I've always been fascinated by traffic. (I grew up in LA, so I had plenty of time to indulge my interest.) City streets are a complex system in which seemingly insignificant changes - a broken street light, a stalled car, a poorly designed highway merge - can have dramatic consequences. In this sense, it's a useful metaphor for all sorts of intricate systems, from gene regulation to neural networks.
Perhaps the single greatest mystery of my childhood was this: Why do freeways get clogged when there isn't an accident? I would fantasize on my way to elementary school - we would often be sitting in gridlock on the 101 - of getting a loudspeaker and ordering every car to accelerate. I figured if we all started moving at the same time and at the same speed then the traffic would disappear.
My latest Seed column - on newsstands now - explores some new research on the origins of traffic jams.
A few years ago, the Swiss economists Bruno Frey and Alois Stutzer announced the discovery of a new human foible, which they called "the commuters paradox". They found that, when people are choosing where to live, they consistently underestimate the pain of a long commute. This leads people to mistakenly believe that the McMansion in the suburbs, with its extra bedroom and sprawling lawn, will make them happier, even though it might force them to drive an additional forty-five minutes to work. It turns out, however, that traffic is torture, and the big house isn't worth it. According to the calculations of Frey and Stutzer, a person with a one-hour commute has to earn 40 percent more money to be as satisfied with life as someone who walks to the office.
The reason long commutes make us so unhappy is that the flow of traffic is inherently unpredictable. As a result, we never adapt to the suffering of rush hour. (Ironically, if traffic was always bad, and not just usually bad, it would be easier to deal with.) As the Harvard psychologist Daniel Gilbert notes, "Driving in traffic is a different kind of hell every day."
But why is traffic so unpredictable? After all, the number of cars on a highway during a typical weekday rush hour is fairly constant. And yet, even when there are no accidents - and most traffic isn't caused by collisions - the speed of traffic can undergo dramatic and seemingly inexplicable shifts.
The key to understanding traffic jams is something known as "critical density," or the number of vehicles that any road can efficiently accommodate. When this threshold is crossed - when too many cars are trying to cram onto the same six lanes of asphalt - the flow of traffic starts to breakdown. At this point, congestion becomes all but inevitable, as even seemingly insignificant events, such as a single driver tapping on the brakes, can trigger a cascade of brake lights. That's when the highway becomes a parking lot.
While the concept of critical density has been repeatedly demonstrated using computer simulations - drivers are surprisingly easy to model as a system of interacting particles - it wasn't until last year that this theory of traffic was experimentally confirmed. A team of physicists at Nagoya University wanted to see how many cars could maintain a constant speed of 19 mph around a short, circular track. It turned out that the critical number was 22: once that density was reached, tiny fluctuations started to reverberate around the track, which caused the occasional spontaneous standstill. As the scientists note, this is actually a pretty familiar scenario for particle physicists, who are used to studying phase transitions, such as the transformation of liquid water into solid ice. In this case, the critical threshold is temperature, which triggers clusters of molecules to slow down and form a crystal lattice, which then spreads to nearby molecules. A traffic jam is simply a solid made of idling cars.
I then explore what we can learn about traffic from ants. And if you haven't read it, I highly recommend Traffic, by Tom Vanderbilt. Finally, here's a cool video taking you behind the scenes at LA Traffic control.
The driving test is designed to see if you know how to operate a vehicle. It has nothing to do with knowing how to drive. Most people don't and they have problems changing speed and direction competently. In addition, our road systems are not designed to facilitate efficient driving. On top of that, driving is an extremely selfish pursuit and it shows in how people behave in traffic.
That pretty much sums it up!
Good blog. I look forward to reading your article.
I've always looked at jams-with-no-accident like this: The car in front of you slows down. You have 3 poossible options: #1 You do not slow down as much as he does - therefore there is now an accident. #2 You slow down the same exact amount that he does - terribly unlikely. #3 You slow down more than he does - the most likely scenario. Now ripple that back 100 cars, and someone finds themselves parked.
I remember reading somewhere the effect a "good agent" can have on congested traffic. A vehicle that accelerates slowly allowing space to develop and brakes gradually to avoid an absolute stop can create a more normal, albeit still slow, flow of traffic behind itself. Apparently the technique dissipates the rolling waves of start/stops and creates a smoother flow. One of the first lessons I learned commuting in LA was to look far ahead for the approaching wave of red brake lights on the freeway.
My dad found this nice solution to avoid traffic. He drives to work on Monday. At the end of the day, he lives the car at the train station in the city where his company is located, and comes home by train. He uses the train all week, and the car just for the 5 mins drive from the train station to his company, and then drives back home on Friday evening. As he is very busy, he has no time to make new friends outside work but he started to see always the same people on the train and made a lot of friends there.
Anyway, I feel that now that we can listen to a bunch of audiobooks and podcasts on our iPods, traffic is not as bad.
An interesing piece about traffic.
Very interesting stuff. The outstanding thing about your excerpt from your column to me was the simple equation of happiness to things other than material wealth. I particularly liked the phrase "a person with a one-hour commute has to earn 40 percent more money to be as satisfied with life as someone who walks to the office." As an avid biker and bike messenger I've thought I was gaining more through my lifestyle choice than the obvious benefits of not spending money on gas and staying in shape, but it's nice to see it articulated even further through Frey and Stutzer.
There's a great video showing the "shockwave" effect of traffic jams here: http://www.youtube.com/watch?v=Suugn-p5C1M&feature=player_embedded
Since I discovered how these "phantom" traffic jams form, I've tried to drive more moderately with less braking and accelerating. Not only is it a great way to burn a lot of extra fuel, it can create a parking lot behind you.
I think Parris's hypothesis (in comment #2) was on the right track but missing something very important - the concept of incomplete information. Let's say you are driving the first in a line of cars. When you step on the brakes, you know how much you're slowing down because your brain formed the plan to slow down, and because you have a speedometer. But the second driver doesn't have any of that information. They must be prepared to react in case you stop completely (for instance, if your engine failed). That makes them slow down more than you. Then the third driver has to slow down more than the second driver for the same reasons, and so on until the ripple effect carries back far enough that someone does have to stop completely.
In other words, if you knew the car in front of you was going to slow from 60 miles an hour to 40, you wouldn't need to slow down any more than that. But they could be slowing all the way down to zero, so you have to slow down more to compensate until you can tell they aren't.
In Bangkok, the traffic lights last around 5 minutes. So a long queue of cars forms itself, but once light switches all the cars start at the same time. It's amazing.
Someone told me it was the same in NYC.
Optical Illusion As Traffic Control
Apparently stripes on the roadway cause drivers to slow down. In one case on a California freeway they worked too well and caused traffic jams.
I am a PhD student in Chemical Engineering and my research is on the phase behavior of colloid particles. I think a better analogy than liquid water would be my colloid particles. The critical number of cars on the road is more like the volume fraction of particles in a solvent rather than the temperature of water. Changing temperature changes the interaction energy between the particles whereas changing volume fraction just changes the density of particles. This seems like a more apt comparison to cars on a highway.
But either change can cause a phase transition from liquid to solid-like (be it crystal, glass, or gel). And people are more likely to be familiar with the water-ice transition than colloids, so the water analogy probably makes more sense for your article.
Love the blog by the way. Always makes me think.
"I remember reading somewhere the effect a "good agent" can have on congested traffic. A vehicle that accelerates slowly allowing space to develop and brakes gradually to avoid an absolute stop can create a more normal, albeit still slow, flow of traffic behind itself."
That's exactly what I do. Good to see that there are other people doing this. But often all the effort is undone by those morons who think they have to change lanes every ten seconds (maybe to feel faster by "doing something") and fill up the "buffer" I created.
Anyone, does this clumping effect(spontaneous traffic jams) shed light on matter clumping as the universe expands and its apparent acceleration? Is implication the effect is statistical and not behavioral?
In case you haven't seen it, Chad has a post about the article: