Re: the Economist piece -- the conclusion you cite is very different from the one they state. After reading the article, it seems to me that the researchers concluded, after analyzing 246 street blocks that could possibly contribute to a path between two points, that closing all but a specific 6 of them would alleviate congestion. I acknowledge that the wording of the article is somewhat vague in this regard; I guess I'll just have to pick up the latest copy of Physical Review Letters to get the whole story.

by Adam on Sep 18, 2008 10:05 am • link • report

The last paragraph reads:

In Boston the group looked to see if the paradox could be created by closing any of the 246 links. In 240 cases their analysis showed that a closure increased traffic problems. But closing any one of the remaining six streets reduced the POA [price of anarchy] of the new Nash equilibrium.

That sounds like 6 of 240 made an improvement, no?

by David Alpert on Sep 18, 2008 10:08 am • link • report

Where would have these ant- McMansion people have been in the history of a growing city?

by Douglas Willinger on Sep 18, 2008 10:37 am • link • report

Trees: i think there is a law in DC protecting destruction of tree's over a certain diameter even when the tree grows on private property. That is, the tree becomes an assesst to the public after it reaches a certain size and it's destruction is prohibited. If CH CH MD had a law like this the scenario described could be mitigated (tear-down allowed but tree desetruction not allowed.)

by Bianchi on Sep 18, 2008 10:40 am • link • report

The article in Physical Review Letters is here, although you either need to pay for it or be in a place with a subscription--I suspect most University libraries would have a subscription. However, as is the case with many contemporary physics papers, a (free) preprint is available from arxiv.org here.

I haven't read it yet, but I will point out that, for better or for worse, these aren't traffic engineers doing the study but complex-systems physicists studying network theory. These sorts of studies are typically more motivated by discovering interesting properties of (mathematically well defined) networks than they are about actually studying traffic. And physicsts are known to make simplifying assumptions which, while making a complex problem mathematically tractable, also (severely) limit the validity of the results.

by thm on Sep 18, 2008 10:59 am • link • report

Well, technically, it's 6 out of 246. But in general, you're right, I did just misread the article. [/sheepishness]

by Adam on Sep 18, 2008 11:24 am • link • report

I think this traffic effect is called Braess's Paradox?

by Michael P on Sep 18, 2008 11:43 am • link • report

I'm surprised Physical Review Letters accepted the paper about Braess's Paradox. A quick look at Wikipedia (thanks, Michael) shows that many of the results in the paper are already covered by the operations research and traffic management literature. The paper's only contributions are the definition of the Price of Anarchy and some minor physics (most of which is already well-known). Their examples are similar to, and less complex than many in, the literature. Real-world examples are well-known.

Montgomery County seems to be most vulnerable to Braess's Paradox. Development along I-270 is concentrated in a linear corridor with two main roads: I-270 and MD-355. Adding a third parallel road could induce Braess's Paradox.

by Chuck Coleman on Sep 18, 2008 8:18 pm • link • report

Chuck, is there a good paper that lays out what are the necessary or sufficient conditions for the paradox to occur?

by Michael P on Sep 18, 2008 8:37 pm • link • report

http://tigger.uic.edu/~hagstrom/Research/Braess/ has a lot of results in several papers. I haven't looked at all of the material, so I can't really answer your question. The papers also contain lots of references.

Another point I want to make, in light of Braess's Paradox, is that simulation studies are necessary to traffic optimization. I'm sure a lot of surprises are lurking out there. Who would have thought that closing 42nd Street in Manhattan would reduce congestion? Mass transit can be added to the simulation studies to provide better estimates of congestion reduction and mass transit demand. When this is done, adequate modeling of induced demand is necessary. Including mass transit in simulations can provide far better cost/benefit analyses of transportation changes (improvements and street closures) than are currently avaiable. Maybe I should write a paper. ;-)

by Chuck Coleman on Sep 18, 2008 9:47 pm • link • report

I would like to see how traffic throughput (measured in people) increases when the occasional street is repurposed from car-bike-bus-taxi-emergency directly to bike-bus-emergency or bike-bus-taxi-emergency. Seems like it could provide many of the benefits of both street closure and BRT without cutting off parking alleys or significantly impacting traffic, if you time the lights right. It would reduce bus stop backups on surrounding streets as well.

by Squalish on Sep 19, 2008 6:26 am • link • report

Closing some street options is all well and good in normal conditions but when that once in a blue moon nightmare condition occurs (like that freezing rain debacle last winter closing the Springfield interchange) all those closed roads turned into cul-de-sacs just concentrated the misery to truly epic proportions. Taking two hours to drive a couple of miles really introduces you to all the road options that once were when you are carefully studying your GPS for other options.

by NikolasM on Sep 19, 2008 11:07 am • link • report