The argument is about where the empty car waits
Waymo has asked San Francisco to create a formal permitting standard for staging — letting its vehicles park at the kerb between rides rather than circling. The case for it, made by Waymo's own policy manager, rests on a figure worth reading twice: roughly 40 per cent of ride-hail and autonomous vehicle miles are run empty.
That number has appeared before in academic work and in campaign material. What is new is who is citing it. An operator arguing for kerb space has to establish that the empty running is large enough to be worth solving, and in doing so has put its own figure on the table.
The modelling behind the request, funded by Waymo and carried out at Berkeley and UC Irvine, simulated around 1,700 vehicles in San Francisco. Forcing them to circle rather than stage added more than 200,000 kilometres of driving a day. That is the cost of empty circulation, quantified, for a fleet no larger than a mid-sized private hire operator.
The rebuttal, from a transport advocate in the city, does not dispute the empty miles. It disputes who should pay for the kerb, and points out that cities never solved the same problem for Uber and Lyft drivers, who have spent a decade with nowhere to wait in the places where waiting is most needed. Its most interesting line is an economic one: a modest charge for kerb space would probably still be cheaper for an operator than driving empty.
That is the moment worth marking. Empty running stops being an efficiency argument and becomes a cost compared against another cost. It joins a per-mile tax on empty vehicles in Washington D.C., operators choosing where to put depots so vehicles run fewer empty legs, and lobbying over the mix of human and autonomous supply. Four different mechanisms, all pricing the same thing.
But the shape of the row is what stands out. Priced staging, free staging, off-street lots, empty-mile fees — every option on the table is an argument about where the vehicle waits. Neither side asks whether it needs to be waiting at all.
A vehicle waits because no work has reached it. That is a demand problem wearing a parking problem's clothes. An operator with one source of work has empty time by construction, and can only negotiate over where to spend it. An operator reachable by several has less of it to spend.
None of which makes staging a bad idea, or kerb pricing the wrong instrument. Empty miles will never reach zero, and Waymo is right that the honest question is how many are acceptable. It is just a smaller question than the one underneath it.
Where we sit
Kerb space is one answer to an idle vehicle. More sources of work is the other, and it is the one nobody in this argument is proposing.
Vehicles sit idle because work reaches them from one place. Omnia is built for the operator with vehicles and not enough work. Connect once, and be reachable by every source of work you choose to accept — on your terms, in your areas, at your times.