source: google research: the power of collaboration: how we can reduce traffic congestion
level: research
google research ran a six-month experiment in 10 major us cities to test whether small changes in routing recommendations could ease traffic. the team modified the google maps algorithm to steer a tiny fraction of trips away from pre-selected congested road segments, using alternative routes with similar travel times. on treatment days, less than 2% of observed trips received altered directions, yet the effects rippled through the network.
the results showed a median speed increase of about 2% on the targeted segments, with fuel consumption dropping by 0.5% to 1.0%. across all affected roads, including those that absorbed diverted traffic, speeds rose by roughly 0.35% on average and 0.5% during peak hours. these gains came from dispersing vehicles away from bottlenecks, allowing peripheral roads to handle higher volumes without slowing down. the study estimates this could save thousands of tons of co2 equivalent per city each year.
the work provides a blueprint for using navigation platforms to manage traffic at a system level. by coordinating a small share of trips, the approach benefits all drivers, not just app users. the researchers used a hierarchical bayesian model to measure outcomes, sharing information across cities and time periods. this experiment-based framework could extend to dynamic signal control and real-time network optimization as smart-city infrastructure grows.
why it matters: shows that minor, low-cost routing interventions in navigation apps can yield city-wide traffic and emissions improvements, offering a practical path for data-driven urban mobility management.
source: google research: the power of collaboration: how we can reduce traffic congestion