Any technology which allows for self-driving cars will allow for self-driving shuttles, buses, and trains. Rumors of transit’s death have been greatly exaggerated.
The main impact of labor costs on route and vehicle choice is to bias operators towards larger vehicles. Infrequent “circulator” buses are less convenient than taxi shuttles, but it’s cheaper to pay a single operator to drive a mostly-empty bus around in a big loop than it is to pay multiple cab drivers to be on call, even if the latter system might pick up a few extra riders.
Likewise, spending a lot of money for a rail system makes sense if each train can replace several buses. Whatever the merits of Houston’s LRT extensions, the original 2004 Main Street Line made sense as a replacement for the numerous buses operating between TMC and Downtown.
But self-driving transit changes this calculus. Remove the operator from the equation a 2x or 3x multiple for LRT vs bus costs becomes a 4x or 5x multiple. There is no way to justify the cost of a guideway.
(Do not respond with “but Ottawa”, or similarly played-out arguments. Jarrett Walker will spend the rest of his career batting down bad anti-bus arguments, and I see no reason to duplicate those efforts).
Across the board, self-driving technology will shift the cost curve in favor of smaller, less-expensive transit vehicles. What it won’t do is promote personal auto usage. The labor cost of the personal automobile is already zero; the driver volunteers to drive himself. And adding self-driving technology won’t make cars any cheaper*.
Rail won’t completely die, either. Very high speeds – 100, 125, or 220mph – require steel wheel on steel rail, or an alternate technology that offers even less rolling resistance (e.g. maglev). Express rail offers a ride that is orders of magnitude nicer than buses, even on dedicated guideways. And the visual presence of trains acts as a sort of civic virtue signal.
Instead of talking about transit’s impending death, we should instead be talking about more busways, and possibly building currently-planned rail lines as busways. These can be bus-only affairs like Pittsburgh, or they can be interconnected networks of managed lanes like what Houston *ought* to be building.