I've had numerous debates about this, I've asked someone who works for my local municipalities traffic office about simulations regarding this (though she didn't have the correct simulators to do it, just for intersections).<p>I know the theory, but people frequently and incorrectly state that emergent traffic jams are proof that leaving large gaps to smooth traffic behind you improves traffic flow. That to me this is an unacceptable leap of logic. Why are their no contrasting simulations showing traffic flow with and without these "smoothing" agents? Creating a simulation that has rolling traffic jams, introduce agents that disperse the traffic jam (can you actually disperse these traffic jams anyway? Again, not demonstrated in the story.) Has throughput increased? It's not a given that it has.<p>My concern is that the people who tried to do what I am saying only achieved negative results, so they did not publish. Regardless of my concerns, I believe this theory might be true. Which is why this is so frustrating!