I'm pretty sure the described tactics should succeed at optimising expected travel time, but they (knowingly) do so at the cost of variability. I regular get t

I'm pretty sure the described tactics should succeed at optimising expected travel time, but they (knowingly) do so at the cost of variability. I regular get to experience the opposite: waiting 20min in the rain, at which point three busses arrive as a group. Yet another extreme is low variability with high scheduled travel time, achieved by including a minute of slack at every second stop or so. It's a result of an inherent instability in bus operations: As soon as the intervals diverge only marginally, faster busses become even faster, because there are fewer people waiting for them and vice versa.