[Ziggy]

It's not hard to think of a mathematical model to demonstrate how this works. Then again, I'm sure this has already been done, and is probably on wikipedia.

Here's the wikipedia article about it: http://en.wikipedia.org/wiki/Braess_paradox

That's a very good point.

With regards to the use of highways, I suspect people endure a slow average speed on the highway as it is the better (relatively speaking) option. While you might only average 40 km/h, I suspect travelling along Finch or Steeles, or whatever, would yield an average speed much lower than that. Especially if you allow for innumerable traffic lights, construction, etc. Also worth considering is that for every day you get stalled in traffic on the highway there is the potential that the next day you might manage a faster time. Whereas with lesser roads, your travel time might be consistent but uniformly slower.

Yeah, that's probably true most of the time in reality, but if Braes's paradox is happening, then removing one of the linkages, say a highway, will actually result in a higher average speed, even with the same number of vehicles travelling through the network. As far as I understand it, it basically happens because of how the congestion balances out. If all the high-speed routes are connected together and become congested, then the whole network will jam up and everyone will travel at a slower speed. If the network is more disjointed, then each individual path through it will have to go through some slow bits, and some fast bits, but the overall average speed will be faster.

 

[xtremesniper]

"The more uncomfortable the driver feels, the more he is forced to make eye contact on the street with pedestrians, other drivers and to intuitively go slower,"

Bah, in my area we already have problems with people who aren't comfortable driving even though they insist on driving everywhere anyway. These are the people who go 50 in a 70km/h zone and cause more traffic delays and potential for collisions due to lane changes than anyone else.

This model may work in other areas of the world, but it wouldn't work in the GTA. Not without some major changes to everything, including licensing process. There are far too many poor drivers on the roads.

That said, I love the idea of tearing down the Gardiner and replacing it with a 5 mile long park. =P

 

[TKTKTK]

It's not hard to think of a mathematical model to demonstrate how this works. Then again, I'm sure this has already been done, and is probably on wikipedia.

I understand how the Mathematical model works - I'm curious to see it applied to the city. What brought this up wasn't an abstract model - it was Jarvis Street in Toronto, and whether its extra lane is speeding traffic up, or slowing it down.

 

[afransen]

Well, that would require a model of the road network, traffic sources and sinks, and determining the delay functions of each link in the network. Then assuming drivers are self-optimizing (time only) and that there is perfect information, it could be ground out. Honestly, I'm surprised that this might not already be done as part of traffic analysis. Then again, I guess the math is somewhat new and only rose to prominence with the advent of computer networks which are closer to the abstract model. I wonder if anyone in mechanism design is working on applying it more practically to private car transportation.