On the zipper merge

tl;dr:

The zipper merge (or late merge) is so misunderstood. The point of the zipper merge is not to increase throughput. The point of the zipper merge is to increase safety (by limiting speeds) and feelings of fairness (by limiting the ability to cut). If you speed down an empty lane to cut in at the end, you are not properly executing a zipper merge - in fact, you are in fact contravening the principles that favor it.

In my view, the best way to execute a zipper merge is to match the speed of the other lane (regardless of whether there is space ahead of you) and then merge at the end of your lane. This procedure is safest and fairest. I want more people to do it.


Spend enough time on the internet, and you are bound to encounter a smug proponent of the zipper merge (also called the late merge).

Springing forth like Batman to the bat signal, at any complaint of a cutting driver, the zipper merge proponent interrupts with smug condescension, in words likely not far from these:

”—Actually, it’s not rude to zoom down the left lane and merge in at the end. In fact, zipper merging is what you’re supposed to do. The Minnesota Department of Transportation website says so. All you virtue-signaling whiners are actually wrong, inefficient, and unsafe. We cutters are merely being rewarded for our superior cleverness and moral correctness when we zoom by you chumps.”

I believe this argument to not only be wrong, but dangerous. Here why.

The Minnesota Department of Transportation lists three benefits of the zipper merge:

  1. Greater safety, because the zipper merge limits the difference in speeds between the two lanes
  2. Less blockage of earlier offramps, because the zipper merge fills two lanes rather than one, backing up only half the distance
  3. Reduced feelings of unfairness, road rage, and ‘forced’ merges, since zipper merging limits the ability and incentive to cut

(Notably, faster throughput is not a given reason. Minnesota’s empirical analysis found no difference in throughput, and other empirical and theoretical work seems to have mixed results, from my brief Googling. A literature review is provided in an appendix at the end of this article.)

Two of these three reasons for the zipper merge are very important, so I’m going to repeat them in all caps.

One reason why zipper merging is encouraged is that it REDUCES SPEEDS BETWEEN LANES.

A second reason why zipper merging is encouraged is that it REDUCES CUTTING AND FEELINGS OF UNFAIRNESS.

Now, ponder this: when a cutter zooms down the left the lane to cut in at the end, are they actually performing a zipper merge? Are they accomplishing the objectives of the zipper merge?

NO.

Zooming ahead INCREASES SPEEDS BETWEEN LANES.

Zooming ahead INCREASES FEELINGS OF UNFAIRNESS.

If you zoom ahead, you are not zipper merging. Zooming ahead DIRECTLY CONTRAVENES the advantages of the zipper merge.

“Okay,” the skeptics among you might admit, maybe one driver zooming ahead doesn’t actually help anything. But if a critical mass zooms ahead, that could help a full zipper merge get going, right? So, you could argue that a driver who zooms ahead actually is doing the moral thing, by trying to jumpstart a critical mass of zipper mergers, no?

NO.

Let me be clear: I am not against the zipper merge. Its benefits are real. But there is a right way and wrong way to zipper merge. Zooming ahead and cutting at the end is the WRONG WAY to zipper merge.

If you actually read what Minnesota, Washington, and other states publish, they do encourage drivers to zipper merge. But notice: in their publications, I have NEVER once seen them say that drivers should zoom ahead and cut.

Say it with me: ZIPPER MERGING DOES NOT EQUAL ZOOMING AND CUTTING.

I present to you now the CORRECT WAY TO EXECUTE A ZIPPER MERGE:

  1. Move into the empty lane
  2. Drive forward, matching the speed of the lane beside you.
  3. YES, EVEN IF THERE IS SPACE IN FRONT OF YOU, MATCH THE SPEED OF THE CAR BESIDE YOU
  4. At the end of the lane, merge

This ‘matching speed’ procedure is the correct way to execute a zipper merge. Here’s why:

  1. It increases safety by reducing speeds between lanes
  2. It reduces the length of the backup by utilizing the capacity of the second lane
  3. It reduces feelings of unfairness and road rage and blocked merges, since you aren’t cutting anyone
  4. AND EVEN BETTER, this has a greater chance of encouraging ‘polite’ drivers to stay in your lane and get a full zipper merge going

To reiterate, THE CORRECT WAY TO ZIPPER MERGE IS TO MATCH SPEEDS WITH THE CAR BESIDE YOU, NOT TO ZOOM AHEAD.

The reason that matching speeds is better is that it actually achieves the trifecta of the zipper merge: greater safety, greater lane utilization, and greater feelings of fairness.

The reason that zooming ahead is worse is that it fails to accomplish the objectives of the zipper merge and in fact makes traffic less safe and flow less efficient.

In conclusion:

The benefits of the zipper merge are real.

But zooming is not zippering.

To properly zipper: Match speeds and late merge. Don’t zoom and cut.

If this post changed the way you think (or agreed with your preconceptions), please share it with others.

Appendix:

When should you not zipper merge?

Do not zipper merge at high speeds. It’s more dangerous and less efficient to wait until the last second. Only zipper merge when traffic is slow and backed up.

Do not zipper merge when a lane is blocked due to an accident or debris. It is safest to give people a wide berth and to give other drivers better visibility so they can brake and avoid it. Only zipper merge at construction zones where it looks safe to do so.

Do not zipper merge in California. California (as of 2017) encourages early merging to increase safety for construction workers. For other states, do your own research. (I believe Minnesota, Washington,

What does the literature say?

What follows comes from my Googling and is therefore extremely prone to selection bias. Beware.

Minnesota’s empirical analysis found found no difference in throughput when the zipper merge was used. (I could not locate the analysis and suspect it to be unpublished.)

Virginia’s theoretical simulations show that for 2-1 merges, zipper merging has essentially the same throughput as early merging (less than 1% difference). At low traffic volumes early merging has a tiny edge and at high volumes zipper merging has a tiny edge. When few trucks are on the road, early merging has a tiny edge, and when trucks are 10% of traffic or more, zipper merging has a tiny edge. (The results for trucks is opposite what I would have expected, but according to my understanding of their simulations, zipper merges do better with trucks because the zipper merge allows multiple cars to drive into the bottleneck as a slow truck accelerates in the opposite lane.)

Nebraska’s empirical work, which was not a randomized controlled study, found that early merging had the highest cost-effectiveness for vehicle volumes below 20,000 vehicles per day, and late merge variants and ‘Smart Zone’ were most cost-effective at higher volumes.

My own intuition is that theoretically early merging should have greater throughput than late merging. If you happen to have a bad merger early, any delays caused by the bad merge get absorbed by the line of cars waiting before the bottleneck. However, if you have a bad merger at the bottleneck, that delay will not be absorbed and instead affects every later car.

Of course, my theory might be oversimplified. If the theory is extended to model the effects of defectors zooming ahead and attempting to merge at the last moment, many of those merges will not be smooth due to blocking, and that could cause more delays than a smooth zipper process.

If you find more studies, please send them to me. My goal is to have accurate beliefs, and I will post any study I can find, whether it agrees with my preconceptions or not.

Other safety considerations

Above, I wrote that the zipper merge is safer because it reduces speed differences between the two lanes. There are other ways that it is safer too:

  1. It reduces speed of traffic relative to construction workers by limiting zoomers
  2. I can also see the zipper merge being safer when you know that merges only happen at one location rather than any possible location. This reduces the likelihood of inattentive drivers not noticing a merge, or getting caught between two merges. It also helps keep speeds more consistent which reduces braking events, and hence likely reduces fender benders and fuel wastage.