Regardless of their perceptions or actual travel speeds, the lane-switchers could be providing a much needed liquidity between the more dynamic right lanes (with entrances and exits) and the more static left lanes. Perhaps it's in doing so that they unwittingly ensure that everyone arrives at the same time. (Although it's almost certainly not the optimal time…)
I've assumed similar things, though I recognize the 'turbulence' argument below. In general, it does seem that as one lane slows down a bunch of people jump to the faster lane, quickly slowing that one down.
Regardless, for me the best strategy is to almost never switch lanes. Even when I do switch, 15 minutes later I see somebody that I had passed earlier zipping by me in another lane. There is no value in switching for me that I can determine, and my commute is much more pleasant when I stop thinking about how to get one car length ahead. Whether the lane switchers improve efficiency or reduce it doesn't change that equation for me.
I said 'almost never'. There are a few places on my commute where free lanes are predictable - everyone is merging right to try to get to get onto some interchange, that so several lanes back up, while the people who are trying to stay on the same road can move left and travel quickly. But even then, what are you optimizing for? Maybe 15seconds? Sure, it is more than that in the moment, as you can sit in the slow lanes for a few minutes, but all you do if you move left is rush ahead to the next back up.
I'm always torn between the idea of liquidity or turbulence. Intuitively, turbulence makes sense, but then what happens to facilitates people in the left lane exiting at their destination?
I'd love to see some work done to figure this out.
Lane switchers are transverse waves. Acceleration and braking are longitudinal waves.
I am of the opinion that lane switchers have very low impact on traffic speed, whereas people who gun the engine and slam on their brakes are most directly responsible for slowdowns.
Of course, this presumes that the lane switcher does not switch in a manner that causes someone to slam on his brakes.
Thus, if my theory is correct, attempting to smooth out the longitudinal speed waves should take priority over transversely vibrating cars that could potentially exploit that strategy to move faster than the rest of traffic, at the expense of everyone else.
I don't know how it is elsewhere, but in the Toronto, Canada area, driving behaviours have gotten really bad over the past 20 years.
We have the worst kind of lane switchers. Here, they often jump into the short lane on the right, which is reserved for cars entering and exiting the road, to cheat back in. This is a real problem, because they slow down traffic even more when they have to force their way back in.
The counterintuitive solution to this is to always make room to let those people back in. You aren't helping them "cheat", not really. You are keeping a potentially dangerous driver in front of you, where you can more easily keep an eye on them, and allowing them the opportunity to get the hell away from your car.
