Hacker News new | past | comments | ask | show | jobs | submit login
Washington, Minnesota officially endorse a “safer, faster” traffic merge (arstechnica.com)
179 points by smacktoward on July 24, 2014 | hide | past | web | favorite | 200 comments

That video explanation made no sense to me. Basically the "interface" presented to the drivers has not changed, so why would drivers deviate from how they currently merge? Because the government said so? I don't see this working.

However there are two solutions that could promote zipper merging:

(1) Force both lanes to the center before moving that center lane to the side that is available ahead:

    _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ________

(2) Present a long-barrier between both lanes well before the merge that only permit late merging, so that those in the continuing lane feel morally obligated to let those queued up in.

    _ _ _ _____________     ____________________

In New Zealand we just have signs like this all over the place:


Its pedantic, but I think this is sublty different than what is being dicussed here. All over the US, there are situations where 2 lanes converge to 1 and drivers "take turns" merging. But I think the scenario that is being discussed is a little different, where there is 1 "working" lane and one "closed" lane, the thought process of drivers in the lane that they know is 'closed' up ahead is to merge early.

I wonder if cultures where queueing (like, in the grocery store) is not as common as in the US have this problem? I suspect not.

I live in the Netherlands, but drive through Belgium when driving home from work.

There is one area where a permanent two-lane road merges into a single lane. The funny thing is that (mostly) Belgians never use the second (left) lane that eventually merges into the single right lane.

This always seems stupid to me, because it increases the length of the traffic jam so much, that the traffic jam goes all the way over a roundabout, causing stops for all drivers, even the ones not going in the same direction as us.

The worst thing is, whenever I use the left lane - which I do all the time (and in return, pass 60 cars waiting in line), people sometimes get so angry that I've had two trucks in the past few months hitting the gas and blocking the left lane so no one can pass anymore.

Crazy people...

Last month a new sign was installed along the road "Please use both lanes, drive all the way to the front and THEN start merging". Even now, people still choose to wait in line at the right lane. It is as if people feel guilty when they see a whole lane being empty and available to drive on.

This is what it's like in Scotland!! People here get SO angry, thinking that you have "skipped" all the other patiently waiting people... And then they don't let you merge! And it makes absolutely no sense to me, because they cause one reallllyyyy long traffic jam then...

We have temporary merge like a zip signs as well for road works, I just didn't see one on Google Image Search. They only tend to put those out for fairly large multi-week road works I've noticed though.

The two lanes in the picture merge into one "center" lane. I've never seen that here in the states. Typically it is one lane merging into another lane. I'd wager the "merge into the shared center lane" makes the zipper protocol more obvious.

I think you misinterpreted the picture - the left lane ends and merges with the right, as when joining a motorway (we drive on the left). You can see the real thing in the photo above the closest two cars.

The zip system is used ubiquitously here - I didn't even know there were other systems, so I was curious to see what new thing the article was about. From what I understood, the alternative is not really a "system"?

To my understanding in intensive traffic it's more about uniform and predictable behaviour.

First option will not improve since (at least in Europe) it just changes lane priority. In linked article's merge case priority traffic is in left lane, since right one merges. In your center merge case priority traffic is on the right (when doing simultaneous merges to center lane, one on the right has priority over the one on the left). So it probably won't change anything.

Barrier option is much better, but it has challanges and restrictions of their own.

Edit: Oh, and I forgot to mention that over here in Latvia I see people starting to respect zip merging. It's all in our head. Not because government said, but because everybody follows simple rule and everybody gets home faster and with less stress :)

Option 2 physically walls off the Late Mergers in addition to mentally (in the minds of the Free Laners), making them easier to ignore.

I like option 1 because it forces everyone to move.

Yeah, option 2 would lead to someone hestitating/stopping in that lower lane. Then that whole lane is stopped because the front driver is too afraid to merge into moving traffic, and the moving traffic is going too fast to stop and let a lower lane driver in.

Good thinking. Apparently your option 2 was implemented (and considered an innovation) around where I live. Here's a picture from 2007: http://www.strassen.nrw.de/_down/pi-070322-03.jpg I don't know if they're still doing it. (Late merging has long been the default in Europe and is taught in driver's ed. However people seem to do it more reliably when reminded of the practice.)

I wonder if there's an option 3, install a mobile traffic light that allows one lane only at one time, in sequences of one minute or so each, i.e Alternating red and green lights, so that there is no need for merging. This way there's no slowdown caused by the merging, I feel that this method may actually make it faster overall.

    _ _ _ _ _ _ _ _ _ _     ____________________
                      Red  /


    _ _ _ _ _ _ _ _ _ _     ____________________
                    Green  /

I was driving in Atlanta a few weeks ago and was in a situation where two two lane roads converged to a single two lane road. Each side had this exact thing and it actually worked really well. Not the same as a construction zone but I think would be a helpful tool.

Temporary traffic lights is definitely an interesting idea. It would however require a long barrier, otherwise, what is to stop people at the end of the stopped lane from trying to merge into the moving lane.

Another idea is a temporary speed bump in the lane that is not ending.

The picture Negitivefrags posted shows a similar setup currently in use in New Zealand (although it's a permanent merge instead of temporary construction)


Except that red means "stop", so presumably that whole "you don't have to stop" thing wouldn't work.

Hey, I'm from Minnesota. Here is a merge spot that I take pretty often: https://www.google.com/maps/@45.008271,-93.203054,3a,75y,63....

Unfortunately, this spot still has lots of early merging. Some high traffic areas have special lanes for zipper merging, but I think the emphasis is on changing drivers' habits through public service announcements rather than redoing the lane structure.

Option 1 leads to "no, after you" behavior, steers two cars directly at each other, and probably would be a disaster.

Speaking about educating drivers, I wish we could just kill that mentality once and for all. It drives me crazy when someone obviously beats me to a stop sign, then waves me through as though they're doing me a favor. No my friend, you're just making it slower for both of us by not following the established rules of the road.

Thankfully my city is replacing as many stop signs as possible with traffic circles, where you'd have to be a complete idiot to yield to someone who got there after you.

> Thankfully my city is replacing as many stop signs as possible with traffic circles, where you'd have to be a complete idiot to yield to someone who got there after you.

In Seattle, we have traffic circles in many of our residential neighborhoods. People still do this here. Another common thing is people taking lefts by just driving on the left side of the circle (I've had people honk at me for taking lefts the correct way)

I've sometimes waited at those circles. Many of those neighborhoods you mention also have streets just wide enough for one vehicle at a time to get through. I'd rather wait at the circle for an oncoming vehicle than have to worry about who will yield to whom and whether there's a sufficient gap in the parked cars to do so.

This can be seen in miniature on my university campus. Thankfully some new surface markings make it slightly more obvious that they are going the wrong way.

Out of curiosity, is your traffic circle configured in a way where taking that left requires no turning while the proper right actually takes effort?

Pretty much. Most of the ones that I've seen here in Seattle are basically small landscaped areas with a curb in the middle of the intersection. Probably not much more than six feet in diameter or so. (At the one nearest my house the sidewalk corners aren't even rounded any to accomodate them.) These really aren't true roundabouts.

But yes, turning left instead of going around them should take no more effort than a typical left turn between small streets, while doing it the proper way requires a fairly tight turn. They seem mainly designed to force cars to slow down.

we have a couple around here near campus. The problem is shockingly few people know how to behave in a traffic circle.

"few people know how to behave in a traffic circle"

Tell them this: imagine a big circular road. Then imagine other roads joining it with T junctions. Then imagine the road is one-way. Finally just shrink the road in your head.

That's all it is!

There's one more rule: traffic on the circular road has priority.

There's one more rule: traffic on the circular road has priority.

This is how it would work in a sane system, but they don't always work this way. I can think of several examples in New Jersey, particularly around Flemington, where the circle manages the intersection of two roads having different importance. In these cases, the important road is given the right-of-way all the way through the circle, and those coming around the circle from the lesser road are required to yield.

That rule already applies. It's no different on a circular road or roundabout; traffic already on a road has priority over traffic turning onto it.

I guess that depends where you live. That thing is intersection like any other and the one on the right has a priority. So, unless there is a sign, the car about to enter the circular road should go first. And yes, all roundabouts have signs in here.

I live in Australia, but I've also driven in the UK and parts of Europe. Any roundabout I've used has worked in the same way -- as you approach the roundabout, you must give way to vehicles already on the roundabout.

Relevant to the circular road with T-intersection analogy, I'm reading the Wikipedia page[0] on T-intersections and I must admit the concept that "vehicles on the right always have the right-of-way" makes absolutely no sense whatsoever. But I suppose that weird rule would screw up the circular road metaphor.

[0] https://en.wikipedia.org/wiki/Three-way_junction#Right-of-wa...

"Traffic already on a road" is a concept that does not apply in this scenario, as all traffic is already on a road in all cases. There is no a priori way to distinguish between the circular road and the roads leading towards it.

In countries where you drive on the right, I believe the typical default rule is that traffic coming from the right has priority. In roundabouts, this would mean that traffic on the roundabout must wait for those joining it. However, the rule is reversed for roundabouts. (This definitely applies to Germany, and I suspect that it applies in most other countries as well.)

They have signs that say "Yield to Traffic in Circle" at each entry point. I agree it is a simple concept yet people seem to struggle with it around here.

We have the same problem. Unfortunately, it's compounded by the fact that the local municipality here likes creating very small traffic circles with a small bump in the middle. The end result is that people end up treating them like 4-way stops and yield as such.

They're trivial to learn. I don't see how people can use that as an excuse. Imagine someone saying "I don't understand traffic lights, they're so confusing."

This is the worst when you're on a bike -- the car tries to be nice and let you go, but you can't assume that they're not going to zip out and kill you if you do, so everyone just fucking sits and waits, when you could have just zipped through behind them. It's insane.

> you'd have to be a complete idiot to yield to someone who got there after you.

You'd be surprised. Or else the vast majority of drivers around here are complete idiots. They basically treat traffic circles like a confusing 4-way stop sign. It's infuriating.

I reckon you could do the following to fix this issue:

(1) add the barrier from option 2 to option 1, so that all the merging happens at the front.

(2) Keep the "make both parties move structure" from option 1, but don't make it 100% symmetrical. Instead, nudge the top lane down just enough so they are forced to slow down and somewhat zipper in with the bottom lane.

I like the idea of #1. But it seems from my experience in the U.S. and Canada, that there is a tendency towards unambiguous right-of-way. That tends to rule out collaborative scenarios where there's no clear "winner" when two cars are competing for the same resource. Which is a shame, because making this collaborative is exactly what will change the mentality (i.e. you can't ignore the drivers trying to merge, because you're trying to merge too).

Even with zipper merging the right-of-way is going to be on the person who is staying in their own lane.

I'm not sure if my experience is regional, or even imagined, nor do I know the reason...I'd imagine it makes assigning blame in an accident much easier, but that's a really crappy basis for traffic rules.

#1 is what happens in Fremont,CA when cops are running a DUI check, works fairly well.

> so that those in the continuing lane feel morally obligated to let those queued up in.

I'm pretty well convinced that we'd never hit critical mass of drivers that feel morally obligated to even moderately inconvenience themselves to benefit the merging of others. That merge pattern is a recipe for stalled traffic in the right hand lane.

First one seems workable, though.

Yes! #1 is what I was thinking too. Remove the dominance/subordination from the merging and it's obvious what one is supposed to do -- zipper right in one at a time.

I like option one, but I don't think it works if there are more than two lanes

Option 1 sounds great in theory, but given that most merges are the result of temporary lane closures, completely impractical (since road work usually goes one lane at a time)

How is it any more impractical than how temporary lane closures are currently done? It can still be done with cones or barrels and the end result is exactly the same.

The point is, suppose they are repaving the road one lane at a time. (Or, repaving half of a 2-lane road)

You simply move up the half-merge of each lane to suit. It's rather clearly explained in the diagram, where you can see it does end in one and only one lane actively in use. You'd do the "merge half of each lane" part while the road still supports two lanes, just as you already have to account for with current temporary traffic control devices.

Ah. I see; I am not a very good ASCII art interpreter, is the problem.

I'm surprised this is written like zipper merging is some kind of recently discovered secret. Zipper merging (or weaving) has been standard practice in parts of Europe for 20 years (maybe longer – I can't really remember anything else). Since March 2014 it is even mandatory in Belgium, and anyone merging early may risk a €55 fine [1].

It has been repeatedly advocated and promoted by various governments in Europe [2] and I doubt anyone questions its effectivity. Some experiments have been done in the past with alternative methods like zipper merging with groups of cars, but this is only applied when traffic can be regulated (with lights or traffic controllers). The nice thing about zipper merging is that with enough public knowledge, drivers can regulate merging themselves.


[2]:https://www.youtube.com/watch?v=TNhnLYI2Efw – Dutch government TV ad from 1989 which explains and promotes zipper merging (albeit very poorly, I must admit).

It's not the concept that is newsworthy, it's the decision to actively promote it that is.

More importantly, the decision by American transportation authorities to promote it is newsworthy. It seems like Northern Europe takes a more systematic and serious attitude towards driving, while the American attitude is more "eat my breakfast while driving 50mph in the left lane, and give the finger to everyone who passes me on the right"

Agreed. Zipper merging has been standard practice in New Zealand my whole (28 year) life. Except we call it "zip merging" because "zipper" sounds awfully foreign around here. I didn't know there was another way to merge until I read this article.

I wish it would work...they teach it since decades in Germany and you still see people not doing it right or worse: people complaining that you drive up to the end. People not letting you in at the end and so on...

I still try it every time because most of the times, it just saves me a lot of time.

I've never seen/heard of people being fined for not doing it or preventing people to merge.

> I'm surprised this is written like zipper merging is some kind of recently discovered secret.

Yeah, it's funny :) In U.S. they have the same with road circles/roundabout. Some time ago I watched TED talk where the speaker was talking about this like it was some amazing innovation :)

Zipper merging works great as long as everyone on both sides is willing to alternate (or if strong enforcement requires them to do so). Otherwise, letting "late mergers" scream past fifty "early mergers" often creates a situation worse than simple early merge. There's plenty of blame to go around.

* The vast majority of the people blasting down the late-merge lane clearly have no interest in alternating or merging into the largest/safest gap at the end. They'll squeeze into the smallest latest gap that they think they can possibly get away with.

* Quite a few of the early mergers feel they've already done their part. Whether they're competitive, territorial, or sincerely concerned with traffic flow/safety, they're not much interested in alternating with the late mergers either.

So how does this play out every effing day at construction projects and toll booths? You end up with a few late mergers playing chicken with a few early mergers at the chokepoint, jockeying for position inch by inch and degree by degree as they try to cut each other off. That's where all the "vigilantes" come from - they're people who are already stuck in the aftermath, they know how it happened, and they have decided that they'd rather settle for forced early merge than botched zipper merge.

Without enforcing proper merge behavior on both sides, zipper merge just becomes a big Prisoners' Dilemma game with a few defectors screwing all of the cooperators.

> "Without enforcing proper merge behavior on both sides, zipper merge just becomes a big Prisoners' Dilemma game with a few defectors screwing all of the cooperators."

Except unlike prisoner's dilemma, it's not a zero (or negative) sum game. The entire traffic flow can speed up if zipper merging is done. The "late mergers" don't have to risk losing their game of "chicken" while the "vigilantes" don't have to deal with the reckless maneuvers of some of the former.

People don't do zipper merging because people are human and drive emotionally [1]. The stereotypes of the conservative early merger claiming the moral high ground and the reckless late merger doing anything to save a few minutes are the cultural results of this mentality.

[1] Another example is the special breed of driver who stops the traffic flow on a major road (which has the right of way) to let someone on a minor cross road or parking lot exit onto the major road, thus inconveniencing everyone behind them for the self satisfaction of having helped the one person they can see.

WRT to your footnote, I also can't count the number of times I've seen that person on the side street come into traffic and get t-boned by the car that didn't stop in the next lane over. Letting people cross traffic like that isn't just illegal, it's downright dangerous[1]. As a motorcycle rider, few situations scare me more than seeing a lane of stopped traffic next to a clear lane.

1. The same thing applies to stopping to let people turn left across traffic, where the left lane of oncoming traffic is stopped and the right lane is clear.

[1] Another example is the special breed of driver who stops the traffic flow on a major road (which has the right of way) to let someone on a minor cross road or parking lot exit onto the major road, thus inconveniencing everyone behind them for the self satisfaction of having helped the one person they can see.

This is similar to the person on a grocery checkout line who lets someone cut in front of them; they are also putting that person ahead of every other person behind them the line.

It's an interesting situation; people feel they are doing the right thing yet apparently never stop to think how they are imposing their decision on others.

There's a blind-spot of sorts regarding the larger effects of their actions.

> [...] inconveniencing everyone behind them for the self satisfaction of having helped the one person they can see.

Or because you know that otherwise they'll be at that crossing for the next hour if nobody's willing to let them by. Perhaps there are some funny allusions to net neutrality to be made here...

Just because you want to turn left, doesn't mean you have to sit helpless waiting to do so. You're allowed to turn right and locate a safe place to turn around.

I have sympathy for left-turners in particularly busy intersections where there is no turn light. Driving in NYC, you pretty much have to nudge as close to the opposing lane as you can until the light turns yellow/red and you can scoot yourself out of gridlock. Furthermore, you must account for streets which are one way, too narrow, full, or active to stop and turn around. However, this was not the situation I was really talking about at all.

The scenario I had in mind while writing my post is that of the local road which crosses a major stretch of highway. Such highways might be an actual interstate, a busy two-lane highway, etc. One of the worst types of road to cross is the one that is far too wide for its own good– an ambitious 6-lane road with not enough traffic to ever be full (Trenholm Road in SC, ahem), but enough so that drivers spread out across all lanes and make crossing very annoying sometimes. There are parts of the Taconic (NY) that pretty bad to cross, though I'm usually the one on the highway. The local cross sometimes gets backed up 5-10 cars, 2-3 of which are stuck in the median. Getting on the highway or going around by going the next town can add 10+ miles. Letting someone cross by slowing down a bit and changing lanes is not, in my opinion, a smug assault on the efficiency of those driving on the highway... it's just something a few people do when they recognize that nothing is very efficient anyway. (Note that I haven't said much about safety because it was only a small part of their comment, and not the part I was responding to. I hope it's clear that I am not advocating for people to test fate by slamming their brakes to let someone curtsy across the pavement.)

Zipper merging is the optimal solution. People early merge because they think it is the optimal/moral/better/'done my part' solution.

If everyone knew about zipper merging there would be no early mergers (and 'late mergers' don't exist within the context of zipper merging).

The problem is most people don't know how to properly zipper merge, so pretending that everything would be hunky dory as long as people play by a set of rules that you know they aren't likely to play by is an exercise in futility.

Well, quite a lot of traffic is already regulated by people "playing by a set of rules", and that ends up being hunky dory for the most part. I think the reason why this one is the exception is that it's quite feasible for people to "fight back" to those they perceive as abusers.

The real problem is accountability. There is none of it(as cops can't be everywhere, all the time) so people end up exercising what little power they do have in order to punish those they view as the abusers.

Don't know why you're being downvoted; I've had too many people deliberately block my merge when I merge late, to ever try it at full speed in the UK :(

The solution is more late-mergers: if the other lane were already full, you wouldn't be "cheating" by passing all the cars who are "waiting patiently." You would just be waiting in a line the same as everyone else in your lane and the other lane.

I may be overly optimistic, but I think a sign saying "use both lanes and zipper merge at end" would do a lot.

People block your merge because they think you cheated and should have merged earlier. If they were educated about zipper merge they would not have the same feeling.

In simulations zipper merging is an optimal solution. In reality, relying on thousands of drivers to properly follow specific directions and be completely aware of their surroundings is a flawed solution. Telling people how to do something efficiently doesn't mean they'll do it that way. People are extremely loose when it comes to following driving rules, e.g. speed, texting, minimum distances, merging, etc..

I agree that telling anyone to do anything doesn't mean they will do it (change their habits).

However, educating people does allow them to make better decisions.

Rules themselves aren't always the best/optimal/safest solution either and being loose may be the best solution.

For example, in one country in Asia if you are "in-front" of someone you have the right to merge/enter a roadway/get on a free way/change lanes over the person already in the lane. This is just cultural and is so much better than what we have in the west (like the USA). If someone changes lanes and "hits you", it is your fault (unless you were in-front of them).

Makes it so much easier to get on the free way, change lanes, etc. as you only have to worry about what is in front of you. Not what is in front of you, behind you, next to you, etc. There are no "blind spots" etc.

Zipper merging is only optimal if traffic is already slowed down dramatically. If traffic is still moving at any kind of speed, then early merge when there is space available will clear congestion faster than everyone driving up to the merge point and hoping that it works out.

No, it's provably optimal in flowing traffic as well. However, the "merge point" is variable depending on traffic speed.

Obviously when you're going 100 km/h, the ideal merge point is significantly farther back than when it's stop and go. Nobody is saying you should wait until you're 20 feet from the barrier and then swerve into the other lane. But it's not like that's unique to this situation.

Zipper merge is synonymous with late merge: https://en.wikipedia.org/wiki/Zipper_merge

If you move the merge point back, then by definition it becomes an early merge, right? I don't understand your disagreement with what seems like a totally reasonable statement by Justin.

My point is that what constitutes "late" is a function of the speed of traffic.

If you define "late" as "the latest point at which you can safely merge without causing negative traffic reactions (e.g. by forcing someone to brake)" then zipper/late merging is better than "moving over as soon as you can", even in light traffic.

In an early merge there is no real merge point, people just merge wherever they can. As traffic backs up further and further people merge earlier and earlier.

This is the commonly held belief that is false. It is false because traffic is caused by lack of capacity, not inefficient driving.

Road capacity and driver efficiency are related. Roads have a theoretical capacity that is rarely reached due to inefficiencies in driver's habits.

Exactly. Drivers as a class tend to think that they are smarter than everyone else, particularly the transportation departments, and think that e.g. speed limits are something you need to be above, otherwise you're an annoying driver. It's a huge cultural problem and one of the main reason I can't wait to see drivers displaced by self-driving cars.

> Zipper merging is only optimal if traffic is already slowed down dramatically.

I agree. But note that the idea of Zipper merging is when traffic has already slowed down, not when it is still moving.

>> Zipper merging is the optimal solution. People early merge because they think it is the optimal/moral/better/'done my part' solution.

It is optimal only when the local driving culture allows it to be.

I live in an area where drivers are extremely selfish and the police don't seem to spend a lot of time ticketing bad drivers, so zipper merging typically doesn't work here.

I live in an area where drivers are extremely selfish...so zipper merging typically doesn't work here.

This is backwards. Zipper merging doesn't work when drivers attempt to be unselfish by merging early, which results in vigilante lane blockage of those who weren't as considerate. In other words, I'm not being selfish when I block late mergers; I'm enacting justice on other people who were being selfish.

The system works when all parties recognize that the best thing for everyone is to merge late.

People used to zipper merge where I live... in the 80s. But the metro region I live in is much more dense than it used to be, and combined with the fact that the police don't spend a lot of time enforcing traffic laws, and that the public transportation system has poor coverage, traffic problems have skyrocketed.

One of the reasons why people early merge here is that many people don't let mergers in. And one of the reasons why people don't let mergers in is that they regularly get cut off by a second car after letting a merger in. It's a vicious circle.

Back in "the day", zipper merging wasn't even an issue. A lot of people early merge here because they'd rather be the one letting people in than hoping to be let in.

You're overlooking the situation where people merge out of the lane in a line without letting people who are already in the lane go through.

> It is optimal only when the local driving culture allows it to be.

Yeah, which is why almost the entire article was about the campaigns to change the culture, not about justifying zipper merging as optimal.

Fair point, but it's easier to successfully campaign a low density region (i.e, a state with around 7M people like Washington) than a high density metro area (the metro area where I live in has 6M people, so traffic is already a daily nightmare).

Those campaigns in Washington and Minnesota just wouldn't fly here.

When our metro population was around <=3M, people still zipper merged (in the mid 90s and earlier).

If enough people are doing zipper merging, then there is no way to cut in line, because both lanes are equally backed up. Then once people get to the merge point, they won't have any sense of unfairness because the other lane was waiting just as long, so there will be less impetus for "vigilantes" to mess things up. I'm not saying there will be no problems, but I don't think that more zipper merging will automatically lead to more of these problems.

The question then is, do enough people zipper merge, and do people really have a proper understanding of fairness? A lot of the driving population may not be acting rationally, or even have enough awareness to comprehend the big the picture. Cars put people into their own bubble, where quite oftenly only their needs are being considered.

They certainly do not. I always leave space for a car at the zip, and damn near every time 2 or 3 will try to push their way in there even though there is only space for 1 car. Infuriating.

>> Cars put people into their own bubble, where often only their needs are being considered.

This is so true. At congested merge points in my city, you have people who don't really care about the alternating needed to make a zipper merge work. First, you have people who just won't let people in, and then you have people who will try to sneak into the lane with the merging car ahead of them.

Yes, of course many drivers will always try to cheat no matter what, but many more will cheat if they feel they have been wronged. My point is that the latter subset would be eliminated if most people used zipper merging.

I tend to start pacing for the zipper merge when I'm (reasonably) close to the end of the zipper. This way, I'm showing I am willing to wait while simultaneously giving myself the option of skipping forward past any particular territorial driver. When I slow down the merging lane to a similar speed as the non-merging lane, I can typically see the zipper start to form behind me when I check my mirrors.

In the same way one or two bad drivers can destroy a lane or two of functional traffic flow, a good driver can create a lane (or two) of harmony.

Hello, my name is Peter, and i'm a late-merger. But where I come from it's just called "Florida driving".

In Florida, we all have a non-verbal mutual agreement that we're all trying to screw each other out of the best spot in the road. We all want to be going faster than everyone else, we all want to get to the front of the line and skip all the chumps, we all want to move into the lane we think will move the fastest at a red light, and we all think turn signals will just warn someone when we're trying to get ahead so they can take our spot on the road.

It honestly wouldn't be so bad if we could keep the speeding and aggressive driving controlled, but there's an unusual number of highways with ridiculously reduced speed limits (55mph on I-95 in Miami, I shit you not.... right around a large stretch of road where 45mph is the minimum speed limit) and a lot of assholes who think they're more important than everyone else. Not much you can do about it, other than a "fast lane", which is basically what they've created for a large stretch of I-95 combined with the Sunpass system.

We have this problem in Texas but there is an additional problem; If you are not an early merger people tend to not let you in anywhere in the chain forcing you to become a late merger. I tend to grab a spot early because too often people would not allow me to merge at any point. Texas drivers are very aggressive, I imagine this scenario is true in other places as well.

In reality, the early mergers are the problem.

They are screwing up the system by merging too early, then adding insult to injury, getting all bent out of shape about people who aren't 'early mergers'.

So first they create the problem, then they get upset when someone does it the right way. Classic!

In practice zipper merging works amazingly well. MNDOT has used signage to tell drivers to maintain speed and stay in their lanes for the past few years of construction in the northern Twin Cities. You wouldn't believe how much faster traffic moves when these simple reminders are posted.

The problem is that without the signs (and enforcement) drivers in the Twin Cities Metro regularly shoot to seal any gap that may allow someone signaling to enter their lane, regardless of the distance to the merge point. This behavior also occurs in on-ramp situations where drivers are entering the highway and slow lane drivers refuse to move over, but don't want you in front of them!

I have driven in many states and never has the mentality that you cannot, MUST NOT, let someone ahead of you been as bad as it is here. Minnesota Nice, indeed.

I live in the Twin Cities and "Minnesota nice" means merging incredibly early when a lane is posted as closed. As a transplant I can always count on getting and extra 1/2 - 1 mile ahead by staying in the closed lane.

It is getting better, but still 40% of the time the car at the end of the zipper merge is NOT happy to see you, and thinks they are letting you in.

Signage is helping at the big construction areas.

"Getting ahead" means you're not following the whole point of this idea, which is to MINIMIZE the speed differential between the two lanes.

If you want to show people how it's done, you go the same speed as the cars next to you in the open lane, and zipper merge at the end.


Yes, THEY are doing it wrong, because they're early merging. (IDIOTS!) But you are ALSO doing it wrong.

I'm reminded of the George Carlin comment that everyone who drives slower than you is an idiot, and everyone who drives faster than you is an asshole.

"Getting ahead" does not mean speeding, and its hard to go the same speed as the full lane when it is at a stand still. I do slow down from the speed limit significantly just in case someone wants to pop out.

The whole point of zipper merging is to minimize speed differences between lanes for maximum efficiency and maximum safety.

The proper way to execute a late zipper merge is to match speeds and merge at the end.

You are not correctly zipper merging if you are passing other cars before merging at the end.

dusing is probably talking about the (rather common in some places) situation where one lane has half a kilometer of slow-to-standing vehicles, while the other lane is entirely free.

In this situation, of course you're going to pass those other cars (please do so slowly and carefully if you pass them on the right). It would be ridiculous to slow to a standstill with half a kilometer of empty road ahead of you.

When zipper merging works correctly, you usually cannot pass cars in the other lane, because traffic automatically flows in the right way. So in a sense you are right that if you pass cars in the other lane, someone is doing the merging wrong. That someone are the people in the crowded lane, however, not the one of who is passing them in the empty lane.

Thank you for your charitable interpretation of my comment, but in fact I don't think it is ridiculous to drive slowly in the right lane. I really do think that's the correct strategy.

Even if the second lane is free, it's safer and more efficient to match speeds with the first lane. And shouldn't we take actions that maximize safety and efficiency? Note that there is no global gain from speeding down the right lane, since the bottleneck is what controls the flow.

I suspect many of you will disagree with me, so let me address some possible concerns.

Is it safer? Yes. Generally driving is safer when all vehicles are moving at similar speeds. Relative speeds being low mean that there's plenty of time to react to danger. It's also safer because other drivers are more predictable when going at a single speed.

Is it more efficient? Yes. The bottleneck is constraining the flow, so gains before the bottleneck end up being useless. The important thing is to make sure that traffic is entering the bottleneck at a fast, consistent speed. Predictable and uncontested merges are the best way to stop the traffic lane from slowing. Going at the speed of the left lane offers two benefits here. First, it's easier and more predictable to merge when you travel at the same speed as other traffic. Second, if you travel at the same speed as other drivers, they won't get annoyed at you and attempt block you out (to everyone's detriment).

Is it weird? Yes. Driving below the speed limit with space in front of you is an odd thing to do. But in this case, you're matching the speed of traffic and there's no global opportunity cost because the bottleneck ahead is the constraint, not your speed. It's weird, but it makes sense.

The reason that the zipper merge is recommended by so many travel departments is not because it's straight-up more efficient than other merging strategies - it's because it's more robust to defection.

But just because it is robust to defection, that doesn't mean the optimal strategy is to defect.

For those of you who disagree, why? Could you point out where you feel my logic fell short?

Hmm, probably we don't disagree by much anyway. Perhaps it's clearer if we put some numbers to it: Suppose that the occupied lane is doing stop-and-go (due to the natural waves occurring in traffic), with a maximum speed of 20km/h.

What should the speed be for somebody in the unoccupied lane? Certainly 80km/h is far too much. On the other hand, it also doesn't make sense (IMHO) to come to a complete stop in the free lane just because cars in the occupied lane have to stop. I feel that a steady 30km/h is a good compromise.

Does that sound reasonable to you?

If the left lane is going an average speed of 10 km/h, then I believe the right lane should drive at 10 km/h. Doing otherwise is being selfish and unsafe, with no global benefit. It feels weird, but it makes the most sense as far as I can reason.

He is helping to correct traffic. Everybody else is messing up the zipper merging by not filling in both lanes equally.

No. The way to correct traffic is to stay in the right lane, moving at the same speed as the other lane, and merge at the end. That is what the zipper merge is. Driving at a faster speed than the other lane does not correct anything.

The zip line is always free though, so obviously you're going to be going faster then the other lane that is backed up with traffic.

Again, the correct way to zipper merge is to reduce your speed to the traffic beside you and then merge at the end.

The whole reason zipper merging is efficient and safe is that both lanes of traffic are moving at similar speeds, making merging quick and easy. If you zoom past the other lane of traffic, you are actually worse than the early mergers in terms of safety and efficiency.

Edit: For those of you downvoting me, could you please explain your reasoning? Do you disagree with the content of my message or its delivery?

You need to have the same speed at the point of merging, not while you're just driving 1km of empty lane leading to the merge point. It would be very silly to drive at a snail pace in an empty lane for no reason. You just have to slow down to the other lane's pace when reaching the merge point.

What would be the point of matching speed with the other lane when you're not actually about to merge?

His reasoning is safety, because it's supposedly (and I guess understandably...) unsafe for cars to be moving at very different speeds next to each other. I assume what people are afraid of (or what actually happens) is that people in the lane moving slower might drift/shift into the lane moving faster and cause a collision. When this happens while both lanes are moving at similar speeds both cars are more capable of avoiding it.

I've been there. I've been in the backed up lane in traffic with the lane next to me moving much faster, near the speed limit. I often try to get into the fast moving lane and it is a bit scary because it can be difficult to see far back enough to get into the lane without cutting somebody off. I still believe it is the fault of the people in the slower lane for being there though, and it is their responsibility to stay in that lane or find a safe opening to get to the next lane. If I see a bunch of stubborn early mergers, I will assume that is where they want to be and I will drive past them like a bunch the bunch of parked cars they are. But I will indeed not necessarily drive 80+ mph and I will pay close attention to the movements of the cars.

As you say, there is a lot of room between "drive 10mph stop and go next to the backed up lane even though everything is wide open in front of you" and "drive 65mph next to the lane going 10mph."

Your speed should not be more than 20mph different from the lane next to you. That's probably at or even past the limit of a safe speed differential.

There are three reasons to drive slowly in the right lane.

(1) It's safer (matching speeds reduces odds of and damage from collisions)

(2) It's more efficient (If you zoom ahead, there is a nonzero chance that drivers in the left lane will be annoyed and attempt to block you, reducing global efficiency. Note that there's no loss in going slow since the bottleneck is the constraint. As a result, going slow can only gain and never cost.)

(3) It's fairest (people who arrive at the same time will leave at the same time, regardless of their lane)

Another comment lays out my logic further: https://news.ycombinator.com/item?id=8084137

I know it feels silly to drive slowly in the right lane, but it's safer, more efficient, and fairest. Why wouldn't you do it?

>'I have driven in many states and never has the mentality that you cannot, MUST NOT, let someone ahead of you been as bad as it is here. Minnesota Nice, indeed.'

We don't rule the lists anymore [1], but I'll put the horrible traffic and drivers of the DMV (DC, Maryland, Virginia) up against anyone in a gridlocked struggle for the title of worst.

Though, I've routinely been told by people more traveled than myself that we hardly rate on an international scale.

1: http://www.weather.com/news/commuter-conditions/worst-traffi...

From my experience, South America, Africa, and the Middle East have horrifying traffic behavior, but at least everyone follows the same unofficial rules of the road. Eventually, you can somewhat reliably predict how someone on the road will behave.

Different areas in the US have very different driving cultures, and the transient nature of the DC area means that drivers bring their habits from wherever they came from, and so everyone is obeying slightly different driving cultural norms. It's a mess.

Grew up driving in DC/MD/VA. I think PA/NJ/DE is much worse.

That's funny. I grew up driving in PA then moved down to DC for a job and I think DC/MD/VA drivers are much worse. It's like none of them have ever driven in the rain before. "Holy shit the road is made of ice, better drive at least 15mph under the speed limit." - The thought process of every DMV driver when it starts to drizzle.

I'd imagine zipper merging situation would change dramatically as soon as long trucks are involved.

If everyone maintains speed it's actually not a big deal.

Unless they didn't maintain enough space for the big truck. Then someone has to slow down.

Or unless more than one lane is closed, or there's, you know, actual traffic. I would love to drive in a place where "you don't even take your foot off the brake" works, closures or no.

As a driver in the Twin Cities area, I can back the stereotype that Minnesotans don't know how to zipper. It speaks so truly of the Minnesota condition.

People begin often merge far too early, and get offended when folks like myself "budge" at the front of the line.

Learn to zipper, darn it!

Edit: This TPT parody, How to Talk Minnesotan, shows how to drive in Minnesota (scrub to 11:58):


If you ever lived in rural Minnesota, you know this is completely dead on.

People think they are being nice. Education can work very quickly and efficiently on such people by showing them they aren't being nice after all.

I live in the US in a "nice driver" area with a number of traffic circles. On a fairly routine basis I come up to a traffic circle and stop because there is someone in it coming towards me, but then they try to "helpfully" stop and wave me in, slowing down themselves, everyone behind them, and even slowing me down as now I'm confused and less likely to move. Or I get stuck behind such a person in the circle and have to stop gracelessly as they unexpectedly stop in the middle. (This being a traffic circle it's not that dangerous due to lack of speed, plus over time I've learned to predict who is going to do this based on the car body language...) I haven't yet witnessed this causing an accident but it's just a matter of time, it snarls things up that badly. They think they're being friendly. If someone would explain that "friendly" is using the traffic circles as designed, they'd probably do the right thing in the future, but so far I haven't discovered the hand gestures that properly express that.

Someone died today because someone didnt know how to merge (the on ramp is to get up to speed) and probably because someone thought they'd be 'nice' and slowed down. My guess is both vehicles at the merge point slowed drastically and everyone on the interstate were too close to each other for the speed and volume of traffic. Sad. http://www.wxow.com/story/26101450/2014/07/24/breaking-part-...

Edit: or perhaps we should fault inattentive driving on the truck driver. Still sad.

It sucks because the speed limit through there should probably be 40 not 55 when traffic is heavy. The problem then is people on the interstate still tailgate and/or don't leave adequate room to allow zipper merging.

Edit: to clarify, the slower speed should be that low only now during construction and lane closure.

If it's a "traffic circle", you may be the one messing it up because entering traffic can have right of way. In a roundabout, it's the opposite. Terminology is important here because they are two different things.

Roundabout, then. Terminology will probably cease to be important as "traffic circles" as you describe them are dying if not dead.

Generally I've found that most drivers in the Twin Cities have not acclimated to "city driving" the way many of us from other areas learned. People here drive as if they are exempt from the responsibility of piloting a 2 ton death roller skate and what it could possibly do to others while they text their friends.

Since the big obstacle to zippering is people's perception that it's rude (since you appear to be "pushing to the front of the line," as it were), one could argue that this is a case of Minnesota Nice hurting more than it helps :-D

"Minnesota Nice" tends to be an in-joke according to many of the people I've asked about it here. :)

If you are offending people, you're likely performing the zipper merge incorrectly.

The point of the zipper merge is maximize efficiency and safety by matching the speeds of the two lanes.

The wrong way to zipper merge is to drive past the other lane and merge at the end.

The correct way to zipper merge is to drive at the speed of the nearby lane and merge at the end.

When you zipper merge correctly, almost no one gets upset.

"Optimal" means different things to different people. What the DoT may be trying to maximize (safety, lack of frustration) is generally not what drivers want to maximize (overall speed) and I think there's confusion about that when zipper merge is discussed. The linked article uses the word 'faster' but the Minnesota pdf they cite actually says "Our analysis has shown that the Zipper System has no effect on travel time through the work zone."

Here is a Michigan DoT paper[1] from 2007 that analyzed zipper merge studies in various states and countries. Only 1 out of 6 showed statistically significant increase in throughput.

"It was found that the system did not increase the throughput or decrease the length of the queue." (Netherlands)

"It was found that the throughput did not increase, but the queuing was more efficient and there were less frustrated drivers." (UK)

"Beacher, Fontaine, and Garber (12) found that there was an increase in the throughput and a decrease in the queue length; however, the difference was not statistically significant." (Virginia)

"However, flow did not appear to be significantly affected by the dynamic late lane merge system." (Kansas)

"The data from both of the studies shows that the typical queue length decreased and the lane usage were nearly equal, but the throughput decreased slightly." (Minnesota)

"The results from Maryland’s deployment show an increase in the overall throughput, a reduction of the maximum queue length, and a more even distribution of volume between lanes."

1: http://www.michigan.gov/documents/mdot/MDOT_Research_Report_...

The problem is, there are two very different ways to merge based on the flow of traffic. If traffic is moving, then early merge is the preferred method. If traffic is not moving, then a late merge makes more sense.

The problem is people who do not merge early when traffic is flowing. These people wait until the last second to merge instead of finding an opening while traffic is flowing and force traffic to slow down in order to let them in.

And the really real problem is that the average driver in the USA has no idea what they are doing and should probably not be driving at all. It would take a huge change in the way we train drivers to get them to recognize when they should use each type of merge. And forcing people into one way to merge or the other is not really the "fastest/safest" way, because it varies on conditions.

I have to deal with such a merge almost daily as the main road I use is under heavy construction. I am an "early merger" mainly because the people in the left lane (right lane goes away) will not let you in if they have a choice and it is widely seen as being a "douchbag" if you fly down the right hand lane expecting to be let in at the end. I'm not saying that I disagree with this article in fact I've sat in near standstill traffic in the lefthand land and I've had plenty of time to wonder if merging at the merge point does make more sense, which I think it does.

I personally wish they would have closed off the entire right lane as there is only a couple hundred meter stretch where this road is 2 lanes and so I wish they had just kept it a single lane all the way though while construction was underway. Here is the road in question: http://s.joshstrange.com/oRq9.png From where you see Louie Pl it turns into a 2 lane road then goes back to 1 lane at Opportunity Way then back to 2 lane where you see it get fatter again. However construction makes the road for a mile or two after Opportunity Way 1 lane only so the 2 lanes for such a small stretch are almost more trouble than they are worth.

As a driver in the UK, I don't understand what exactly the difference between a merge and an overtake is. It is quite common here for roads to change from 2 lanes to 1 lane. There's no signage telling you to merge. There's no rules telling you what to do. You just operate by the same logic as an overtake, and you'll find that, quite naturally, everybody 'zip merges'. Perhaps the reason for the inefficiency is having rules in the first instance.

The difference is localized. The UK is much smaller, and therefore has less regional tastes about merging. In the Midwest and Eastern seaboards of the US you have polite and diverse people respectively. In California, you have experienced drivers. California works like the zipper merge already, and the rest of the US is flummoxed by what to do - depending on where you are.

No, California traffic is just so stationary that it devolves into zipper merging ... sort of.

Zipper merging also generally only works when there is movement after the zipper. In many places in the US, the point where a zipper is occurring is followed by more traffic injection (um, there was was a reason for losing that lane, after all, and it was probably the fact that there was a new lane coming in) in less than a mile and any advantage is lost.

I don't think it's so much about being smaller, more that there is only one set of rules for the whole country and those rules are more strongly enforced. Also, from what I can tell, the driving exam is far more rigorous in the UK.

Speeds in the UK are way slower than the US, and cars are way more expensive to run in the UK than the US.

Consequently, everybody and their dog doesn't expect to drive a car in the UK.

Demonstrably and comically false (except on the running costs front - you're spot on there). Not only are UK motorway speed limits 70mph (and traffic speeds higher still), 70mph is also the limit on most dual carriageways, and major undivided roads have a 60mph limit. In the US you'll rarely find a road with no central divider with a limit above 45, and only major multilane highways go to 65 and higher limits (depending on state).

I'm not sure we drive any slower in the UK. A residential road is 30mph (very rarely 20mph), country roads are 60mph and arterial and motorways are 70 mph. That seems comparable with speeds in the US.

Also a UK driver. Before reading this article it had never occurred to me that there was any other way of merging on lane reduction.

So Negitivefrags is probably right that the best top-down way to get this to happen is to place signage. But it's sort of fun to think of the best bottom-up way to get this to happen, assuming your local government isn't putting up signs.

For example, I love going a constant speed in start-stop traffic (letting gaps appear ahead of me then closing them as traffic stops) so as to smooth out the ride of everyone behind me.

So what's the optimal way to get this behavior to spread? It's probably not just to gun it to the front of the line and yell 'zipper!', everyone just thinks you're an asshole.

VikingCoder suggests "If you want to show people how it's done, you go the same speed as the cars next to you in the open lane, and zipper merge at the end." I love it, because it's really awkward, and people will note it enough to maybe think through what's happening, what your logic is.

But I dunno, maybe there's something else more clever out there, too.

Locally (metro NYC), I've heard this referred to as "alternate merge" and it's usually mentioned in the context of someone saying something along the lines of "there was a traffic jam due to an accident and then when it was MY TURN to go, the person in the other lane didn't let me in". Usually there's a comment about how the plates were from out of state, too.

This is not to be confused with, e.g., a road that splits and the left lane split it moving fine and the right lane is backed up and a driver in the left lane waits until the last possible second to move over, often times coming to a complete stop, thus backing up what should be a free-flowing lane. No, in that case you didn't move over in time, you don't get to slow down traffic, you take the lane you're in and keep going, you'll have to take the next exit.

This works pretty well where I live. Here in Vancouver, a major bridge that takes traffic into the downtown core is only 3 undivided lanes. There's a huge amount of traffic flowing over this bridge in the morning as it connects work to home for somewhere in the range of 60,000 - 70,000 commuters daily. From the residential side of the bridge, traffic zipper merges from 4 lanes into either one or two depending on the time of day. The traffic is coming from two different sources (west and east) so each source will zipper merge and then the zipper merged lanes will zipper merge onto the bridge. I honestly have no idea what this would look like any other way.

Zipper merges are obvious and natural if you set them up right - what it takes is the knowledge to do it. Once people know that they should be yielding to the mergers, the whole merge process speeds up considerably.

I see this as well whenever crossing the Lions Gate southbound. I suspect if there weren't 4 lanes merging together it would be a different situation (I don't see people doing this correctly in other parts of the peninsula that merge).

I thought the zipper merge was the standard way to merge? You stay in your lane till the merge section(end of your lane) and its every other car. Maybe its a California thing.

I've rarely seen people in (northern) California do proper zipper merges. CA has some of the most clueless and inattentive driving I've ever seen.

I rarely have seen issues in California with zipper merge, when the lanes are actually merging. Most of the time the issue I've seen is when people are cutting over from a turn only lane to try and force their way to go straight.

Turn-only lanes (using a turn-only lane to go straight) is where I see the worst cheater/vigilante behavior. (It's always BMW's who do it, too... )

Those turn-only lanes are just wasted capacity that could be used to carry more traffic into a zipper merge. I'm guessing that at some peak hour the turn lane has utility, but at reverse-peak they should be able to switch to a regular lane leading to a merge point.

Zipper merging inside an intersection will almost certainly lead to gridlock due to incomplete mergers before the light changes.

Referring to exit-only lanes on highways, not signaled intersections.

Yeah, that situation is where I see it 2nd-most; first most is people trying to change lanes into an exit lane right at the exit when they're in a non-exit lane and want to exit. Those are lane changes, though, not merges.

Disagree. I used to reverse-commute through the Golden Gate Bridge and the Caldecott Tunnel, where the freeway went from 4 lanes to 2 in both places. 99% of drivers manage to zipper-merge properly. Likewise, almost everyone zippers at the South 101 Central Freeway merge.

California drivers may or may not be bad, but they're much better than the ones in Minnesota.

I wish some of the "exit only" lanes would change to be zipper-merge-able-- in SF, think about the 7th street exit on 80E just before the Bay Bridge-- two empty lanes!-- and lots of "cheaters" whereas with a zipper merge everything would move faster.

IIRC California officially endorses early merging, though?

You could just make two exit lanes in that case, and merge them as or after they exit. People exiting at the last minute from a non-exit lane though aren't really merging, but rather changing lanes. And it doesn't really result in a proper zipper merge since not everyone is exiting, so you can't predictably alternate cars.

I guess I mean that they shouldn't be exit lanes at all, but should be used as extra capacity to facilitate zipper merges. In the case above, the exit lanes stretch back almost 1/2 a mile, which is waaay above the actual demand for that exit, esp in the evening hours.

Oops, I read your case backwards (I'm not familiar with that exit). I thought you meant that there were too many people exiting for the one exit lane, so you were proposing people from the 2nd-to-right lane should be able to zipper-merge into the exit lane right before it exits.

The reverse (merging out of the exit-only lane to stay on the freeway), I think usually works better with a late-merge, but you do still have the problem that you can't do a proper zipper merge, because the lanes aren't fully merging, so you can't alternate one-right/one-left. Instead some cars want to lane-change and some want to continue in the lane, which adds at least a little complexity.

It could be partly improved by signage changes though, it sounds like, to just mark the exit lane later.

I still commute occasionally across the bridge. Getting people to zipper merge 4 to 2 lanes properly going onto the bridge almost never happens.

SF drivers are great at the waiting till the end of the lane part, but not about the letting you in. Sometimes I have to wait for three or four cars to go by before someone notices me.

Then again people also try to pass on the right by entering the merging-in lane and immediately getting stuck. More than once I've seen people try to pass inside the merging lane by going on the shoulder. Is it something about the 358i?

Same in PA. I've always remembered merging like this.

Since this is a solution that requires "everyone" to change their behavior, my guess would be that it'd be more productive (and faster?) just to wait for self-driving cars to be prevalent enough that they "set the tone" for what constitutes normal driving in these situations.

I've recently sold my car (a crappy Opel) and replaced it with a motorcycle, and my behavior in traffic has definitely changed re: merging. The other day there was a mile-long queue in front of a highway exit, traffic was up on the hard shoulder or whatever the term is. If I were still in my car, I would've been polite and just queued up - would probably cost me 15 minutes to get to the exit and further on.

Now though, I did full D-bag and just drove past most of the line. At the exit, the queue started breaking up, leaving plenty of room for me to merge into.

Actually if everyone took a motorcycle instead of their car, most of these problems would be gone. But that's probably a silly argument.

This sounds like Italian driving.

If you haven't driven in Italy, expect zero braking during merges. I really do mean zero.

Trucks will move over (at speed), cars will move over (at speed). The "zipper" action is seamless and fluid, even when the vehicles are temporarily extremely close to each other.

Just look down the road, see a merge may happen. If you are a truck move over, if you are a car you may need to accelerate to move over.

The first time you experience this it looks and feels scary, especially as people merge with only 50cm-1m between vehicles. But once you're used to it you realise that you virtually never slow down and merges are really quick and easy.

Late merging works fine in Seattle, and in my experience the lane that's closed moves substantially faster than the open lane because Seattle drivers will actually let you merge in. That's not how I recalled it working when I lived in Texas though. In Texas if you got caught in the closed lane you were pretty much stuck there for a while, while the guys in the open lane whizzed by too fast for anybody in the slower (possibly stopped) closed lane to merge safely.

If the person merging late had to stop, then it means they went too far without getting a space (because they didn't take the chance they were given or no one gave them space). This is the problem with zipper merging: if everyone gets it right it's beautiful but the failure mode is near catastrophic (effectively stopping one lane) and easy to get into. One or two bad actors can ruin it.

...and in my experience the lane that's closed moves substantially faster than the open lane because Seattle drivers will actually let you merge in.

I think that's actually because more people are early-merging than late-merging. So you have more car volume in the open lane than the soon-to-be-closed lane leading up to the merge. Definitely a way better situation everyone early-merging, but sounds like there's still room for improvement: ideally both lanes should move at about the same speed.

Seattle drivers are some of the worst mergers on the planet. You'll see a half-mile-long line of early mergers and then an entire unused lane. Absolutely blows my mind.

Traffic management is a fascinating intersection of sociology, fluid dynamics, behavior change, and urban planning.

If you're interested in this field, I can't recommend the work of Tom Vanderbilt enough. His book "Traffic" brought urban planning to the masses in the same manner "Freakonomics" made social and public policy accessible.


I've always found it fascinating how a single bad driver can cause a massive traffic jam ("ghost traffic jam"). I think this should be taught when you're learning to drive as part of some sort of mandatory basic training - zipper merges should be in there too.

Critical Mass by Philip Ball has some nice bits on traffic.


I'm always blown away but how smoothly some areas handle congestion when overpasses are being built or repaired.

A lot of people and a lot of articles discuss resistance to this as a matter of misguided "politeness", and a LOT of the support for it carries a tone of smugness---there is definitely a weirdly emotional component to this.

But here's the fairly objective thing I want to know: the people who have been (very negatively!) described as "early merge vigilantes" generally operate by staying in the lane that goes away but pacing the cars in the other lane, blocking the people behind them from zooming ahead; and then quickly merging just before the lane goes away (usually an enabler in the other lane will make this work better so neither has to slow down or stop at this point). So here's the question: isn't this exactly what the zipper merge is supposed to accomplish, with both lanes travelling the same speed? Why is this so-called "vigilante" behaviour being described as bad? If a driver that's smug about their late-merging behaviour being optimal is going noticeably faster in their lane than the lane that merged, doesn't that mean that what they're doing is exactly not a zipper merge?

Perhaps the answer is that the ideal to aim for is not "late" merging, so much as "speed matching" merging: when there is a merge ahead, if you're in the faster of the two lanes (whichever one that is!), slow down a smidge now and then you won't have to screech to a braking halt at the merge point (and you'll also make the flow better for everyone behind you). Zooming ahead because there's a briefly clear section of road may be "late" merging in some sense, but is not the same thing as "zipper" merging and is not improving flow.

The point is also to fill both lanes, so your last suggestion doesn't work very well.

And I've been known to be a "vigilante" before. I'd give it up in a second if people would actually start filling the other lane.

There's a merge here in Seattle -- the southbound I-5 collector-distributor with the onramp to 90 where it merges back into I-5 -- where people zipper halfway reasonably well and I actually routinely stay in the less-filled lane and late-merge.

Washingtonians are awful at late merging/zippering though. Even when people are late merging you get tons of people merging early and then the line moves forwards and it screws over the lane that is being merged into. People just won't simply drive forwards to the merge spot.

Dividers might help in some cases, but when the road is empty and everyone is doing 60+ mph the dividers are going to cause accidents. You could try to put down double lines on the road to block merging, but nobody pays attention to those here either.

Getting zippering into drivers ed and getting them early, spamming the TV with ads, and putting signs up all over the place and carpet bombing the message might work.

Happy to see them doing this.

Oh yeah, and there's definitely an emotional component to it. When I'm stuck behind 50 cars and some asshat in a BMW goes speeding past to late merge the world goes completely red. I would love to stop doing that. If everyone zippers the asshat just has to pick a lane like everyone else, and there's no way to "cheat".

And yes, "asshat" and "cheat" are all in my own brain. I find it extremely difficult to stop thinking that way when I see it though.

It's actually really simple: early merge with cars going the same speed in all lanes is always better. You'll always get maximum throughput in the lanes remaining open, because the merge happens over a distance and so is smoother.

But that's in theory, with people actually cooperating. In reality no matter what education campaign for early merge some few percent of people are just dicks, and when they see an open road ahead will zoom through at twice the speed and this is what causes the merge problem (see the green car in video).

When people espouse late zipper merge it makes me sad because of what it says about human nature.

I think you are misunderstanding what is being called "early merge vigilantes". They are people who merge early and then don't let people who didn't merge early like them in because they are trying to punish them for "cheating".

Well, this is (has always been) the standard in Germany and it works pretty well. We even have one of those great German words for it: Reißverschlussverfahren (= zipper procedure).

Concerning people not letting you in: In case of ending lanes (except ramps) you are obliged to let people in. There is always the occasional douchebag that doesnt but it almost never happend to me. At worst you have to wait for one car to pass.

I see the negative effects of early merging regularly. The freeway onramp closest to my home leads to a frequently congested portion of the freeway. People merging ASAP often end up having to stop on the onramp, blocking traffic behind them which can then back up onto surface streets. They do this despite the onramp merge lane continuing for quite some time after it first connects to the freeway.

Perhaps instead of merging one lane into another, the animosity toward late mergers can be alleviated by making all lanes move.

If no one feels like they are in the free lane, they'll be forced to cooperate.

So, on a two-lane highway, make one center lane and have both lanes funnel into it. then you can move that free-flowing lane wherever you need.

malandrew types faster than me. And has better ASCII art.

This is pretty brilliant, but how do you encourage this without a light or years of including it in driver education?

Even including it in driver's education isn't likely to help. Driver's education teaches people to use blinkers, but from informal observation I'd estimate 1/3 of drivers in my area don't use their blinkers at all.

In Minnesota there are posted signs such as "Stay in your lane until the merge point" which seem to work pretty well.

In contrast to the states promoting zipper merging, Indiana recently redesigned some interchanges to effectively eliminate the single merge point. They added extra lanes so that one "preferred" lane (the one that formerly backed up everyone for three miles) is no longer the sole access for any of the off-ramps. Now, everyone has basically three miles in which to merge into, out of, or across this lane, rather than piling into it all at once.

This picture doesn't show the full extent of the redesign, but the yellow highlighted part is the former problem lane.



Alternatively... Since the government forces me to bring my vehicle to one of their locations every other year for emissions testing, have them install a speed control unit with a beacon to communicate with other cars. (Yes it will be harder to install on some cars than others -- even just the beacon would be valuable.) Over the course of five years, retrofit every vehicle with a beacon and have it already installed on new cars, have severe penalties for vehicles caught driving around not broadcasting their signal, and now everyone can have a bit more automation in their automobiles since the systems will enforce safe merge gaps and other nice things like computer-enforced speed limits. Plus it prepares everyone for a fully automated experience that can come sometime in the 2020s.

I can dream right?

"Merge rudely."

It's only rude if no-one else is doing it (i.e. you're speeding ahead of everyone merging early). If everyone stays in their lane until the merge point, no-one's zooming ahead, so it stays fair.

On most military bases, and therefore in military heavy towns like San Antonio, the zipper merge is used to great effect. I've even seen it used at T-intersections (where the perpendicular road must stop) - on almost every military base, when traffic is backed up, you'll see drivers immediately implement the zipper - one car from right of way, one from the intersecting road.

I really wish people would actually do this. I call this "driver karma". For every person you let in, you get to go in front of another person. If everyone followed this heuristic, we wouldn't have a problem. Oh, and if you let a bus full of people in front of you, your score is increased by the number of people on the bus :).

This 2008 article in Wired[1] changed the way I merge forever and it has paid off when I drive in Chicago's endless construction and traffic.


Early merge feels more polite, late merge is safer and faster on the overall traffic (and much faster if you're the only one doing it + you're seen as impolite)

They want to make it more common. Pretty sure everywhere in EU everyone late merge. In west USA at least, I only see early merges.

Seems great in theory, but what if the other drive isn't willing to let you in - or - vise versa?

The nicest thing about zipper merging is that everyone can see the fairness of it, when it works. It's the lack of fairness that most infuriates people with the free-for-all system.

Minnesota still does a very poor job of letting people know where the expected merge point is.

I grew up and learned to drive in Washington; I moved away 20 years ago but come back regularly, and if there's one thing I know about Washington drivers it's: they can't merge for shit.

With the low cost of portable lighting and signage why not traffic light both lanes and alternate who goes through, just like we do with highway congestion on ramp lights?

I've never understood early merging. Why eliminate a lane of traffic before necessary.

In Germany we've been doing this for decades. Welcome to the future.

I'm the asshole who merges in at the last possible second. It's really annoying that they're doing this as I routinely save 10 minutes off my commute by jumping into lanes that will eventually merge in.

Want faster and safer traffic merging?

> Part of DMV road test should cover highway and emergency (road block) situations. > Utilize responsive testing (how fast person reacts to something). Could be an in-house application. > Older drivers should have extra testing once they hit certain age. > Re-test drivers every 10 years. > Driving and doing your make up? Reading a book/iPad? Eating chipotle? License suspension. Bye bye.

In general every situation where I had to steer out of the way was caused by an older person who should NOT be driving.

Zipper merging on New York roads? Fuhgeddaboudit.

> In general every situation where I had to steer out of the way was caused by an older person who should NOT be driving.

Do you have any statistics about driving accidents? Because all the stats I've seen show that older drivers are safer better drivers than young people, and this is especially true when comparing to young males under the age of 26.

Those statistics do not include cases where older person switches lanes into your car and you luckily avoid it.

They also do not include cases where other people cause accidents and go on their marry way while you wreck your vehicle avoiding them.

Like the other poster stated, it's young (novice drivers) and old drivers.

It's also a scientific fact that you reaction time decreases as you grow older.

>Because all the stats I've seen show that older drivers are safer better drivers than young people.




> Age-related declines in vision and cognitive functioning (ability to reason and remember), as well as physical changes, may affect some older adults' driving abilities.5

> Per mile traveled, fatal crash rates increase starting at age 75 and increase notably after age 80. This is largely due to increased susceptibility to injury and medical complications among older drivers rather than an increased tendency to get into crashes

> This is largely due to increased susceptibility to injury and medical complications among older drivers rather than an increased tendency to get into crashes

So it's not that they are more likely to get into crashes, it's that they are more likely to die from the injuries that a crash causes while a younger person would survive if they received the same injuries.

It's been awhile since I've looked into it, but I seem to remember it being more or less that the young, and very old are both less safe than the middle.

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact