Cool experiment but somehow I feel like I intuitively knew this already. Also, this:

http://trafficwaves.org/trafexp.html

By the way, did you know that the other lane really does move faster?

I thought this was great. I've watched and pondered these things for a while. At some point I want an app to collect GPS data from my phone so that I can try various routes and map out avg speed between home and the office for different strategies.

One thing I've noticed on freeways is that off ramps generate 'waves' because people move from the far left (HOV) lanes to exit, and the right lane gets less dense. Then the following on-ramp has people braking as folks merge on, and as they try to move over to the HOV lane.

It would be awesome if there was some way to have a third dimension, lift up and then exit.

There's an interesting book on the topic, focusing on the way human behavior affects traffic patterns. It's similar to other behavioral anthropology books like Dan Ariely's 'Predictably Irrational'.

Tom Vanderbilt - Traffic: Why We Drive the Way We Do (and What It Says About Us) (http://www.amazon.com/Traffic-Drive-What-Says-About/dp/03072...)

Are most traffic jams caused by this phenomena, or simply because miles ahead of you there are 4 lanes of cars trying to take a 2 lane exit?

This, ideally, shouldn't cause an issue. The cars shouldn't start to decelerate until they are already out of the flow of the thru traffic. Of course, you have the people who decide they want to be on that exit at the last possible second, which is what you might be referring to.

Now I have seen some exists where the light at the bottom is really poorly timed, causing traffic to backup unto the interstate, which causes backups even for the thru traffic…

A commenter on the article does bring up an interesting topic, that I've noticed in my rare times driving on the interstate, that people don't use the cruise control. This annoys the heck out of me.

Why shouldn't that cause an issue? If you have 1,000 cars all making the same trip from the suburbs to downtown, and 90% of the trip is on a 4 lane road, but the last 10% is on a 2 lane road, then the 2 lanes are your bottleneck, and assuming the speed limit is constant, the total trip time of all 1,000 cars will be the same as if the entire trip was on a 2 lane road.

Obviously, real road systems are more complicated, with more cars merging onto the main road along the way, others exiting at different times, etc. But I think the general idea holds true: when thousands of cars are all going from suburban areas into a downtown area, there will inevitably be a bottleneck close to the downtown area.

This could make the lanes of slow traffic infinitely long in your "ideal" situation, which means there would be cars entering the highway with the same destination that never had the choice to get out of the flow of traffic before decelerating.

Earlier discussion: http://news.ycombinator.com/item?id=3000305