrative may not fully encapsulate the transformative potential of games in educational settings. I am presenting an argument for a radical reimagination of the term "serious games" within academia and research to align more closely with the intrinsic qualities of gameplay that foster deep learning and skill development. The traditional focus on content and information acquisition as primary learning outcomes of game-based learning (GBL) is increasingly misaligned with the demands of the 21st century, where dispositional traits and complex skill sets are paramount.
The richness of gaming, as experienced by gamers who are deeply versed in game mechanics, dynamics, and game thinking, offers a more nuanced and potent framework for learning than what is often captured in current GBL research. My experience has made it clear that individuals who straddle the worlds of gaming, learning science research, and teaching bring invaluable perspectives that can rejuvenate the discourse on serious games. By shifting the focus towards leveraging games for developing critical 21st-century skills such as adaptability, critical thinking, collaboration, and creativity, which are inherently nurtured through natural gameplay, we can start to make GBL research more meaningful.
Re-evaluating the outcomes traditionally measured in GBL research means transcending the conventional emphasis on rote learning and content acquisition (what “is”), and instead focusing on skills and dispositional traits (what “could be”). The advent of individualized learning through AI coaches presents an opportunity to offload content-based instruction, thereby freeing up educational spaces for more targeted game-based interventions focused on skill and character development.
In advocating for this paradigm shift, I am calling for a collaborative effort among educators, researchers, and game designers to forge a new path in game-based education research. This path must honor the complexity and depth of games as tools for learning and recognize the diverse capabilities they can develop in learners. This approach aims not only to elevate the academic discourse on serious games but also to harness their full potential in preparing learners for the challenges and opportunities of the 21st century.
From Nodes to Knowledge: Mapping Game Dynamics with Eigenvector Centrality
In the intricate dance of game-based learning, where every choice and action contributes to educational outcomes, understanding the relational dynamics between these elements becomes paramount. Enter the concept of eigenvector centrality. This is a mathematical measure from network theory that offers a window into the importance of individual decisions within the complex web of game mechanics. By integrating this concept into the game design process, we embark on a journey to not only map out the endless pathways of learning but also to spotlight the most impactful routes that lead to deeper understanding and skill acquisition. In this section, we'll explore how eigenvector centrality can serve as a guiding star in the design of educational games, ensuring that each decision made by the player is not just a step in the game, but a jump towards meaningful learning.
Where do we begin?
We start by identifying the essential learning science theories and concepts we want to use as foundational rationale elements, such as SDT, SCT, Maslow's Hierarchy of Needs, EVT, and others. These theories can help inform the design of game elements that promote engagement, motivation, and learning at the subatomic level (as moderators and mediators of individual choices)
We can map atomic (ludemic) and subatomic levels of decision-making to broader game elements (mechanics, mechanisms, dynamics, components, etc..) and learning outcomes. For example, “exploration” can be linked to self-determination theory (SDT) by promoting autonomy and competence, while “challenges” can be linked to expectancy-value theory (EVT) by creating tasks that have value and appropriate difficulty levels for the learner. Each of these seeds the decision-making process.
Define nodes and edges: In the network of actions, each node represents a specific action or decision made by the player, and the edges represent the relationships between these actions. Some nodes will have higher eigenvector centrality, indicating their importance in the learning process. For example, a node representing a critical thinking task might have a high eigenvector centrality because it influences other nodes related to problem-solving and decision-making.
Perhaps we model expert and novice sequences: Create separate networks for expert and novice players, where the expert network represents the ideal sequence of nodes (actions) that lead to the desired learning outcomes. The novice network, on the other hand, may contain less effective sequences, highlighting areas where intervention or guidance is needed.
Compare and analyze sequences: Analyze the differences between expert and novice networks to identify areas where additional support or scaffolding is needed. Use this information to tailor the game design and provide targeted feedback or assistance for individual learners based on their unique sequences of nodes.
Iterate and improve: Continuously refine the game design based on observed learning outcomes and feedback from players. As learners progress and develop expertise, their sequences of nodes should more closely resemble those of an expert, resulting in improved learning outcomes.
D&D Equipment Example
(Using an example related to “equipment” in the context of a D&D game)
Imagine a player has just created a new character (a rogue with a background as a criminal). The player's starting wealth is determined by the rogue class and the criminal background, as stated in the equipment section. The player rolls the dice and calculates that their character starts with 150 gold pieces (gp).
Based on this information, the player decides to purchase the following equipment: leather armor (10 gp), a rapier (25 gp), a shortbow (25 gp), a quiver with 20 arrows (1 gp), a set of thieves' tools (25 gp), a burglar's pack (16 gp), and a dagger (2 gp). The player spends 104 gp in total, leaving them with 46 gp.
In this example, the player makes a series of decisions in choosing the character and then based on their character's class, background, and available resources (gold pieces). These decisions are strategic and aim to optimize the character's equipment to enhance their abilities and chances of success in the game.
Loop in Eigenvector Centrality Theory and other players:
If we have a social network of individuals (a party), represented as a graph with nodes and edges. Each node represents a person, and each edge represents a connection between two people. Eigenvector centrality is a measure that calculates the importance of a node in the network, taking into account not only the number of connections (degree centrality) but also the importance of the connected nodes.
We represent the ludemic actions as nodes in a graph where edges represent the relationships between these actions. The nodes in this graph would be:
Choose Rogue class
Choose Criminal background
Roll for starting wealth
Purchase leather armor
Purchase rapier
Purchase shortbow
Purchase quiver with 20 arrows
Purchase thieves' tools
Purchase burglar's pack
Purchase dagger
To calculate the eigenvector centrality of each ludemic action, we need to establish the relationships between these actions. For example, choosing a rogue class may influence the choice of a criminal background, which in turn influences rolling for starting wealth and purchasing equipment.
Suppose we compare the player's equipment choices with those of an expert player strategist. The expert might prioritize different equipment or may allocate gold more efficiently. We can create an alternate set of nodes based on the expert's choices and compare the eigenvector centrality of each action in both scenarios.
In the global context, we can evaluate how the player's equipment choices impact overall gameplay and problem-solving. For example, the player might have a lower eigenvector centrality for "Purchase rapier" than the expert, indicating that this action is less influential on gameplay success compared to the expert's choice. This could be due to a more optimized selection of weapons or better gold allocation.
In the resource management context, we can focus on the equipment choices and gold allocation. Comparing the eigenvector centrality of each action in this context, we can see how the player's choices impact their ability to manage resources efficiently. For example, if the player has a higher eigenvector centrality for "purchase leather armor" than the expert, it could indicate that the player's choice of armor is more influential in managing resources but might not be optimal for overall gameplay.
By comparing the eigenvector centrality of ludemic actions in both global and resource management contexts, we can evaluate the impact of each decision on gameplay success and resource management efficiency. This information can be used to guide players toward making better choices in future games or adapting their strategy during the game.
By analyzing the eigenvector centrality of each decision, we can estimate the importance of each decision in relation to the others. This can help players identify which decisions have the most significant impact on learning outcomes, allowing them to focus on improving those aspects of their gameplay.
Eigenvector centrality values do not represent a net impact of all decisions. Instead, they indicate the importance of each decision in the context of all other decisions, with higher values suggesting a greater influence on the outcome.
Different eigenvector values represent the sophistication of the gameplay. Higher eigenvector centrality values indicate that a decision has a greater impact on the overall outcome, and this can be seen as a reflection of the player's level of expertise or mastery in that particular aspect of the game. As the player becomes more experienced and proficient, they are likely to make decisions with higher eigenvector values, which can be associated with a higher displayed level of the skill defined by the connected learning outcome.
The learning outcome related to this example flow could be called something like "Strategic Resource Management." This learning outcome would assess a player's ability to make optimal decisions regarding equipment selection and allocation, taking into account factors such as character class, background, and available resources. Strategic Resource Management also tracks with decision-making that occurs in learning scenarios. By tracking the eigenvector centrality values of the decisions made by the player, we can get an insight into how well they are developing their strategic resource management skills and identify areas for improvement.
Multiplayer (Parties) and the Introduction of AI
Tracking multiple players and using AI as a “co-pilot” for decision-making.
Imagine we have four players in a party: Player A, Player B, Player C, and Player D. An AI expert player acts as a party co-pilot during gameplay, providing guidance and suggestions based on optimal strategies. Each player encounters ludemes throughout the game, which require decisions related to strategic resource management (described earlier) at the atomic level. These decisions are influenced by micro (sub-atomic) factors, such as motivation, goals, prior knowledge, and skill, as well as macro-atomic factors, such as the game context and decisions made by other players. The game's central learning mechanic (overall outcome to be measured) focuses on strategic resource management, and the ludemic level includes multiple related individual decisions about allocating resources, including equipment and abilities, in the most effective manner. Note: In this model, the learning outcomes include improved resource management, decision-making, and strategic thinking skills, which are all part of “strategic resource management"
Player A is highly motivated (sub-atomic) and has strong prior knowledge of RPG mechanics. When encountering a lude-level decision related to equipment allocation, Player A makes a decision to prioritize the most powerful items for the team's current situation, considering both individual and team benefits. The AI expert player suggests a similar approach, validating Player A's decision. Player A's decision has a high eigenvector centrality as it strongly impacts the overall gameplay and aligns with the expert player's recommendation.
Player B has a strong goal orientation (sub-atomic) but limited prior RPG knowledge. When encountering a ludeme-level decision requiring the allocation of abilities, Player B decides to focus on maximizing their own abilities, neglecting the needs of the party. The AI expert player suggests a more balanced approach, considering the strengths and weaknesses of the entire party. Player B's decision has a lower eigenvector centrality, as it does not contribute significantly to the team's overall effectiveness.
Player C has good metacognitive skills (sub-atomic) and is aware of their own learning needs. When encountering a ludeme-level decision related to resource management, Player C chooses to actively communicate with other party members to coordinate actions and share resources efficiently. The AI expert player supports this decision as it aligns with the game's learning outcomes. Player C's decision has a high eigenvector centrality, as it influences the overall team strategy and aligns with the expert player's recommendation.
Player D has limited prior knowledge and motivation but recognizes the value of learning from others. When encountering a ludeme-level decision requiring collaboration in resource management, Player D decides to follow the AI expert player's suggestions closely. This decision is influenced by Player D's motivation to learn from others (micro factor) and the game's focus on strategic resource management (macro factor). Player D's decision has a moderate eigenvector centrality, as it contributes to the team's overall effectiveness but may not be the most optimal choice.
In this scenario, the eigenvector centrality of each decision represents the significance of each player's resource management choices in relation to the expert player's recommendations and the overall gameplay. Players with higher eigenvector centrality values exhibit more sophisticated gameplay and are more likely to achieve the desired learning outcomes, such as improved resource management, decision-making, and strategic thinking skills.
Or…without the AI influence
In the absence of the AI expert player's recommendations, the party's decision-making process will be solely based on the individual choices of each player. The eigenvector centrality values will still represent the significance of each player's decision in relation to the overall network of choices made by the party, but without the influence of the AI expert player. Let's assign imaginary hypothetical values to each player's decision:
Player A's decision to prioritize powerful items for the team's current situation: Eigenvector centrality value of 0.45
Player B's decision to focus on maximizing their own abilities: Eigenvector centrality value of 0.15
Player C's decision to communicate and coordinate actions with the party: Eigenvector centrality value of 0.30
Player D's decision to focus on a balanced approach to resource management: Eigenvector centrality value of 0.10
In this scenario, Player A's decision still holds the highest eigenvector centrality value (0.45), indicating that it is the most influential choice in achieving optimal outcomes for the party.
The eigenvector centrality values also theoretically demonstrate the relative importance of each decision in achieving the desired learning outcomes, such as improved resource management, decision-making, and strategic thinking skills.
Players with higher eigenvector centrality values are more likely to exhibit sophisticated gameplay and contribute positively to the overall success of the team, while those with lower values may need to adjust their strategies to better align with the game's learning objectives.
By comparing the eigenvector centrality values of the players' decisions, we can identify areas where the party could improve their decision-making and better work together to achieve the game's learning outcomes. Additionally, this information can help guide instructional interventions or scaffolding to support player growth in specific areas of the game.
Whether we use human gaming experts or AI-driven decision making as a guide, integrating eigenvector calculations using the Co-Pilot system offers players a unique opportunity to develop critical thinking, resource management, and problem-solving skills in a dynamic, engaging environment. This fusion of mathematics with interactive (fun) gaming paves the way for a new era of serious games, where entertainment and learning mix seamlessly, promising a richer, more immersive experience for gamers and learners alike.