Game theory: pedestrians versus autonomous vehicles

Originally published at:


The kids are gonna love this game!


May you live in interesting times :smiling_imp:


Yeah but when there’s a critical mass of self driving cars a pedestrian will be able to j walk without slowing traffic or fearing getting hit by using school of fish like behavior.


Game theory is like the mathematic formalization of “Christ, What an Asshole”.



Interesting article about Otto self-driving trucks in MIT Technology Review this month -

The human driver piloted the truck to and from the highway the old-fashioned way, because the technology doesn’t drive on small rural roads or in cities.

In related interview, Otto cofounder/president Lior Ron says:

“It’s not going to happen overnight. It’s going to happen in very discrete baby steps,” he said.

For instance, he expects to see fully autonomous vehicles first traversing city streets during the wee hours of the night when not many pedestrians are around, then more often during the day and in various driving conditions (which can still be quite tricky for self-driving cars and trucks to safely navigate), and so on.

So, they are headed that way, and would solve these problems incrementally, advancing to more complex situations as the solutions settle in. Seems really reasonable.


Remember that segment Bill Hicks did about the LA Riots and the pedestrian right-of-way law? Oh Bill, where are you when we need you…


I was thinking about autonomous cars yesterday. I bet people will find all kinds of “interesting” ways to mess with self-driving cars in a mixed environment of self-driving and human driven cars.


I think places with decent urban infrastructures will have an easier time with pedestrian and car traffic. Places with more hectic environments and less regulated traffic laws will be the real challenge. Go to 3rd world countries and tell me that an autonomous car will work great there.

However i can’t see how local governments in US cities would pass up the chance to issue tickets via facial recognition to jaywalkers. They already love raking in money by setting up cameras at intersections, i don’t see why it’d be any different in the future with foot traffic if it were to be a problem.


Like any rural area in the USA


My kind of game theory.


Roger Zelazny and Fred Saberhagen riff on this idea a little in the middle of their 1982 SF novel Coils, which is nicely prescient in a few minor ways and quite a lot of fun to boot.


This is why we can’t have nice things.


This is what we need flying cars for.

1 Like

“Finally, according to a human driver scenario, the slower travel time incurred by using a self-driving car would outweigh the benefits of a passenger lifestyle.”

This presumes that the travel route is someplace with a lot of pedestrians. For most US commuters, I rather suspect that’s not the case. People still aren’t going to be running across highways and freeways, even with autonomous vehicles. Even on non-highway streets, so much infrastructure and urban design are already pedestrian-hostile that there are very few people walking, much less jaywalking.

Well, human-driven vehicles don’t work there, unless you allow the definition of “work” to include “mass fatalities.”

(Some) Rural areas in the USA may be 3rd world in their infrastructure/living standards, but you don’t have anything remotely like this:


Personally I can’t wait until self driving cars are in the majority. Just think what Reclaim the Streets could do.




Self-driving cars can only work because there is infrastructure, which implies rules. You can’t have rules of the road without actual roads.

I think making self-driving cars work with any existing infrastructure for human-driven cars is a fool’s errand. A separate infrastructure has to be rolled out to support self-driving vehicles. Otherwise, you have self-driving cars adapting to an environment that will become increasingly irrelevant as autonomy takes over.


No, in London, there are no ‘jaywalkers’. You’re allowed to cross the road whenever and wherever you want.


You end up with a serious boot strapping problem there. The first self driving cars are likely to be more costly (they have extra stuff!), and even if not they will be new cars. Now add onto that that they can only drive on new roads (or roads reclaimed from former person-driven roads) and you get a very limited use. Who wants to pay extra for a new car that can only self drive somewhere they don’t go? Then because people don’t buy it you don’t get economies of scale, so the price never comes down (in fact it doesn’t go where it was originally predicted).

You might be able to get it to go somewhere if rather then merely allowing self driving cars the government mandated them and started taking roads away from person driven cars, but I don’t see why they would do that.

On the other hand the harder programming job of making self driving cars interoperate with person driven cars on existing roads gives you a way easier economic situation. Buying a self driving car gets you a car that can drive itself. You get time back (note: I’m referring to the fully self driving cars, not something like the Tesla ones that require a person to remain vigilant even after lulling them into a false sense of security). If that is thing that interests you, then you buy it. It doesn’t have much of an effect on others. If safety promises are correct it actually makes other people slightly safer (and yourself a fair bit safer). If they are correct insurance companies start giving them a discount (because one of them surely will, and the rest will have to follow or lose customers). They may eventually become cheaper then normal cars once you take the insurance discount into account.

So do it the hard way, the easy way is too hard.