Solving big problems, one burrito truck at a time
The Burrito Optimization Game—a free, web-based app created by Lehigh University researcher Larry Snyder and software developer Gurobi Optimization—draws thousands of global users and sparks classroom engagement
image: The Burrito Optimization Game was developed by Gurobi Optimization in partnership with Lehigh University Professor Larry Snyder. This interactive game challenges players to deliver burritos in a way that maximizes profit through supply chain optimization.
Credit: Courtesy of Gurobi Optimization
When trying to teach a complex subject, sometimes the best strategy is to wrap it in something familiar.
Like a burrito.
An ideal candidate for such an approach? Mathematical optimization, which uses math and computing tools to make decisions about complicated systems. It has the power to help a range of people from professionals in data science, engineering, and business, to individuals looking for a better route to work. But to the uninitiated, its promise can be a hard sell.
“I think that people who know a little bit about optimization either think it’s trivially easy, or it’s impossibly hard,” says Larry Snyder, Lehigh University’s deputy provost for faculty affairs and a professor in the Department of Industrial and Systems Engineering. “What they don’t realize is that there is theory and software available that can help solve really hard, really complicated problems.”
Enter the Burrito Optimization Game, which recently surpassed 50,000 plays by users around the world.
In 2022, Snyder collaborated with Gurobi Optimization, a company that develops mathematical optimization software, to create a free, web-based learning tool that would make the concept of optimization more approachable to more people.
Here’s how it works: The game opens with a map of a city that identifies buildings where there is some demand for burritos. Players can drag and drop their burrito trucks at locations where they believe they will sell the most wrapped wonders, and best balance their costs and revenues. When they are done placing their trucks, their solution to that day’s demand is compared with the mathematically optimal solution.
“The player might learn that their solution is 20 percent worse than optimal, for example,” says Snyder. In the game, suboptimal translates to lots of disappointed eaters, much missed revenue, or too many trucks per hungry mouth. “They’ll then get feedback in the form of lightbulbs that appear on the map, which reveal little tips like, ‘You missed some demand here,’ or ‘You should have located a truck here,’ so you understand where you went wrong.”
The game progresses in difficulty across days (as opposed to levels). One day, the map might become more complex with more locations to consider. On another, it might rain, which limits how far people would be willing to walk to a truck. Or, perhaps, a disruption in the supply chain means cheese suddenly costs more. As the days go on, more uncertainty is added. Early in the game, you may know that 12 people in a particular building are die-hard burrito bingers. But later on, all you know is between 10 and 14 people might be craving one.
“All these scenarios change the decisions you have to make,” says Snyder.
The team designed the game initially with data scientists and other “optimization-adjacent” people in mind, but quickly realized the format made it an effective tool for users of all ages. Professionals use it to learn the basics of optimization, but Snyder and other faculty at Lehigh have employed the game in introductory mathematical optimization courses and as an outreach tool in local high schools to bring optimization into their young orbits. The game’s “championship” mode allows users to compete against each other, a feature that’s well-suited for both classrooms and, says Snyder, academic conferences.
“My kid played it when they were 10 years old,” he says. “So it’s easy to start the game without knowing anything about optimization.”
Ultimately, the team set out to create a tool that could introduce people to optimization and its unique ability to solve what Snyder calls “overwhelmingly complicated” problems.
“We wanted people to have a sense of why these problems are hard to solve by hand, and to realize that there are algorithms that can solve them for you,” he says.
And it’s resonating. Snyder has received emails from users around the world, in particular from those who incorporate the game into their teaching. Sometimes they have suggestions for features, or requests for data, and sometimes they just want to tell him how much they love it.
“I’ve been really pleased with how much traction it’s gotten, especially in the classroom,” he says. “Games just make learning more fun. They’re really useful in helping someone make the jump from not understanding something at all to understanding enough of it to spark their interest in learning more.”
Related Links