This article looks at the very real possibility that AVs will actually move slower in central cities than cars do today.  This is based on two notions – first, the idea that AV’s risk averse algorithms will understandably slow them down or stop them whenever a pedestrian or cyclist crosses the street.  Second, the idea that pedestrians and cyclists – now sure that  cars will be stopping – will step off the curb or into traffic whenever they please, creating havoc for the efficiency of automobiles.  Author Adam Millard-Ball asks us to imagine AVs trying to get through Manhattan while obeying all traffic rules and stopping with every pedestrians crossing at will.

The article points to a key issue regarding AVs in dense environments and how the interaction with other modes will severely hamper some of the largely claimed increases in speed.  It would seem that these increases will most probably exist in suburban and exurban areas, but not as much in central cores.  How does the speedy highway leading into the city deal with the congestion glut as cars enter slower networks downtown?