When I was a child, my father would wonder, often on the way to our annual visit to Disneyland (beginning when the park opened in 1955), why the traffic on a freeway unobstructed by accidents or construction could nevertheless sometimes slow down and even stop while in the course of its general movement.
This was bewildering to me also, and it was many years before it occurred to me what was going on. The cars were acting like molecules in a gas, and the tightness and looseness of their packing, or the swiftness or slowness of their movement, was responding to the equivalent of 
pressure waves that were passing down the column of traffic. Pressure waves, which occur in gasses or liquids, feature alternating bands of high and low density areas, with the molecules moving more slowly in the former, more switfly in the latter, areas. (See the discussion of pressure and shear waves in relation to light.)
When traffic is diffuse enough, it does not seem to behave in this way. Thus, there is a threshold of density that must be reached before the phenomenon emerges. At first we see "slow-and-go" traffic, where the column seems to unaccountable speed up or slow down, and then "stop-and-go," when the slown down culminates in actual stoppage. As densities increase further, the traffic behaves as an increasingly viscous liquid, approaching a solid. The low density, speed up zones may thus come to disappear, and the whole column of traffic becomes a kind uniform, slowly moving paste.
I was reminded of this dynamic under rather different circumstances, driving down I-5 back to Los Angeles after spending Thanksgiving in Northern California in 2010. The traffic was not really experiencing densities of the familiar slow-and-go or stop-and-go freeway traffic, but there was a marked new dynamic that emerged in the relationship between the number 1 and number 2 lanes on the two lane interstate highway.
The number 2 lane is where the trucks drive. In California, they are supposed to be limited to 55 mph. Other traffic on I-5 can drive at 70 mph. The trucks usually drive considerably faster than 55 mph, and the other traffic is often going 80 mph or faster. I've seen people getting tickets, and California Highway Patrol (CHP) cars are often lying in wait behind bridge abutments; but the overall traffic flow seems little affected by the threat of speed limit enforcement. The long straight highway in the southern San Juaquin Valley (just "the Valley" to Locals, as in American Graffiti [1973]) makes for a great raceway, although the characteristic winter fogs of the Valley ("Tule" fogs, from the tulares of the ancient damp meadows) can trap the unwary into catastrophic multiple-car pile-ups.
In most conditions, the flow of traffic in the number 1 lane can be suddenly slowed when one truck pulls out to pass another. This is a constant threat, and one must be prepared for the sudden slowing that results, until the passing truck is back in its proper lane. If a truck driver misjudges a grade (as I-5 passes over a few hills at the edge of the Valley), or there is an unwillingness in the other truck to be passed, it may be a somewhat extended process before the number 1 lane is clear again. Otherwise, one mostly only need be concerned about the occasional fool who drives slowly, or merely at the speed limit, in the number 1 lane. Unless they pace an equally slow truck to their right (which some misanthropes seem to do deliberately), one can sweep by them easily.
In the heavier post-Thanksgiving traffic of 2010, I noticed the emergence of a new dynamic. Since most passenger drivers want to go faster than the trucks, they prefer the number 1 "fast" lane. When the traffic reached a certain density threshold, however, the line of traffic in the number 1 lane itself began to experience a slow-and-go effect. This could ironically mean that at times the truck traffic in the number 2 lane was going faster than the passenger traffic in number 1. The "fast" lane was losing its advantage as the fast lane.
Of course, if the number 1 lane slows down and the number 2 lane is empty, one may as well get in the number 2 lane. But this was not always of obvious value, since there were usually slow trucks up ahead in the number 2 lane. On earlier drives on I-5, I had occasionally seen something like this happen and generally did not consider it worth while to try and beat the line of traffic down to the next truck. I had noticed, however, that cars that did pass me on the right in this way generally did gain from it and disappeared off into the distance. In a way, I considered this rude, since it involved "jumping the queue" in the number 1 lane.
In 2010, facing heavier traffic than previously, I became impatient. This is not a good attitude to have in traffic, but I began to recollect the cars that on previous occasions had gotten ahead of me by passing through the number 2 lane. So I finally began to do it myself. One of the first tries did not go in the most agreeable way. As I came down to the limit of the truck ahead in the number 2 lane, I cut in in front of a car in the number 1 lane. This guy did not like that, honked his horn, and dispayed some other signs of incipient road rage. I had indeed become the rude queue-jumper. He soon passed me on the right, with evident relish.
I did not like the tone of such a business, or the element of danger attendant on forcing my way into the number 1 lane. Soon, however, I noticed something. I had already experienced a common occurrence on the interstates, when I thought I was following a fast driver in the number 1 lane but discovered instead that he was allowing the traffic to get ahead of him. This often felt like a kind of betrayal, since driving with a group of one or more cars at similar speed often seemed a bit like running with a group of fellow runners in Track, where one runner sets the pace and the others follow, which is easier than everyone running their own race. There is also safety in numbers. If a group of cars is (relatively) modestly breaking the speed limit, the cops may ignore them, or perhaps only pull over the first car. The place of the pace car thus involves elements of leadership and courage. But if my pace car (whether or not they had any idea they were performing such a function) is not doing its job, I feel like I have been let down.
What is a vice in a pace car is a virtue in a slow-and-go number 1 lane. I soon discovered on I-5 that I did not need to tempt fate and rouse tempers by crowding in from lanes 2 to 1. If the line of car is long enough, there will always be someone who is allowing the traffic to get ahead of him. They have then left a comfortable gap with more than enough room ahead. If that occasionally does not happen, it is a loss easily made up in the next pass through the gaps in the number 2 lane.
Thus, the moment came when I sailed past the gentleman whom I had earlier annoyed, stuck in a slowing number 1 lane, found a nice gap far ahead, and then never saw him again. He may or may not have seen me passing. I had no further need of provoking road rage. Where I-5 and the former US 99 merged at the Grapevine, the multilane freeway to the south eliminated the conditions of the particular dynamic that I had been experiencing north of there.
The behavior of traffic is thus a curious business, with variations in circumstances resulting in varying expressions of the "autos as gas molecules" model.