display | more...
Rutgers University has what I believe to be one of the most poorly managed public transportation systems in all of Western civilization.

Rutgers New Brunswick is made up of four campuses: Livingston, Busch, Cook/Douglass, and College Ave. These campuses are spread out over several miles -- it is not possible to walk between them, and bicycling is extremely difficult for a variety of reasons.

A large number of different bus lines are run... AX, between College Ave and Busch ... B, between Busch and Livingston ... EE, between College Ave and Cook/Douglass ... F, also between College Ave and Cook/Douglass, by a different route ... G, between Busch and Cook/Douglass ... GG, between Livingston and Cook/Douglass ... H, between College Ave and Busch, with a longer route than the AX ... L, between College Ave and Livingston, with a longer route than the LX ... LX, between College Ave and Livingston.

This setup creates a fully connected graph; all campuses have bus lines running to all others (with College Ave being doubly-connected).

If only it worked. In theory, each bus runs on an interval schedule; on any given line, at any given stop, a bus is to arrive every n minutes, where n is specified by the applicable schedule. In practice, this is implemented by putting m buses on the route and having them depart the garage n minutes apart - with little subsequent communication or coordination - where m is (supposedly) determined by dividing the line's trip-time by the desired n.

Result? Chaos! Several variables interact to make this system collapse by midday.

• Vehicular traffic is not uniform. Variations in road conditions mean that a route segment that one bus takes four minutes to cover might take another bus only two - thus making the arrival intervals irregular.
• Passenger demand is not uniform. Classes let out at specific times, and so a bus arriving at a stop at 11:06 might drive by, not even having to stop to pick up or discharge, while one arriving at 11:11 might not be able to accomodate all the people wanting to board. Again, this leads to irregular intervals arising.
These irregular intervals by their very nature grow worse. Suppose a steady stream of students is arriving at a bus stop, and buses are arriving at regular intervals. Then, each bus takes the same number of passengers and spends the same amount of time at the stop. Suppose now instead that the buses are not arriving at regularly. Then, they instead come in pairs -- say, two buses five minutes apart, then a fifteen minute wait for the cycle to restart. What happens? The first bus in the pair takes fifteen minutes worth of accumulated passengers, whereas the second takes only five -- and spends less time stopped. This again draws the pair closer together, until you get to the common sight of a completely empty bus following close behind one with students mashed into the stairwells like sardines.

What can be done? The UMass (University of Massachusetts Amherst) transit system is no less complex than Rutgers', spanning five campuses (UMass, Amherst, Hampshire, Smith, and Mt Holyoke) with nearly a dozen different lines; yet buses run, for the most part, on schedule. The trick? UMass buses have a schedule, and they stick to it. Unlike Rutgers' interval-based system, UMass bus drivers have a specific timetable detailing the exact times they are expected to depart their stops. These times are sufficiently generous to ensure that buses always arrive with time to leave as scheduled; buses wait at their stops if they've arrived early.