EQUILIBRIUM game: a virtual field trip through a complex system

Abstract

The introduction of systems thinking and modeling in school curricula is an ongoing endeavor that recently gained momentum through the stepwise implementation of the Next Generation Science Standards (NGSS) in the United States (Skaza et al., 2013). However, despite an undeniable proliferation of interest in systems-related topics, empirical investigations do not yet reflect much of an impact on general systems thinking capabilities. Most empirical investigations show that students and decision-makers poorly understand complex dynamic systems. For instance, studies carried out at Belgian high schools revealed “a rather poor general level of students' systems thinking ability” (Cox et al., 2017, p. 1). Even students at the MIT Sloan School of Management, who have an unusually strong background in mathematics and sciences, performed poorly in a paper-pencil test on system dynamics and delay (Sweeney and Sterman, 2000). Similar results were obtained from laboratory experiments on renewable resource management in the fields of reindeer herding and fishery. Despite their experience, practitioners systematically misinterpreted nonlinearities, overinvested into equipment, and overutilized their resources (Moxnes, 2000). To resolve such misperceptions, systems-centered education through illustrative materials, simulations, and other resources are indispensable. But these resources are often lacking. In the context of the NGSS, there is, for instance, increasing concern about, on the one hand, the readiness of teachers, students, and staff and, on the other, the availability of needed classroom technology resources to support the implementation of these new standards (Harris et al., 2017). Supportive teaching materials are essential because the underlying philosophy of the behavior of complex systems does not become a part of a student's thinking when experienced only through reading and listening (Meadows, 2007). Moreover, it is quite unrealistic to expect students to make correct inferences about a dynamic system from investigating conventional, static teaching materials (Wheat, 2007). Although still not fully integrated into most educational curricula, the use of interactive games as a means to master such challenges is not a new idea. The Beer Game is one of the earliest examples of the application of operational gaming in an education context. As the usability and performance of computers increased, popular games like STRATAGEM (Sterman and Meadows, 1985) and FishBanks (Meadows et al., 1993) found applications in classrooms and seminars around the world. Despite a reported crash in the market for educational games in the late 1990s, there is now new energy and perspective behind the idea of learning games (Klopfer and Osterweil, 2013). It is also in that spirit that the last few years have witnessed a comeback of participatory simulations (e.g. Taillandier and Adam, 2018, Maharaj et al., 2011, or Berland and Rand, 2009). Educational games with a systems science background have also entered the commercial market (e.g. GaiaTown or ECO). Blockbuster videogames like SimCity or Minecraft (Smaldone et al., 2016) were turned into educational versions. The continuing trend of global environmentalism (Falkner, 2012) fuels such developments and encourages expanding the traditional boundaries of educational gaming. Whereas educational games generally incorporated many different genres of gaming, educational games in the field of systems science and systems thinking are mostly limited to a type of game that may be called Interactive System Simulation (corresponds to Participatory Simulations as promoted by complexity scientists). In such games, human users interact with the simulated world by controlling the world's parameters or by steering the behaviors of actors and agents in a dynamic system (Taillandier and Adam, 2018). Examples are the simulation game Tradeoff (Verutes and Rosenthal, 2014) or the role-play simulation World Climate (Sterman et al., 2015). Although very sophisticated in its implementation, the underlying structure of the system typically remains hidden in such simulations. As a result, students learn how the system behaves, but they struggle to explain why it behaves in a specific way. In this article, I introduce a game called EQUILIBRIUM, which supplements the analysis of behavior by the important explanatory aspect of systems analysis. In other words, the game objective is to explore what structures might have caused the observed dynamics in a systemically coherent way rather than analyzing the system's sensitivity to change. The following section is an introduction to the game objective, game interface and intended “solution.” The second part of the article reports on lessons learned from implementing the game in university courses and gives instructions on teaching systems thinking with EQUILIBRIUM. The design of EQUILIBRIUM is similar to classic murder mystery games. In such games, players are confronted with an incidence of crime and a number of hints, which are designed to reveal more and more information about the murder. In such a game the underlying process is not actually implemented (the incidence of crime is not actually reenacted in the game), but evidences and hints (e.g. a bloody knife) gradually reveal what happened. In EQUILIBRIUM, hints (i.e. game contents) successively reveal the structure of a human-environmental system that is destined to collapse. This incidence takes place on a fictitious island that was once inhabited by a small human population. At the core, the scenario was inspired by Jared Diamond's popular book chapter on the environmental collapse of Easter Island (Diamond, 2011) and Barry Richmond's exercise on the mental modeling of tree stocks (Richmond and Peterson, 2001). Students discover the island in a kind of virtual field trip. Their journey takes them to places where they can find evidence for processes that once shaped this environment. Eventually, students need to find out what happened to the human population and what systemic relationships may have caused their disappearance. To answer this question correctly, students need to come up with a coherent hypothesis. The game does not make explicit any systemic structures, which leaves some room for alternative interpretations of the system-in-focus. Nevertheless, the same dominant structure is inherent in any coherent interpretation of that system. This structure is the main criterion for assessing the students' performance in the game (see section Student performance and lessons learned). To share thoughts about potential explanations of systems failure, students need to be very clear about the system's causal structure. In this manner, participants practice the communication of systemic mechanisms to consolidate their systems thinking capabilities. By inspecting interesting locations, as indicated by the bull's-eye map, players come across important items that reveal systemic relationships. The player will discover these items in the same order as presented in Table 1, provided they move along the main road on the island. Apart from the items a player can find, some additional hints are more subtle, like noisy swarms of mosquitos that appear in very high concentrations, particularly at locations that are far from bird-nesting sites. This supports the conclusion that birds feed on mosquitos. In addition, the vast majority of trees on the island is the same coniferous trees that are also bred on the tree farm (see Table 1, Item 6), which indicates that islanders planted them in the course of a replanting project. Individual perceptions may contribute to slightly different interpretations of the hints that are provided in the game. Despite this vagueness, the behavior hints indicate a self-enforcing loop that is a common feature of every potentially coherent interpretation (see Figure 1, loop is in blue). This dominant loop structure contributes to a gradual environmental degradation, which, in the long term, negatively affects the human population on the island. The following paragraphs outline the function of the dominant loop and present a hypothetic and coherent interpretation of the loop's impact on the human-economic system. The birds feed on mosquitos, which in turn subsist on the blood of larger mammals like horses. Mosquitos also disseminate a viral disease that threatens the domesticated horse population on the island. The island community strongly relies on the export of racehorses to the mainland. Apart from horse breeding, a small timber industry also exists on the island. Horse and timber economies are interrelated: Horse farmers raise large wooden monuments as a kind of mojo, which they obtain from local artists. These artists, in turn, receive the raw timber from lumberjacks. The economic prosperity of the timber industry and the local artists is consequently highly dependent on the horse breeder's demand for wooden monuments. As more and more large trees fall victim to the production of wooden monuments, singing birds suffer from a significant shortage of breeding sites. Birds need sufficiently large trees for building nests for their offspring, otherwise chicks are caught by predators. Eventually, the bird population diminishes because of the shortage of breeding sites. A decrease in the bird population benefits mosquitos, which spread out and exploit new habitats. Mosquitos carry a deadly horse flu. The number of viral infections increases proportionally to the mosquito population. The horse breeders believe that wooden monuments would protect their horses from the horse flu and raise even more monuments made from wood. This self-enforcing loop causes a continuous increase in horse mortality rates, which eventually forces horse breeders to emigrate. Timber workers and artists follow shortly after, due to the lack of demand for their product. Moreover, the islands' residents tried to counteract deforestation by replanting one seedling for each tree that was harvested (see Figure 1, green). Upon closer inspection, however, this turns out to be ineffective. To preserve the ecological equilibrium, the number of large trees needs to be maintained because only large trees are suitable bird nesting sites. The replanting policy is ineffective because it leads to long-term depletion of large tree stocks before the first seedlings reach maturity (Richmond and Peterson, 2001). The time it takes for the implemented policy to balance out the tree-logging activities is equivalent to the time it takes for the tree seedling to grow to maturity (about 100 years in EQUILIBRIUM; see Table 1, item 3). So, the stock of large trees continues to diminish (see Figure 2). In summary, the hints provided reveal a story on the overuse of a slowly renewable resource (large trees). The islanders were unaware of driving forces and mechanisms (a self-enforcing loop). As a result, they implemented ineffective (replanting tree seedlings) or even harmful (construction of more wooden monuments) strategies to fight environmental degradation. Despite the provided hints, it is unrealistic that students will reach a complete understanding on their own. The necessary guidance depends on the student's prior knowledge and objectives associated with planned teaching activities. In preparation for the game, the following section provides some suggested teaching activities and contents. After students played the game, it is recommended to work out the solution in an instructor-led student discussion. It is essential, however, to eventually provide students with the game solution: “as described in section Student performance and lessons learned Game ‘solution’: Explaining the loss of equilibrium.” EQUILIBRIUM is well suited to consolidate prior theoretical knowledge on (i) causality; (ii) cascade effects; (iii) feedback; and (iv) delay. For the remainder of the article, I will refer to these components as knowledge blocks. As a minimum requirement to play the game, a fundamental understanding of the knowledge blocks 1, 2 and 3 is essential. According to my experience, the introduction of causal loop diagrams (format is depicted in Figure 1) as a basis for the discussion of system structure and behavior is most efficient in teaching this content. Causal loop diagrams (CLDs) illustrate the fundamental characteristics of causality (covariation and direction), facilitate the recognition of feedback and chains of causality (cascades) and support the qualitative analysis of system behavior. Whereas simple causal structures are sufficiently explained through CLDs, the recognition and understanding of delays (knowledge block 4) inherent to the forestry system, calls for a closer inspection of quantitative system dynamics. Stock and flow models are well suited to investigate such dynamics. For instance, planting tree seedlings is a flow that replenishes the stock of small trees (no bird breeding). As small trees grow to maturity (flow), they enter the stock of large trees (bird-breeding sites) which in turn becomes depleted by the felling of trees (flow). Lacking student knowledge on stock and flow dynamics will hamper the correct interpretation, as time delays are inherent to the stock and flow structure (see Figure 3). A fundamental understanding of the nature of stocks and flows, the perception of a world made up of stocks and flows, as well as an awareness for time lags between system compartments should enable students to draw correct conclusions. To achieve such a level of understanding, a potential approach is to discuss the static EQUILIBRIUM landscape with students as a pregame activity. This involves the discussion of questions like what physical landscape elements represent stocks (e.g. number of mosquitos or number of birds at a point in time), how they are linked by flow processes (e.g. deforestation is a flow that depletes the tree stock and replenishes the stock of construction materials) and what quantities of flow can be expected. To further advance student's understanding of stock and flow dynamics prior to the game, the incorporation of a graphical integration exercise as presented in Sweeney and Sterman (2000) is advisable. The value chain in EQUILIBRIUM (see Table 1, Item 5) lends itself nicely to derive examples for such an exercise (see Figure 4). Despite an introduction to stock and flow dynamics, knowledge block 4 (system behavior as an effect of time delay) seems hard to grasp for most students. Accordingly, knowledge block 4 is more suitable for advanced courses, while knowledge blocks 1, 2 and 3 cover more fundamental aspects of systems thinking. The following section reports on experiences gained through an application of EQUILIBRIUM that is focused on the fundamental contents of knowledge blocks 1, 2 and 3. To date, the game has been used in two elective courses that are part of the master's program on Applied Geoinformatics at the University of Salzburg. Overall, 13 out of 14 students who actively participated in the exercises agreed to provide feedback for research purposes by clicking an opt-in checkbox. On average, participants spent 50 minutes playing the game, whereby the time effectively invested ranged between 15 and 90 minutes. These discrepancies resulted from the voluntary nature of the exercise. Except for one person, who indicated some prior knowledge on geographical systems analysis, none of the participants had a background in system dynamics or deliberate knowledge of systems science. The majority of students indicated geo-related studies as their major field of prior education (Geography, Geoinformatics, Cartography, GIS and the like). In a few cases, a considerable lack of English writing skills seemed to affect the performance negatively. The variation of language skills may in part be explained by the diversity of the group in terms of nationality (Poland, Kyrgyzstan, Germany, Austria, Hungary, China, Pakistan, and India). Despite the distinct focus of the introductory session on feedback loops, spontaneous recognition of the dominant loop structure was rare (see Figure 5). Only two students closed the self-enforcing loop and referred to its harmful function: “The people tried to protect their horses and to plant new trees, but they weren't able to break the circulation. They left the island with their last horses.” “This chain of events and factors can be seen as a self-reinforcing feedback loop.” Most students missed a link in the dominant loop structure, which made it impossible to explain the system's failure in a coherent way. Whereas relationships such as “less large trees – less birds”, “less birds – more mosquitos”, “more mosquitos – more flu dissemination” and “more flu dissemination – less horses” were well recognized, only a few referred to the upsurge in monument production and deforestation as a cause of increasing horse mortality. Other interpretations did not contribute to a coherent explanation: “The disease attracted mosquitos”; “The mosquitoes may have destroyed trees, too.” Students also seemed to be predisposed to topics associated with climate issues. The lack of rivers, something that looks like a dried-out creek and a few patches of brownish grass, triggered the following independent conclusions: “Drought seems to have been a serious problem, too”; “So because of the increasing temperatures and the less water leads to people leave the island.” The religious belief in the protective effects of wooden monuments as a key driver of the system was mentioned in two instances. One participant explained that mechanism in more detail: “The inhabitants also believed on that, if they raised up monuments (a kind of wooden menhirs) the Gods will be gracious to their horses, and will stops the spreading of disasters. The problem was that, this monuments needed huge wide wooden beams (picture in wood processing plant).” Just like most of his peers, this student was able to connect system elements. Students even described cascades that were composed of four to five consecutive links but eventually failed to close the dominant loop. Similar misperceptions were reported in Cox et al. (2017) and Sterman (1989). Another explanation of weak performances is that some of the game items remained undiscovered. As it turned out in the discussion, the nonsequential nature of the game poses another challenge to participants. In other words, students will discover respective arrows of the dominant loop not necessarily in sequential order (shown in Figure 6). Students who follow the main road on the island will most likely know about the links “trees-birds” and “mosquitos-horse infections” before they come across the link “birds-mosquitos.” For this reason, they need to memorize two independent relationships before they pass by an item that reveals a third relationship to connect them. Whereas the nonsequential order of discoveries is an intended feature of the game, misperceptions that result from an omission of essential information need to be addressed. A potential solution to this problem is proposed in the following concluding section. In this article, I present a game to support teaching systems thinking in education. The game is designed as a point-and-click adventure that displays an island ecosystem that is characterized by environmental degradation and human exodus. The students' task is to find out what caused this situation. Student hypotheses are evaluated against a dominant loop structure that is inherent in every logically coherent explanation (see Figure 1, loop is in blue). So far, EQUILIBRIUM has been used in two elective courses at the University of Salzburg. The game triggered controversial discussions among students and revealed considerable misperception of system structures by students in this particular group. While students could link cascades that were composed of four to five connected relationships, the vast majority failed to close the dominant loop. This confirmed empirical observations that open-loop or one-way causal thinking is quite common (Sweeney and Sterman, 2007). To overcome potential misperceptions, an introduction to basic principles of system dynamics, a supervised discussion, as well as a comprehensive debriefing are highly important elements of every lecture that uses EQUILIBRIUM. Similar constraints associated with unguided gaming sessions were reported by (Riess and Mischo, 2010). Apart from potential limitations in terms of natural systems thinking capabilities, some of the students failed to identify important game items, which explains gaps in their argumentation. In the game, the search for such items is largely unguided. Only landmarks on a bull's-eye map indicate their locations. To address flaws that result from such a design, a scoring system may be implemented in the future. In this new version, students would earn points for clicking and viewing important items. A score below the maximum then indicates that they have yet to find potentially interesting sites. Despite this problem, EQUILIBRIUM fulfilled its principal function of triggering discussions about fundamental aspects of systems thinking. The different opinions and interpretations that resulted from the discussion reflect the actual strongpoint of EQUILIBRIUM. The game leaves room for discussing different interpretations of dominant loop structures. Students tend to passionately defend their own hypothesis about the system in a kind of simulated scientific discussion. The effects of this type of learning via knowledge building will be subject to future empirical investigations. The EQUILIBRIUM resources are ready for download under https://bit.ly/2PLYWi4. Start the game directly from the executable file. There is no installation required. The game runs on conventional home computers and notebooks with CPUs that are equivalent or better in performance than Intel i3 (see Table 2). When launching the game, make sure to select an appropriate graphics quality in the provided dropdown menu. Depending on the specifications of your system, loading the 3D environment takes up to 2 minutes (indicated by a rotating hourglass). In the introductory section of the game, you will find an anonymous online survey. Please ask your students to contribute to this survey. It will facilitate the continuous improvement of the game. I would like to thank Andrea Pődör for testing the game in a course at the Óbuda-Universität. Many thanks go to Markus Schwaninger and Yaman Barlas whose comprehensive feedback helped improve the manuscript.