The researchers studied how social interactions affect energy use in buildings by creating a social network from sensor data in an office building. They used a special model to analyze the data and found that energy savings of up to 20% were achieved while keeping occupants comfortable.