I haven't read the article, but I ride the bus a lot. Here is my subjective take: bus drivers get into a very distinct and slow pattern of their bus driving behavior. Slow to do everything, never not taking their break, never attempting to make up for lost time.
So in their daily routine, the traffic between stops is variable, but their little rituals and apathy to the travel plight of passengers is not.
They also like to take up time socializing between driver swaps, which aren't accounted for in bus schedules. Thus the largest factor is bus A may have no traffic, and bus B had a lot, but their stops at each place take the same amount of time, thus they end up clumping when traffic let's one drive between stops faster than the other, who will still be on his break at the Bart station.
The point of the article is that you can generate this behavior in an ideal system without taking into account differences in drivers.
Put simply, small stochastic differences in arrival rates of passengers or initial speed of buses will amplify themselves. If you take three buses and shift the center one in time, you'll end up with one longer gap and one shorter one. The bus with the longer gap will tend to find more people waiting at each stop, while the bus with the shorter gap will find fewer people and move faster. Ultimately, the faster bus will catch the one in front of it and follow it, with zero people getting on at each stop. There will then be a -huge- gap, followed by a very slow bus that has to pick up a ton of people at each stop.
No, it doesn't. The article says that bunching will happen even with perfect robot drivers. Your comment says that bunching happens because drivers are lazy.
(Also, the article is a tiny flash game and like five sentences of explanation. This isn't tl;dr territory.)
So in their daily routine, the traffic between stops is variable, but their little rituals and apathy to the travel plight of passengers is not.
They also like to take up time socializing between driver swaps, which aren't accounted for in bus schedules. Thus the largest factor is bus A may have no traffic, and bus B had a lot, but their stops at each place take the same amount of time, thus they end up clumping when traffic let's one drive between stops faster than the other, who will still be on his break at the Bart station.