Talk:Braess's paradox

Why is adding a crossroad considered "adding capacity?" The phrase "added capacity" to me sounds like adding a third route of equal capacity to the first two, which would indeed solve the problem. The phrasing "adding choice to the system" or something similar would be more accurate. --Monguin61 21:13, 10 December 2005 (UTC)

This is actually pretty good but "For a better explanation, please see this link:" sounds, well, too humble. Either it's been explained adequately in which case there's no need for apology, or it hasn't, in which case, skip the apology and expand the article.radek 06:58, 12 March 2006 (UTC)

Add: How far from optimal is traffic at equilibrium

I think the following information should be added:

- If latency function are linear then adding an edge can never make total travel time at equilibrium worse than by a factor of 4/3

- At worst traffic in equilibrium is twice as bad as socially optimal

--Grondax (talk) 15:33 11 Novermber 2008 (EST)

The example is flawed.

This example is flawed..

Suppose I'm a 'rational' traveler, and the third route gets added. I know that everyone is taking the new route, thinking it will be faster, so I take Start->B->End instead. Start->B takes 45 minutes. B->End takes 0 minutes, because everyone else is trying the new 'faster' route. I get home in 45 minutes. Everyone else eventually realizes that this new route isn't any faster, and everyone ignores it, and eventually I suppose it gets closed to reduce maintenance costs.

And this example is flawed, and I dare say the entire theory, because people will COMMUNICATE with one another. —Preceding unsigned comment added by 71.193.207.197 (talk) 00:41, 3 May 2008 (UTC)

I agree the example is flawed, but I don't think its necessary for people to communicate with each other to find the optimal solution. If x people take Start->A->End, y people take Start->A->B->End, and z people take Start->B->End, then (x, y, z) = (1301, 896, 1803) yields an optimal solution of 63 minutes on average and 71 minutes in the worst route (the z route). Given enough time and sensitivity, isn't it likely that uninformed drivers would naturally arrive at this equilibrium? --Beefyt (talk) 17:52, 4 September 2008 (UTC)

Only if the z drivers were unaware that they could get from Start to B in 21.97 minutes by taking the route via A instead of going directly to B, which takes 45 minutes. Similarly the direct route from A to End takes 45 minutes but going first to B makes it 26.99 minutes. Once they realize this the drivers will start switching their routes in favour of the shorter ones, which will in turn become longer due to the increased traffic. We eventually end up in the situation described in the article. In other words, your equilibrium is unstable, given the assumption that drivers can choose their route freely and that they are minimizing their own time spent on the way. I can't see any error in the example. 81.83.2.11 (talk) 21:42, 30 November 2008 (UTC)