“These are games where the idea is to learn ideas,” Dr. Clayton Lewis of the University of Colorado at Boulder stated to clarify the topic of Friday’s MCS Colloquium. “Not to learn a skill, but to learn some important concepts, concepts in math or science or something like that.” The presenter has for several years administered a Computer Science course on educational game design at CU, with an end goal of getting students to develop games that both engage the player and strengthen their understanding on hard-to-grasp concepts in evolutionary biology, parallel computing and other abstract areas.
“The romantic fantasy that I’m not alone in having is that you could create a game that people would voluntarily play themselves,” Lewis explained, giving the example of children playing with LEGO blocks. With a similar mix of engagement and education, Lewis posited that “you wouldn’t have to require digital logic for EE because people would come to school already knowing about digital logic because of all the fun they had with Robot Odyssey,” a computer game from the 1980s that required players to learn how to design increasingly complex digital logic circuits in order to progress through its levels. Unfortunately, Robot Odyssey was a flop, and Lewis admitted that “What I’ve learned is that this problem [of creating engaging, educational games] is a lot more difficult than I thought it would be.”
Lewis noted that, in areas other than the purely conceptual, games do quite well at inculcating players with facts, skills and experiences. For memorization activities, “cheesy” games–those whose game content is completely unrelated to the material being learned–work admirably, and games teaching a skill (like touch typing) were likewise well-received. Lewis also stated that “immersion-type” games, like the U.S. military’s America’s Army, did their job while still remaining engaging. However “concept games” were generally unsuccessful in staying true to the form of the learning experience while still remaining engaging for players.
“One of the problems that we found”, Lewis explained, “was that students made bad decisions…about what the learning goals of their games should be…The students would choose largely pointless learning goals because they would think, ‘I can make a game about this.'” Such games, Lewis said, ended up “teaching” players concepts that were obvious even without the game, and allowed users to tinker with the game until they won, with no need to learn any abstract concepts.
A major problem in initial iterations of the course was finding difficult concepts to design games around. At one point, students were assigned a K-12 teacher, who would describe concepts that their students had problems absorbing. But Lewis found that “K-12 teachers generally don’t have god learning goals in mind because they find it difficult to tell you things that their students find it hard to learn. If you think about it,” he continued, “teachers don’t spend their time on things that are hard to learn. They spend their time on things that their students can learn. If they were working on something that there was a good chance students couldn’t learn it, they wouldn’t be doing their job.” As a result, learning concepts were drawn from the college level, with fields for study narrowed to two: biological concepts from Dr. Mike Klymkowsky (of CU) and parallel computing concepts from within the Computer Science department.
Even then, creating games to explain difficult concepts was difficult. One of Lewis’s examples of this shortcoming was with a set of student-created games on the Dining Philosophers problem from the computer science world: a virtual dinner table is set with an equal number of utensils and philosophers who must eat, and each philosopher needs two utensils to eat. The purpose of the problem is to illustrate the problems of starvation and deadlock in operating systems and parallel computation, with the solution being a programmatic way of preventing both problems. However the student-made games allowed the user to directly manipulate which philosopher picked up which utensils at a given time, leading Lewis to state that “I have no difficulty arguing that you would learn nothing about parallel computing from these games. To really understand what’s going on in Dining Philosophers,” he explained, “you have to work at the level of policy, and these games operate at the level of manipulation.”
Lewis noted that other problems arising from the class included the issue of premature commitment. “We saw this big time in early editions of the class,” he explained. “The students decided what they wanted to do for their project, and they would work on that the whole semester…and one of the reasons that the results were disappointing is that people had made bad decisions early on about what they should work on.” Lewis and his associate with the class, Alex Rapenning, solved this issue by instituting “gamelet madness,” a five-week beginning-of-semester period where students were expected to write an educational game each week, with each game tackling a very different topic from the others. A student’s final project was only decided upon well into the semester.
Another issue was that of student expectations: in Lewis’s course the focus was on “gamelets”, short-form activities relying on a simple programming framework rather than on large, complex titles. “We’re thinking of Pac-Man, versus Grand Theft Auto,” Lewis explained.
A final issue that Lewis mentioned was the fact that engagement with a particular task had nothing to do with the learning content of that task, nor were constant rewards the best way to keep players engaged in a game. Lewis used gambling, movie plots and spectator spots as examples of the latter concept: a constant state of tension, with small wins and losses mounting to larger overall advances, was more engaging than constant rewards. His conclusion: “Any effort you put into building engagement…is not going to solve the tough problem, which is getting [the game] so someone will learn from it. Then of course you have the other side of that: all the work that you put into [increasing learning content] doesn’t help you with the engagement part.” Lewis then gave the example of Will Wright’s Spore, a relatively recent game that ended up sacrificing trueness to evolutionary form in favor of placing the player in the role of an intelligent designer to foster engagement in the game. “Engaging rubbish wins over conceptual content,” the presenter lamented.
Lewis was, however, open to the idea that a solution for creating engaging, conceptually rich educational games was possible, noting that the process of game creation actually proved to be a very good learning process for the game designers themselves. “Instead of trying to make games that people learn from, let’s see what people learn from creating games,” Lewis explained of Scalable Game Design, a course where this concept was tested. “The answer is that people learn quite a bit.” He also noted that users discussing the content of a concept-focused game did actually learn those concepts, in contrast to users who simply played the game. Lewis conceded that, if such discussion were encouraged in a classroom setting, conceptual games might actually have a positive impact on student learning, though he opined the vastness of the gulf between that reality and his romantic dream of having people voluntarily play a concept-focused game and learn a complex concept in the process.