Bus bunching turns a 10-minute frequency into 0-0-30 minute gaps because one late bus creates a self-reinforcing cascade

Bus bunching is the phenomenon where buses scheduled to arrive every 10 minutes instead arrive in clusters of 2-3 followed by a 25-30 minute gap. It's caused by a self-reinforcing feedback loop: a bus runs slightly late (traffic, a wheelchair boarding, a long dwell at a busy stop), which means it picks up extra passengers who would have caught the next bus. More passengers mean longer boarding times, which makes it later still. Meanwhile, the bus behind it has fewer passengers to pick up, so it runs ahead of schedule and catches up to the late bus. The result: two buses arrive together, then nothing for 20+ minutes. The NYC Comptroller's office found that more than one in ten MTA buses fail to maintain even spacing, with riders on the worst routes waiting an average of 3.62 minutes longer than scheduled. The rider experience of bunching is far worse than raw delay numbers suggest. A rider who just missed the first bus in a bunch now waits 25-30 minutes for the next service, even though the schedule promised 10-minute frequency. They have no way to know this in advance — the schedule says 10 minutes, the app might say 10 minutes, but the reality is half an hour. This unpredictability is more damaging to transit usability than a reliably slow schedule, because riders can't plan around it. A bus that's reliably every 20 minutes is more useful than one that's "every 10 minutes" but actually arrives at 0-0-30 minute intervals. The bunching pattern also means the trailing buses in a bunch are nearly empty while the leading bus is dangerously overcrowded. Bunching persists because fixing it requires real-time operational control that most agencies lack. The technical solutions are well-known: hold the early bus at a timepoint for 60-90 seconds, or instruct the late bus to skip a low-ridership stop. But implementing this requires dispatchers who can see all vehicles in real time and communicate instructions to drivers, plus drivers who follow those instructions. Most agencies' dispatch operations are stretched thin (one dispatcher monitoring 50+ vehicles), and driver culture resists being told to wait when they're trying to stay on schedule. King County Metro reported progress reducing bunching in 2025 through automated holding algorithms, but this required a complete overhaul of their CAD/AVL system — an investment most agencies haven't made.