By reinventing what the car means, self-driving cars will change how cities work, too. Many “urbanists” are wary of self-driving cars, seeing them as “just more cars,” as they see cars as the root of the problems in cities.

If all that gets deployed is personal vehicles, though, only some of the benefits happen. There can be even more negatives–as robocars make car travel better and cheaper, that naturally means it will be more popular. On its own, that could mean more car-miles-travelled, more congestion and even more crashes, in spite of the abilities of the robocars to reduce congestion and crashes on a per-mile or per-vehicle basis.

A mistake, however, would be to focus on a particular solution. The right thing to do is focus on goals, and motivate others so they will figure out how to make those goals happen. It’s not usually a city’s job to innovate; their job is to protect the public interest and the interests of residents.

I’ve written about the goals, and my shortest summary of them is as follows:

People want to get everywhere quickly, reliably, safely, cheaply, comfortably and easily, any time, while not overly disrupting others. They also want to serve various specialized individual needs, from moving cargo to sightseeing. Society wants to help all the people get this, with low externalities (like emissions and street disruption,) with low public spending and lasting into the future.

The good news is that self-driving vehicles are much better behaved than the human driven cars (and pedestrians and cyclists) we are familiar with. If a rule is made or a goal is set, they will follow it. You can tell human drivers not to speed, fine them if they speed, but they will still speed. With self-driving vehicles, there will only be a handful of different systems in the city. If one type of vehicle is misbehaving, they can be talked to. All the different teams can gather in a room and work out problems and solutions.

If we propose a goal such as limiting traffic on a street network, making it happen with human driven vehicles has been complex. Typical approaches have included congestion charging cordons, transponders for managed lanes or metering lights. All of these have high infrastructure costs and face political barriers. With robocars, however, one need simply inform fleets of a new rule, and they can follow it with no infrastructure cost.

They will oppose any rule that disadvantages them, however, and that’s also the last thing you want to do, to punish the best-behaved vehicles. Robocars can promise things ordinary cars can’t promise, like never slowing down to stare at a crash, or never engaging in road rage.

Airports are pioneers in managing their traffic (though not always in a good way, since their motives are often revenue rather than improved service to the public.) They charge high fees for TNCs and robotaxis that come to the airport, while usually not charging the fees to private cars. That’s backwards, because the robotaxis can offer much more to meet the airport’s goal of traffic management. For example, a robotaxi service can promise that whenever possible, one of their vehicles will try to drop off one passenger group and pick up another on the same trip. Private cars never do that, and even TNCs have trouble, though taxis are good at it. This means twice the passenger volume per car, and should be rewarded, not punished.

Fortunately, the arrival of navigation apps, either in-car or in-phone, has changed the patterns of most drivers, especially at rush hour. Today they use them regularly and obey them almost like robots.

The new regulatory opportunity is to just define the goal. For example, “No more than 50,000 cars/hour in the CBD” might be a goal, aimed at reducing congestion and making streets more pleasant. Congestion charging is one way to attempt that, but it means the wealthy get privileged access to the roads over the poor, which is not politically popular nor aligned with public interest goals.

If you can cap the number of cars so that they don’t exceed the capacity of a road or network, you won’t get much congestion. Especially if the cars are well behaved and rarely crash. But if you can implement such a cap, the marketplace naturally solves the problem of how to move people on its own. There is no need to make additional regulations or build new infrastructure. If you count the empty seats going by in our current infrastructure, you realize there is already more than enough capacity, if we just used it optimally.

Tell people that they must do it and they will. If 4,000 wish to travel but only 2,000 cars are allowed under the cap, they won’t use today’s approach of just taking 3,000 cars and hoping traffic is not that bad. As long as it’s done fairly, people will figure out how to share the 2,000 cars to move the 4,000 people. They will pool, or set up vans and buses, or ride existing transit, or work from home, or change their routes or commute times because they simply must. The city doesn’t have to solve how this works–the people will.

The only other thing the policy has to do is try to stop them from solving it simply with money, if that’s not desired. Solving it with money does work, of course. The rich are very good at making sure they get advantages for their wealth, and sometimes they should.

Another important move to to prepare for a world where places for pick-up/drop-off are more important than parking, which currently rules our curbs and streets. PuDo is needed both for private robocars and robotaxis, TNCs and larger vehicles like vans and buses. Fortunately, as parking becomes less important, it frees up more than enough space for PuDo.

With time, other steps can arise. With PuDo replacing parking, there’s also the opportunity to put in more bike lanes, and to make high streets pedestrian-only or mostly pedestrian. Some of this already happened during Covid (though some of that was reversed, but we got a dress rehearsal.)

Over the much longer term, new development will adapt to the new mobility choices. The good news is that as we make better use of existing infrastructure, we don’t need to build new infrastructure for quite some time, until population grows so much as to overwhelm what we have. That’s a long way away, and in fact once demand is managed, we can also decide it won’t happen. When demand is managed, induced demand can be impeded to keep things under control.

What’s clear is that 21st century cities and transportation will not be like the prior century, and it’s an error to base plans and thinking on the 20th and 19th century technologies like cars, buses and trains.