Neither a Sidezoomer nor a Lineupper Be

Cynthia Gorney has a great piece in The Times Magazine about the ethical dilemma we all face when the left lane of a multi lane highway is blocked and drivers need to merge to the right. Should you be considerate, move over early and queue up (and feel like a sucker as other drivers whiz past) — or should you not move over and try to squeeze in at the last second (and feel a bit sleazy for passing others)?

Gorney refers to people who sleeze by as “sidezoomers” and to those who move over early as “lineuppers.” Levitt has written about this and definitely sides with the lineuppers (and even supports traffic fines to deter sidezooming).

Barry Nalebuff and I thought about this problem a few years back when we were writing Why Not? (shameless plug). Here’s a solution that we drafted but left on the cutting room floor:

Since neither approach is satisfactory, we should look for a better answer. By this point it should be second nature to see that the problem is one of poor incentives. People who jump the queue don’t care sufficiently about how jumping slows down the other drivers.

Is there a way to prevent people from engaging in this behavior? Indeed there is — quite literally. Why not neither move over nor pass, but simply track the speed of the slower moving right lane? You can block people from passing you without taking advantage of the people who are already moved over. The drivers in the right lane understand how you’ve helped them and always gratefully let you in at the end.

Try it, it works. We know; we’ve done it. The only people who lose are the ones who wanted to jump the queue — but they are usually too self conscientious about their intentions to honk (or pull out a gun).

Our solution of neither merging nor zooming doesn’t solve the problem of the fake exit sidezoom (because you are blocking legitimate exiters) or the surface road zidezoom (again because you are slowing down legit surface road drivers). But it does help in those closed system situations where a highway has a bottleneck without an exit.

It’s also a solution that doesn’t need a lot of collective action. Gorney’s article suggests a third way — but it is a tad too utopian for my tastes. I’ve done the “neither move over nor pass” option dozens of times and my unilateral action promotes both fairness and efficiency. As Barry and I drafted:

A particularly nice feature of this proposal is that individuals can implement it on their own. Thus, the new norm can propagate by example. Every time you do this, the other people in line learn about it and can put into practice themselves.

Of course, there might be a reason why we ultimately decided to cut this solution from the book.


In several places, including my state of PA, there are signs instructing drivers to use both lanes until merging. Sometimes one lane is clearly faster so I take it. It seems like this would be the authorities trying to get drivers to use all of the lanes as some other commenters point out.


Saying that the sidezoomers are using the road efficiently ignores the whole concept of a bottleneck. The bottleneck is the only part of the road that needs to be used efficiently, so the key to keeping traffic efficient is to make sure that everyone is at a smooth, similar, as-high-as-possible speed when they enter the constricted area, as speed differences lead to unnecessary speeding and slowing, rather than efficient cruising. Sidezoomers cause a great deal of start-and-stop traffic which is worse on everyone's cars, gas mileage, brakes, and moods, and it has no positive effects and likely some negative effects on the traffic actually inside the bottleneck, where it matters.

M Todd

Sidezoomers are the reason it bottle necks at the end. Instead of a steady stream of traffic people are jamming on their breaks because zoomers are forcing their way in.

In our state, truckers will line up side by side and force everyone to merge into the proper lane. At the end the truck running the block is let in and everyone tightens up the line so the sidezoomers cannot merge. In real problem areas the police will ticket those who zoom to the end only to find a ticket waiting for them because they ignore the merge signs.

At the beginning of the warning to merge I will let people trying to get over in, but when one zooms by me past the half way point forget it they can wait until at the end of the closed lane until rush hour is over or they die of dehydration, either way problem solved.


RE: #3: "People who move over early only promote the inefficient use of of the highway"

I think you are misunderstanding, if not economic theory, then at least the dynamics of what is happening in traffic.

The "scarce good" in this situation is not free lane space (hey, there's probably hundreds of miles of the stuff right behind you) but is in fact the opportunity to be first through the choke point.

By delaying merging until the last possible moment, you are the other sidezoomers (is there a reason this is a near-anagram for "Sodomizer"?) are creating an artificial shortage of merging opportunities right at the choke point. This is the reason traffic usually speeds up once you are past the merge point.


There are two problems with sidezoomers -- at least the ones on the DC Beltway, where I commuted for almost five years of observed lunacy:

1) Having arrived at the chokepoint more quickly than the lineuppers, they seem to feel it is their sovereign right to merge instantly, even though it may force the entire queue of lineuppers to brake to avoid a collision, starting with the person the sidezoomer is cutting off, and;

2) Many of the sidezoomers are in fact also members of a third group, the imspecialists, so no matter how far along the last guy cut in, they'll try to go a little farther, which is why the MD state troopers do such good business in tickets at the I-95/Beltway merge from people driving on the shoulder because they clearly believe the law about not doing so really doesn't apply to them.

Jason Kramer

Your proposed solution would eventually lead to the proper solution, which is to use both lanes the entire way to the closure and then alternate cars from both lanes.

In your proposal, a car in the left lane tracks the speed of a car in the right lane, backing up sidezoomers and lineuppers in both lanes, until the blocker moves over at the point of closure. Then cars from both lanes begin to alternate. In other words, your solution solves nothing, except to make the driver of the blocking car feel smug and righteous.


Much of this analysis seems very utopian. Many of you argue that sidezooming and merging at a set point is safer - but as long as people are pissed off at the side zoomer, I think aggressively trying to keep them out of the lane could make things MORE dangerous.


Hmmm. Perhaps this works and is reasonable in situations where 2 lanes go to 1, but what about congested exit ramps? This would just be totally absurd as you would be slowing down traffic that did not want to exit. Additionally, without the zoom-uppers to fill traffic holes which are always left by lineuppers, you would not be globally maximizing traffic utility.

In fact, I would contest that even in the two-to-one lane situation, by driving slow in the closing lane, you are basically enforcing a tax on the utility of the other drivers: those who zoom are losing their "getting-there-faster" utility while those who line-up, you are hurting their "I don't-have-a-guilty-conscience" utility.

So go ahead with your communist driving practices if you want. But from an economist, this recommendation seems a bit odd.



I agree with #2,3 and 4. No one should move over early. This is the most efficient solution.


I couldn't agree with the third comment by Mike B more. Until the left lane is closed, it's still an open and usable lane. Congestion is prolonged needlessly when people merge before they have to. My wife is an early merger and she does so for one reason- she's so nervous that she's not going to be able to get into the middle lane that she'd rather just find a way in earlier and save herself the worry. To each their own, but actions like this slow down traffic for everyone else.


aaron (assuming all of the comments from "aaron" come from the same person),

the finding in the book you site (#20) (which i am believing, since i have not read the book) is interesting. i always assumed that sidezooming was the same as lineupping, or slightly worse due to last second congestion at the bottle neck (obviously the first 10 or so poster who said that it is more efficient had no reasonable support for their arguments, as has been pointed out). if it is indeed the better, the acceleration issue (#30) would be a logical reason why. maybe i should buy the book.

but your claim in #42 that speed is the same, but somehow the long line makes the wait longer is silly. if cars are getting through the bottleneck at the same speed, the average wait in both situations is the same.


Jason makes an interesting point about maximum efficiency being achieved, in the case of a traffic choke point that opens back up, folks getting out of the busted lane early.

I don't think this is really the problem -- if you're at an unexpected accident, the lane that's closed (with the car wreck, for example) is going to be the slowest, so side-zoomers wouldn't be drawn to it anyhow.

The problem is when a freeway loses a lane and doesn't gain it back. If everyone were to diligently merge over ahead of time, as can happen on the approach to the Golden Gate bridge when coming from the north, a major slow-down can be avoided. However, this is really only possible if there's no traffic in the first place, as once there's traffic, the incentive to side-zoom becomes very powerful, and even if you were doing the side-slow technique, you still wouldn't be able to restore the traffic flow to normal speeds -- you'd have perhaps a little moral victory, but not a practial one. Once the traffic's already there, you're not going to improve things by driving slower or merging earlier -- so use that merging lane and follow your self-interest.



Here in Germany the official recommendation is to merge only at the end of the blocked line right before the obstruction and there are often signs explicitly saying to do this. It is also taught when you are getting your drivers license.

The main reason to do this, I think, is not that it significantly changes throughput at the bottleneck but that it shortens the total length of blocked highway. Having exits every few miles means merging early and increasing the congestion length causes all kinds of secondary problems further behind and thus decreasing overall efficiency.

The explanation for mergin early I hear by far the most often is that people want to avoid the stress of being afraid to not be let in at the end. On the other hand this means that if you do sidezoom you will always find someone willing to let you in at the end because they emphasize and actually want to help you.

Personally I find it very counter-intuitive to merge early and not use a free lane. As said before the other extreme to merging at the very end is to always use only the right lane making multi-lane highways unnecessary. I would argue that given cooperative drivers merging at the point of slowest speed, which is right before the bottleneck, is easiest and thus most efficient and least likely to cause any accidents.


d smeeze


I agree with you completely (especially about the being civilized part)... except for when you say that merging is a zero sum gain. Not only is the defector's gain inflicted as a loss on other drivers, but that loss is usually even greater than the defector's gain, so that the entire system suffers a net loss in efficiency.


@48. JB, first off, simulation models don't "prove" anything. A model is only as good as the assumptions underlying it. Secondly, I have seen simulation models used to uphold both arguments.

@49. Ecks, improving the flow rate at the bottleneck as the only objective is a myopic view of the problem. What is happening miles before the bottleneck? How does early merging affect that? What is the mean wait time per driver or passenger? What about variance? There's a lot more to this problem than being [explitive] civilized. Otherwise, you wouldn't have much argument.

Joe P.

The *entire* reason why this is considered "cheating" is because the late-mergers get an advantage. However this advantage only exists because there aren't enough late-mergers; if more people merged late, then there would be less advantage to doing so, and less disadvantage to being in the main lane early (since fewer people would be queued there).

I believe that balanced queues are always the most advantageous situation, with one designated merge point. If this is correct, then that designated merge point may as well be the point at which the orange cones force it, and no earlier. The entire reason that people feel cheated is that they assume that the merge point is much earlier than it actually is required to be.



Jim, it's because the cars will be travelling at that same slow speed for a longer distance.


Have you ever gone to the seafood market and watched the live crabs? Every now and then one crab tries to break free, gaining a claw-hold on the edge of the tank. As soon as he looks like he's going to get away, all the other crabs grab him and pull him back down. We sidezoomers are that crab yearning to be free, the rest of you are the oppressive majority, blindly following the mob. Fascists, the lot of you!


I was an early lineupper until a few weeks ago. It is simply not true that sizezooming slows the rate of traffic through the bottle next. All it does is lengthen the queue causing every one to wait longer. This was determined by real world studies.