I remember a while ago reading about how bus arrival times in some South American country were studied and showed to follow psuedorandom distributions. I don't

I remember a while ago reading about how bus arrival times in some South American country were studied and showed to follow psuedorandom distributions. I don't remember the source, but the key differentiator [in that country] is that there is a person at the bus stop that tells the bus driver the time that the previous bus was there. This will cause said driver to increase/decrease speed in effect to counter bunching with the next bus. Unfortunately, I also don't remember how rigorous the study was.