[cortesoft]

This sort of traffic jam dissipation is only applicable to phantom traffic jams. Here in Los Angeles, traffic jams are caused by too many cars being on the road.

At some point, every available square foot of roadway is filled with a car. AI can help limit the amount of extra space taken up by air, but that will only moderately increase the number of cars that it will take before traffic hits. That number of cars will be reached, easily.

[Retric]

Yea, I remember getting stuck in a huge traffic jam and feeling dumb for driving on 66 at rush hour. Until, I remembered it was Saturday. In northern VA traffic is often bad at from around 5AM to 10PM. But, we can have traffic jams at any time day or night 7 days a week.

Granted, bad night time traffic is often aided by construction. But, that does not help when your parked on a freeway at 3AM.

[jandrese]

I've been stopped due to congestion on 66 at 1AM in the morning. That road is an outright disaster.

[jemfinch]

> At some point, every available square foot of roadway is filled with a car. AI can help limit the amount of extra space taken up by air, but that will only moderately increase the number of cars that it will take before traffic hits. That number of cars will be reached, easily.

The thing to keep in mind is that AI can limit the extra space between the cars as well as increase the speed those cars can safely travel. The cars will spend substantially less time on the road because they'll be going faster. Double the speed and you've converted a 60 minute commute into a 30 minute commute, doubling the number of commuters the roadway can support even if you don't adjust inter-car distance art all.

[jeremysmyth]

Don't forget Amdahl's law though. You might be able to optimize away the reaction time with good AI, but you can't optimize away mechanical factors like engine failure, tire grip, and road surface friction. This affects the maximum speed and traffic density just like a human's reaction time.

This means that if one car in a densely packed column fails (blowout, engine seizure, ...) then the car immediately behind it is still subject to its mechanical ability to avoid a collision (swerve into empty space or brake to a hard stop at speed) no matter how fast it reacts.

[cft]

AI will eliminate truck night stops, allowing them to be driven 24/7. Hopefully more trucks will be moving at night. That should reduce the jams.

[paulmd]

Increasing the efficiency of a process also tends to paradoxically increase its utilization (Jevon's Paradox). So it's entirely possible that automated trucking will actually increase total traffic volume.

Still though, even a few vehicles "eating traffic waves" can break up a traffic jam. I've heard the number put at 3-5% of traffic being self-driving would prevent phantom traffic jams.

So overall I would say that self-driving cars could help reduce phantom traffic jams, but they may actually increase real traffic jams caused by overcapacity.

[maxerickson]

Tariffs. If people want access to a congested zone, charge them for it. Dynamically adjust the tariff based on the level of congestion.

Of course people hate this and wail that it isn't fair, but it is among the more fair solutions to the immediate problem and inflicting traffic jams is probably not a good way to address economic inequality.

[SilasX]

Agree. Tolls (assuming they means the same thing as tariffs) can also create the critical mass necessary for commuter bus lines to work, thereby allowing people to spend only a little more than they do now on commuting but get there faster since the roads are cleared. (The per person effective toll would be lower on a bus, even if they are charged by size.)

But they would have to be much higher than in most proposals, and they should be much lower or non-existence during hours when they're not choked despite the zero price.

[maxerickson]

Yes, toll is the better word. For a system with dynamic pricing, once you have a critical mass of vehicles with some sort of system to notify the drivers, you can even set the toll experimentally.

[aab0]

Tolls could even be done after the fact. Imagine a special tax on self-driving cars and trucks: 0.0x cents per mile if driven between 9AM and 5PM, based on mileage logs and all miles assumed to be driven then. If a corporation wants the discount, it can produce GPS logs showing its vehicles were logging all that mileage during off-peak hours.

[rando18423]

So the poor person who needs to get to the doctor should take the slow lane, and the rich person who wants to get to the next bar faster should take the fast lane? Public utilities INTENTIONALLY are not and should not be run according to the principles of profit maximization.

[SilasX]

The poor person should take the bus, which could actually be a fast option if the streets weren't choked at critical times. That's what a sane utility looks like. (If they're not choked, no congestion charge.)

I'm not a big fan of the mentality that the transportation system should suck just so we can avoid the rich have better options than the poor. Taken seriously, that would mean a ban on air travel because the poor can't have as much.

If you meant emergency vehicles, those always have priority :-p

[SwellJoe]

Not only that, AI will eliminate huge amounts of wasted road even during peak times and with traffic that can't be scheduled for off hours (like people going to work/school). A networked self-driving car can know where the traffic is and route around it. It can provide tools to everyone (from city planners to individuals) to maximize the availability of resources and minimize the time wasted on waiting. Red lights can cycle at the most efficient rate at all hours of the day, reversible lanes (like many large cities have on their primary highways) can be closed and opened on a dynamic schedule rather than on the preset schedule. When all cars are AI-driven and aware of all other cars, they can drive at exactly the right speed, allowing faster merges and exits, etc. Humans are wildly variable in their response time and unpredictable in their behavior in so many situations on the road, that replacing them seems likely to increase the capacity of roads by a dramatic amount. Maybe doubling? Maybe tripling? I don't know, but it will be a huge difference.

[j_s]

Yes, all (but one?) lane going in the "to/from work" direction based on traffic needs will make a huge difference!

[SwellJoe]

Theoretically, one doesn't even need directional lanes, anymore. You just have a road, and cars go where they most efficiently fit; lanes won't have any meaning anymore, just like most road signs won't have any purpose. If every car is communicating with every other car, and every car knows where it needs to go and how to get there in the shortest time, it can look like utter chaos to the human eye, and still be entirely safe (assuming we go to great lengths to insure reliable software and hardware).

Of course, I expect it to take decades for us to reach that point, which is really unfortunate. That's a lot of senseless auto collision deaths due to human error while we wait for the future to arrive. If governments were capable of looking forward, at all, they would recognize the massive savings an all AI road represents, both monetary and health, and would be pushing hard for it (both enabling private industry to implement it and guiding public funded research).

[cortesoft]

Not sure how much you have driven in Los Angeles....we also have traffic jams at night.

[ec109685]

If the cars were able to travel faster (by organizing themselves to flow more optimally), they wouldn't all need to be on the road together.