When you sit in traffic, you realize how most people are driving solo. It’s horribly inefficient, but there’s no reason to believe people will ever double up or that traffic will ever get better. That’s just not what having a car is about. The whole point to having a car is that it provides the freedom for you to go where you want on your own schedule. No need to ask someone else for a ride or to schedule your outings around another person. Your car is your earth capsule; a climate-controlled extension of your home base.
It seems like a certain population density is required to make mass transit a rational choice. It has to be prohibitively expensive in terms of travel time and money to encourage people to use mass transit options. In places like New York City, it is rare and expensive to have dedicated parking. Likewise, in such a dense city environment, it often does not make sense to drive yourself. As a result, people take the subway or a cab rather than deal with the hassle of driving.
In almost every other part of the country outside New York City, it makes sense to drive yourself as most mass transit options are simply worse than driving. In the future, I think the solution to horrible traffic will be cars controlled by artificial-intelligence and demand-based toll pricing.
Automated cars will allow more efficient driving and a more pleasant commute. Cars controlled by AI will be able to drive much better than human drivers and they will be able to interact with one another to avoid road hazards via ad hoc networking or a centralized traffic control system. With AI drivers, cars will be able to travel at a high speed and will be able to coordinate to maintain efficient traffic flow. Just like timed traffic signals facilitate traffic efficiency by coordinating access and cross traffic, cars controlled by AI will be able to work together to maintain the most efficient flow of traffic throughout the system.
Demand-based toll pricing is another way to improve traffic flow. As our infrastructure needs increase, governments will struggle with providing and maintaining the roadways. When you add new roads or highways you would think this would reduce traffic congestion, however traffic congestion quickly increases to use up the newly available capacity. In other words, adding new roads does little to reduce traffic congestion. The only real answer is to control demand and to encourage more efficient use. Gasoline taxes are a blunt instrument in that people who contribute little to traffic congestion (for example, people who drive within their immediate area) bear the same tax burden as commuters who clog the highways trekking into the city from the exurbs. The most efficient way to control traffic and to pay for the roads is to encourage the expansion of tollways, especially demand-based tollways. For example, most tollways already support toll tags or some similar way of automatic payment. Why not alter the toll prices based on congestion? In other words, the more traffic, the higher the tolls. Late at night, when there is no congestion, allow free travel. During rush hour, keep raising the toll amount until congestion shrinks to allow a set average speed of say 50 miles per hour.
While the highway system was built as a common resource it is indisputable that some people use this common resource more than others and that some contribute a lot more to traffic congestion than others. It is reasonable to expect heavy users to bear more of the cost.
Oddly enough I was thinking about this the other day as I was struggling to get home through heavy traffic. I read an article a few years ago that claimed the cause of traffic congestion is not “too much” traffic – the theoretical maximum traffic capacity of our freeways is much higher than we see in practice – but instead “overbraking”. This explains how traffic grinds to a halt without any apparent cause (e.g. an accident). How it works is that someone changes lane to merge or get a better position, and the car behind them brakes slightly to accommodate. The car behind them brakes a little bit more, since drivers always overcompensate, then the next car and so on until eventually so much speed is lost that traffic starts to crawl. The solution, as you say, is computer control. Computers will not overcompensate, and through communication with other vehicles and better reaction times can travel a lot faster and closer to other vehicles. It’s claimed that computer control could at least triple the capacity of our roads … the question is how do we get there? Until most cars are computer controlled we will see little benefit from the technology, so it’s a tricky economic proposition that will probably need government subsidy.
Keith
I think we’ll see this in another 20 years. First the cars will “assist” drivers, alerting them to potential collisions or road hazards and braking if they get too close to a stationary object, etc. Then we will gradually cede more authority to our guidance systems until we’re just passively riding around. I think the role for government is in encouraging this technology (a la the DARPA challenges) and helping to build the infrastructure to allow this technology.
The problem, as always, is transitioning … how does a roadway work when you have a combination of the “new” automated cars and “old school” human-driven cars … this mix will have to be addressed. We can’t simply switch over all of the sudden.
Perhaps, it’ll start with special lanes (like HOV lanes) that are only for these new auto-automobiles? Or, maybe it’ll start with cabs, buses, or other non-personal vehicles.
Regarding what Keith was talking about … if you watch time-lapse footage of heavy traffic, those transient moments of stoppage or near-stoppage migrate “upstream” even while all the traffic is moving “downstream”. This produces an accordion effect or a “wave”. Definitely not efficient…but this type of pattern is seen in nature, so there’s something fundamental about it.
You can see this phenomenon well in some of the footage in the film Koyaanisqatsi.
Brian: I think AI will have to take into account destructive or unexpected behavior like vehicle failure, etc., so it will need to be robust enough to work around crazy human drivers. We would definitely see mixed systems. In one sci-fi book I read recently, all the taxis were driven remotely from call centers. Something like that could be an alternative to AI.