Interesting read. My dad is a traffic engineer, and has studied these phenomena and discussed them with me.
I was never able to understand how these waves of traffic started without a wreck, or lane reduction, until one day in 2007 when I witnessed one happen. It was a near miss of one guy who went into the median, almost hit the center wall, then swerved back into traffic and went an extra lane and a half. No one wrecked, and it all happened at speed so no complete stopping or significant speed reduction. But the momentary loss of capacity started to propagate and accumulate.
The wave pattern is a very real phenomena and there are many ways to predict the longitudinal speed and amplitude (the speed of cars in the wave) on various roads, and the driving habits of the people in that region.
Some of his philosophies are not quite correct. His merge method of leaving gaps, and letting the multiple minor road entrants before you is worse than close cars going 1 and 1. By allowing more space between cars, you lower the traffic density of the road, and thus back up traffic behind you. The merge works better, but the arterial has significantly less capacity. In general the best way to merge is to limit the minor road traffic influx, and mandate gaps in their traffic (with ramp meters), and not regulate the main arterial flow. It is slightly worse for the people getting onto the main arterial, but much better on aggregate. Once the arterial is grid locked, it is best to have a full unhindered queue, and merge 1 and 1 at the end of the ramp. And merging early (not going to the end of the line) is actually worse for traffic (assuming you are not clearing a lane that will exit and can alleviate traffic by removing cars from the road), than merging when you are forced to do so.
His 'wave smoothing' will work in the immediate area to smooth the wave, but will not make transit times any faster, and can make traffic worse. You do eliminate or reduce the traffic wave, but lower average speed, thus not decreasing travel time. The wave will still propagate backwards, just with less amplitude, but a longer wavelength. They can also increase traffic behind the person, or initiate another traffic wave.
In general the only way to make traffic better is to increase the immediate capacity of the road. That means driving as fast as possible, with as little headway, safely and smoothly. Because if you start driving too fast, or too close you may cause other drivers to hit the brakes or follow further from someone. But in gaps are bad, speed is good. If the road isn't close to saturated you don't really have to worry.
For a general structure of traffic you really have to think about ideal road capacity, versus actual road capacity. Road capacity is normally measured in vehicles per hour. It has 3 components, number of lanes, average speed, and average headway.
In general a road will work fine as more cars are added and it approaches ideal capacity. As you get closer to ideal capacity average headway shrinks until you hit ideal capacity. As more cars are added, headway can not shrink, and cars behind this section of road have to lower their speed. This lowering of speed lowers the capacity of the road, and thus compounds further and further back until you start having people having to completely stop on the interstate. So when a road is close to ideal capacity, a single person scrubbing 1-2 mph of their speed can cause people hours from then, and miles behind them to come to a complete stop.
This is also associated with other problems where people will slow down in areas where there normally is traffic, even though they need not slow down. As they expect traffic they will slow slightly, starting up traffic behind them. This is why you get the exact same traffic patterns, on days with lighter but still heavy traffic.
Surface streets are much harder to model, as they have signals that control flow. And this control is much more prevalent than the emergent traffic phenomena and many times the distances between flow control devices is not long enough.