Hacker News new | comments | show | ask | jobs | submit login
How traffic actually works (jliszka.github.io)
263 points by jliszka 1336 days ago | hide | past | web | 181 comments | favorite

What an arrogant article. How can you title something 'How traffic actually works' when you don't examine actual traffic. Instead, make up a simple model, run some code, claim it shows what you want.

It's not like modelling traffic is so insanely hard that we need such simplistic models. Pick as complex a model as you like; we have the computing power to simulate it!

He's derived the fundamental diagram of traffic flow: https://en.wikipedia.org/wiki/Fundamental_diagram_of_traffic....

Up next: why F = ma really, really matters for space launches. Experts don't know this!

"1 weird trick for successful space flight"

Lockheed Martin hates this guy!

If you RTFA, he compares the results of the model to actual data collected by the Federal Highway Administration. Why use a complex model when a simple one suffices?

Not a fan of the use of "RTFA".

What author is trying to argue against is the idea of "smoothing-the-wave", which he does prove in his overly complicated article.

Yes, there is too much rambling for making a point, yes the tone of title is quite arrogant / link-baity.

But this terrible essay did prove the point: smoothing the wave is not going to make you go faster.

I'm not sure who actually thought that "smoothing the wave" did make you actually go faster though, since a moment's thought shows that "going faster" would necessarily involve going through the bumper of the car in front of you. I think it's about a "laminar" traffic flow being a lot less annoying and dangerous to drive in, if you are going to go slow.

I think he fell down on his own argument when he claims that doing that causes a "big traffic jam" downstream of you; how can it, if you're only 10 or 20 seconds max further behind than you otherwise would be? You can only cause that much additional "traffic jam" in the back, and it may well be worth it to create a more laminar flow.

>I'm not sure who actually thought that "smoothing the wave" did make you actually go faster though,

Probably all the people that passed around the critiqued link and suggested that it gives a way to improve traffic flow?

The gist I get from the link is that the act of "smoothing the wave" will not benefit the one who performs the act, but it benefits the drivers behind the lane. You can't fix the traffic jam in front of you, but you can do your part to prevent one from forming behind you.

And that's why people pass around the link.

Right: like I said -- and like the GGP (jerf) just denied them saying -- the claim is that it will hasten the traffic flow. Individual vs group benefits is beside the point.

He utterly fails to make that point. His simple model explains the idea of maximum throughput / congestion.

It utterly fails to explain the extreme variance in real traffic throughput during congestion. The original article does.

OP here. That's a really good point. I just think the original article overreached in saying that zipper merging is a "simple cure" for traffic jams, without taking into account that the road might at or near capacity (and you hit capacity in a hurry when you lose a lane).

My article doesn't account for variance in flow rates in congested traffic, but variance in car length (% of trucks on the road) might explain it. I'm not convinced that merging behavior is the culprit; the original article only speculates that that's the case. I'm putting forward a reasonable explanation for why that isn't the explanation, and a basis for evaluating whether it might be — in particular, whether a zipper merge results in higher flow rates after the bottleneck.

Isn't variance in flow rate the crux of the issue? Merge lanes are a special case in traffic, where another flow is entering the channel. This creates turbulence around the merge point that does not exist where there is pure laminar flow and all lanes can be treated equally. Would it help if "simple cure" was renamed as a "simple improvement?"


Sure, there exists a point where the input flow is great enough that traffic must move slower. OP's entire argument is based off of this. The problem is that there are many things you can do to slow down the flow rate even further, and merging poorly creates this "turbulence" which wastes further flow.

And, it's clear that no improvement in merging behavior can beat the maximum road occupancy, however we can approach that limit much more closely.

I think that conceptually modelling traffic as fluid flows is quite clever, and the "turbulence" idea is particularly satisfying.

FWIW I tried this out, seeing what happens when there are varying average car lengths on the road.

http://jliszka.github.io/assets/img/traffic/speed-vs-occupan... http://localhost:4000/assets/img/traffic/flow-vs-occupancy-2...

Also added as an update to the article.

It also won't make you go slower. And rolling at a constant low speed is much less frustrating than constant stop and go traffic. Especially for people driving standard.

It's also more fuel efficient.

I've always tried to smooth the wave on the fuel efficiency argument. I figure that maintaining a constant low speed has got to use less gas and be easier on my car than all the stopping and starting.

I think lots of people, on both "sides" of this debate, have predetermined notions and are trying really hard to justify them.

The initial point about rate of vehicles through a specific point per lane per hour couldn't be more simple and valid.

Agree 100%.

From the article:

>Suppose you’re on a 2-lane (each way) highway and one lane is closed up ahead due to construction. Now the flow rate of your lane is cut in half (or there are twice as many cars in line in front of you, depending on how you want to look at it).

What I observe is that the speed is reduced to one twentieth of the speed, not just half. This is because people are merging very slowly while jostling resulting in needless braking. If everyone could agree upon a proper zipper merge, the speed for everyone would go up many times.

Or imagine a traffic signal at the merge point that only lets one lane go through for a minute each. The overall flow rate would be much higher than what it is without such a signal.

Discounting this based on theoretical flow rate (as if removing one lane reduces real traffic flow by only 50%) shows that the author totally ignores real traffic scenarios completely at odds with the title of the article.

I think you're may be thinking of this as one mechanism instead of two separate mechanisms.

You are dealing with a queue as well as a through-rate at the merge-point. The through-rate with one lane can still be 1/2 of the through-rate with two lanes, but because you have a queue waiting to reach the merge-point you can end up waiting much longer. More spacious merging will not change this because of the principle stated in the first paragraph of the article.

Increasing the speed at the merge-point will not decrease the queue. It will only decrease the density of the queue but move it back further in traffic. Your time to cross the merge-point will be basically the same.

>Increasing the speed at the merge-point will not decrease the queue. It will only decrease the density of the queue but move it back further in traffic. Your time to cross the merge-point will be basically the same.

I have to disagree with this. If you increase the speed at the merge point, someone who is newly joining the queue will definitely cross the merge point in lesser time than with lesser speed at the merge point.

That doesn't matter though, as if they're going faster they're more spaced out. Only 2000 cars per hour can pass the merge point

Sure, but when traffic jams happen you often see not just ONE lane of traffic slowing down at the merge point but ALL lanes of traffic slowing down. On a four lane freeway narrowing to three lanes it's not the case that the two right lanes are stop & go and the two left lanes are 80MPH. EVERYONE slows to 10MPH. Furthermore the notion that you'll get 2000 cars per hour at the merge point irrespective of speed is ludicrous. Once everything slows down people act really douchey and don't let each other merge, etc.

I have witnessed eight lanes of traffic slow from 70MPH to 10MPH over a single poorly designed merge when there was more than enough aggregate free space for the merge. That happened because drivers don't accelerate hard enough on onramps and people don't redistribute themselves prior to shitty merges.

Show me a society that has no traffic jams and I'll show you one that's ready for socialism. Or vice versa.

It's not about velocity it's about rate of cars through a point per hour. If you have a mergepoint with 10 lanes. Each lane can support say 20 cars per minute. If you have 200 cars approaching the merge-point per minute the cars can travel effectively at the speed of their choice.

If you close one lane, reducing the capacity of the mergepoint to 180 cars per minute while 200 cars are approaching a queue will build. The speed with which the cars mass the mergepoint is not relevant because the rate of cars per lane per minute will stay basically at 20 cars/lane/min.

As to your point about all lanes slowing down - cars will always redistribute as you can imagine. People tend to merge left as there's an additional traffic stream merging on their right. The writer made points about the capacity of the mergepoint vs. the cars approaching- not individual lanes and speeds.

Indeed. The more interesting points to be discussed are

1. If the flow rate changes with speed, thus merging at 5 mph instead of 50 is bad.

2. What will prevent accidents. I suggest start-stop traffic causes more accidents than free flow.

3. Fuel efficiency, which would depend on many things, eg the percentage of cars in traffic that stop their engines when the car is stopped.

In the context of the article your first point is the key question. The author argues that the flow rate is relatively constant even with varying speed. I would tend to agree that it would be fairly constant but if there was extremely efficient merging I could see the flow rate increasing by a few percent. Theoretically the difference in flow rate if cars maintain a 2 second delay from the previous car should be the difference in the length of time that it takes the actual length of the car to pass through the merge point.

So if the car is travelling at .5 carlengths per second then the duration per car should be 4 seconds whereas if the car was travelling at 10 carlengths per second the duration per car should be only 2.1 seconds.

Since occupancy determines flow rate, there’s not much benefit to trying to “cancel out” a traffic wave

This is so fundamentally mistaken as to make the rest of the analysis useless. Traffic waves account for the majority of highway traffic (excluding on and off ramps), and cause the flow rate for a road to be decimated. It should be obvious that avoiding rate decimation is (literally) an order of magnitude more effective than preventing people from merging in front of you.

Road signs commonly ask you to use both lanes up to the point of the bottleneck. That’s reasonable advice, but it’s not going to get anyone home faster.

No, but if your line is twice as long it is twice as likely to cause contention on the roads behind you. ie, if an exit road has a traffic line twice as long, it is probably causing traffic on the non-exiting highway. So, do everyone a favor and (a) don't tailgate (b) ignore people merging in front of you (c) don't drive in rush hour, and (d) ignore traffic advice from people who don't drive.

(http://tomvanderbilt.com/traffic/the-book/ has actual good advice)

>Road signs commonly ask you to use both lanes up to the point of the bottleneck. That’s reasonable advice, but it’s not going to get anyone home faster.

If you simply kept reading:

>This is good because it is less likely to affect other traffic by spilling out onto onramps and surface roads. Also maybe there’s someone on the highway who’s planning to exit 3 miles before the bottleneck. If the backup is 2 miles instead of 4 miles, that person doesn’t have to wait in traffic.

His point is a little different: that someone exiting the highway might exit before reaching the traffic if the line were shorter. My point is that a line on an exit ramp which reaches back onto the highway will cause additional traffic for everyone on the highway. Since his model doesn't take traffic-causing-traffic into account, he never reaches my conclusion.

Exit lines are particularly costly because adjacent lanes of traffic generally only sustain a 10MPH difference in speed. Once the right lane is stopped, a 4 lane highway has a maximum safe speed of 30MPH in the left hand lane. This reduces the flow capacity of the highway and may cause the left lane to lock up because left lane drivers get frustrated at slow speeds and are the more likely to tailgate and cause traffic waves.

His model also is only concerned with the self: "can I reach my goal any faster", and he reaches the mostly correct conclusion that he basically cannot (even though avoiding traffic waves by driving in the rightmost non-exit lane would help him significantly). He does not appear to value his contribution to the delay of the people behind him, which is where all the really interesting conclusions are found, including the reason why traffic wave busting is beneficial.

There are cases where leaving a greater-than-normal lane spacing makes sense, in the interest of safety. Safety is something utterly ignored in the fascinating analysis in this article. For example, if I'm in a left lane buzzing along at wide-open freeway speeds and the right lane next to me is congested and backed up and not moving, I leave more than normal spacing in front of me. Why? Because it's very likely that someone stuck in the right lane is going to dart out in front of me or the driver in front of me, and very likely that I'm going to need to make a very sudden stop. It's just unsafe to cruise along at 75 mph next to hundreds of cars that are looking for an opportunity to jump in front of me.

The prisoner's dilemma also applies to this. Behavior that is "optimal" for a single individual in terms of getting to destination is not what's optimal for the collective in terms of safety. And optimal behavior for getting to destination is not the optimal behavior for safety or lower stress driving, either.

You probably shouldn't be cruising at 75 mph next to a lane of stopped traffic regardless. Leaving more space between you and the next car wouldn't seem to make that practice significantly safer.

I agree that the optimal behavior isn't universal though. One example I'd cite is that I always leave more space when the vehicle in front me is so big that I can't see what's in front of it since that limits my reaction time to what that one vehicle is doing rather than what's going on further ahead.

I'd argue that safety directly translates into speed as well, as a single accident can shut down 1-all lanes of the freeway and result in several hours of delay for the parties involved. So even being selfish you'd have reason to err on safety's side

Shouldn't you leave less space, to dissuade people from jumping in the gap?

No - because you don't know what's going on in front of you, and you may need to stop suddenly.

If you want to prevent people from jumping the gap, speed up.

If you want to be safe from random edge cases (like someone in front of you slamming on the brakes), slow down.

I'd leave at least a minimum safe distance between my vehicle and the one in front.

But Parent suggested that it's a good idea to leave more space in case someone jumps in. I'm not sure I understand that - if they jump in it's always dangerous because you don't know what the person behind you is going to do and you don't know when the jumper is going to jump in. Thus, you want to try to prevent them jumping. This is achieved not by giving them more room (which encourages jumping) but by leaving less gap (reducing jumping) but still leaving the safe minimum.

The stopping distance of my car (a Lexus IS300) at 70mph is 171 feet. So I know I need over 14 car lengths in front of me for me not to hit anyone with brakes at full power.

Do I actually leave that much space in front of me every time I drive 70mph? Hell no. But the slower the cars beside me are going, the quicker i'm going to hit one if it pulls in front of me. If they were going 70mph too, I would only have to brake down to, say, 65mph, which would should only take a second. But if they're going 10mph, I now have to come to an almost complete stop to prevent from hitting them, which as you can see above is a lot of car lengths.

In every case, if you can't predict what people around you are going to do, it's better to leave more space in front. If the guy behind you is following closely and might rear-end you, leave more space so that you can slow down gradually, instead of slamming the brakes and inviting collision. If the car ahead of you might have to slam on the brakes, leave enough space that you can slow down gradually.

If someone in the lane beside you might cut in front of you, leave more space so that you can see them making their move and react in time. That reaction might be to speed up and prevent them before they change lanes, or to slow down and let them in, but in either case, leaving more space is a better idea than tailgating.

I prefer to be pessimistic about other drivers' ability to judge whether they have enough room to cut in front of me. I'd rather not try to go fast enough to "discourage", as there will inevitably be some person who either doesn't notice me, or misjudges my speed (via a side view mirror) and mistakenly thinks he can cut in front of me safely.

A lower speed increases the chances of this, but reduces the risk of a catastrophic result. ;)

No! Read the original article this author is trying to critique.

The central assumption for this argument is that the average time between cars is fairly consistent no matter the environment; small changes to this time do not matter.

However, on a congested road very small changes can have huge effects. IIRC, one study showed that a 1% decrease in the number of vehicles resulted in a 28% decrease in commute time.

This. I've studied traffic extensively and if you ever wonder how it works, it is actually quite similar to any other network. Imagine road traffic is the same as traffic through your servers. There are hard bottlenecks at various points, but in practice what matters is planning for variability. Real-life situations like someone cutting you off, people rubber-necking, the natural commute tendencies etc tend to be much more impactful. Unlike with servers, you can't just spin up a new lane, and "packet loss" is pretty much unacceptable :)

EDIT: For all you infra folks out there, imagine not being able to dynamically provision resources, no packet loss and worse, the packets can basically do what they want. That includes things like crashing into each other and stopping any other packets from making it through.

re spinning up a new lane: the A38-M Aston Expressway into Birmingham, UK dynamically multiplexes its lanes together into variable width channels depending on the required upstream/downstream bandwidth :) - http://i3.birminghammail.co.uk/incoming/article1294911.ece/A...

There are highways in the US that do this with dividers. A specialized vehicle drives over them and shifts them one lane to the left/right twice each day.

A good example would be the Golden Gate Bridge (at least they used to do this, I haven't been to the west coast in several years).

Still does, but it's only over the bridge. The approaches have fixed bandwidth and can back up.

I was watching Motorway Cops - or whatever it's called on BBC - just last night. There was a major crash, with a Mini and a Jaguar.

I didn't know that the direction of the Expressway changes, it's pretty cool, but has kind of put me off driving down it.

aww hell no. I'd be terrified of merging into that lane with no highway divider between me and oncoming traffic. Small two lane highways are bad enough! :P

There are less terrifying, less dynamic alternatives:


Nice! yeah, this would be preferable. If there's a catastrophe in the opposite lane, I'd really love it if it could stay over there while I'm going 70 miles an hour in the opposite direction :P

Most of the time, there's an empty lane as a separator. When it is used, they don't allow motorbikes in it just in case, but it's really no worse than a two-lane undivided road.

It's also not necessary for you to use that lane if you don't want to -- it doesn't go anywhere special :).

As a correction, as this road appears on the BBC's Motorway Cops regularly, and there was an incident featured in this weeks' episode (1): There is always an empty lane, which shifts depending on the time of day. Accidents are apparently very rare, and the speed limit is 50 mph, compared to 70 on a regular motorway and 60 on a 'normal' road.

[1]: http://www.bbc.co.uk/iplayer/episode/b03c7h9y

That is sort of terrifying but also quite cool

So the older/well circulated article that this author complains about would be describing Van Jacobson's algorithm?

Am I missing something about what the author is saying about the empirical data?

The stark uniformity of the empirical data up to the congestion threshold compared to the wild variance in heavier traffic is exactly the point of the older article this author is critiquing. Of course congestion happens when you approach capacity. This article is rejecting the original's explanation of that variance, without offering any new explanation.

This doesn't sound like the same study, but it's similar: a TED talk about how adding a fee for bridges in Stockholm reduced the number of cars by 20% and effectively eliminated congestion: http://www.ted.com/talks/jonas_eliasson_how_to_solve_traffic...

"Yeah, so, it turns out that charging for something prevents its overuse."


I think the point is the nonlinearity of speed as a function of the number of cars, i.e.,

"Charging for something reduced use by 20%, and overuse-related problems by 95%"

"WOW! Why don't more people do that?"

Did you read the whole article? That's explicitly discussed, even with graphs!

> At low occupancy (cars per mile), drivers can go as fast as they’d like. As occupancy increases, so does flow rate, even though speed decreases somewhat due to everyone trying to maintain following distance. At a certain point, when occupancy becomes high enough, speed dips low enough to where drivers are unable to maintain their minimum following distance, and — catastrophe! — the flow rate decreases dramatically.

The catastrophe theory model seems compatible with the anti-traffic solution to traffic 'standing waves' - by slowing down a little when approaching stationary traffic, allowing that stationary blockage time to 'evaporate', you avoid slowing down below the catastrophic threshold that puts your rear bumper more than 2 seconds behind your front bumper, keeping the road in the higher capacity regime. Same reasoning is why UK motorways signpost 'advisory speed limits' - and some even have variable enforced speed limits - to manage road capacity.

OP here, excellent point. Since so much is determined by occupancy (cars per km), I think the only way to switch domains is to lower the road's occupancy, which is exactly what you're doing when you leave extra space in front of you. However, I tried to point out that this probably comes at the expense of increased occupancy somewhere else, so you're just moving the jam around.

One thing your analysis leaves out is the human/psychological element. I've never bought into the idea of "erasing" the waves causing an overall throughput increase directly, but rather that it adds buffers so the traffic is a lot more predictable. The more people need to slam on their brakes, the more they deviate from the optimal-flow following distance, effectively wasting throughput. I guarantee you that when humans are involved, you'll get a lot better throughput from a line of nicely synched cars humming along at 25mph with a 2 second gap than the same cars constantly speeding up and slowing down, but maintaining an average speed of 25mph and attempting to maintain a 2 second gap despite wildly varying traffic conditions.

But if you can spread the jam over a longer area with higher speeds then you should be able to prevent any part of the jam hitting that catastrophic threshold.

...so you just moving the jam around.

If there is a reliable way to move it behind me, I'd like to know!

For a blog post that that claims to explain "How traffic actually works" I find it disappointing that the author doesn't even briefly touch on the dynamic properties of this problem.

No, this is not how traffic works, the model is much more complex and you're missing very large dynamic effects.

This is such interesting stuff. If you go just one step further and plot the carrying capacity of a road against the speed of traffic on it you get a hump which tends towards zero as speed reaches infinity.

The significance of that hump is that it represents a maximum carrying capacity for a road - an optimum speed. If you do the numbers I believe it's about 50mph. What I think is really cool is that when the road is completely at-capacity, traffic is stable above 50mph (as cars can just slow down and capacity increases) but collapses to a jam below below 50mph (because as the traffic slows down to absorb new cars, capacity falls still further).

This is way why, when the M4 motorway funnels down to only two lanes as you approach London, there is a 50mph speed limit - it literally allows the road to carry more vehicles per hour. I also suspect that it's another benefit (other than safety) of setting 50mph speed limits in contra-flows and during roadworks - it allows the road to carry more vehicles despite having fewer lanes.

I'll have to try and find the article, this was years ago, but there was a study about people just tapping their brakes lightly that would cause the person behind to do the same, a little longer, etc. etc. that would cause a big traffic jam over time. The study calculated some figure that showed if people just kept their distance up to the person in front, and instead of lightly tapping the brake as a reaction to a merge, and just let the car idle a bit to slowly decelerate, you could reduce traffic slows / jams by a lot. This seems more reasonable than the solutions presented by this article, don't let people merge.. really?

I believe the article you seek is the one being critiqued by this author.

Currently I drive for about 15 miles on a major highway (during rush hour at least once) on most days of the week. I rarely need to use my brakes.

The article claims that the maximum theoretical throughput for traffic per lane is 2,400 vehicles per hour.

What if every vehicle is separated by 2 seconds of stopping time, but all vehicles are moving at the exact same speed of 200 Miles per hour?

It would take one vehicle 0.005 hours to cover a mile (5,280 feet).

Assuming each car had a length of 10 feet, it would take 0.005/528 = 0.0000095 hours for a vehicle to cover it's own length at 200 mph.

2 seconds is the equivalent of 0.0006 hours. So each vehicle would take approximately 0.0006095 hours to cross any given point.

Providing a number of 1/0.0006095 = 1,640 cars per lane per hour according to my calculations.

I expected the number of cars per lane per hour to increase with speed, which is why I did the calculations, but apparently it does not!

If you could actually drive 200 mph on your way to or from work however, you would get there sooner than you typically do in traffic. It's an odd problem to think about.

Hmm? This shouldn't be surprising at all given the rather large assumption that one car passes a fixed point each two seconds. Go as fast as you like... an observer on the roadside must still wait two seconds for each car to rocket (or crawl) by. Imagine having an ethernet card that could only send one frame a second to a party on the other side of the globe: would you care if the additional network latency due to switches and the speed of light were 1msec or 200msec?

Where we may see great improvements in road throughput is in driverless cars that greatly reduce that gap.

It'll be really interesting to see how driverless cars negotiate things like natural merges (a freeway losing a lane), and how that affects traffic flow. Ostensibly they'll be as efficient as possible, but if two lanes become one, there's going to be some slowing regardless. Furthermore, in a vehicle that also ostensibly wants to get you to your point of interest in as timely a manner as possible, how does it decide whether to subject itself to staying in a right lane where it might be subject to moreane merges, or does it ever decide to go around a particularly bad or congested spot for the sake of getting you to the destination?

Yes, that seems to be the only possible metric we can reduce to increase the flow of traffic.

Cut that number in half from 2 to 1 seconds, and you would approximately double your throughput.

I love it! It's a counterintuitive result, but if you think about it, it makes sense.

I'm having a very hard time accepting the claims of original article, "The Physics Behind Traffic Jams."

Recently, I've tried to perform a few basic hypermiling techniques, especially "timing a light": if I'm the first car to encounter a red light, I leave extra space and can sometimes be moving through the light at 20mph or more when it turns green instead of stopped. It seems analogous to leaving space in front of you during a traffic jam, so that you can exit the jam at a faster speed.

The problem with this is that other drivers will often take advantage of your behavior. They think the most appropriate thing to do is race up to the red light. So instead of profiting from the faster speeds out of the light, they instead cut you off and start the light cycle at a predictable 0mph. You haven't done anything except swap places with these jerks.

...which also handsomely shows the real complexity of modeling systems where the human mind is involved, and the folly of over-simplified "1 car per 2 seconds" systems that look good in a computer model but fail to capture the real experience or decisions of piloting a car.

Flying through a light that just turned green for you is a good way to hit a pedestrian or someone coming in from the left or right trying to catch the tail-end of their yellow.

Please don't do this.

Surely most (all) lights leave a significant gap between one set of people going red and the next going green. Of course, people noticing this gap may think "well it only went red two seconds ago, so I have time" - which would mean not shooting off at the green makes you safer - but if people get used to everyone else waiting 2 seconds after it turns green, then suddenly they have even more time after their turn goes red to make it through...

Vehicle sensing systems mean the basic assumption of light timing no longer holds: when you get to the light can affect how long you have to wait.

I understand, but I'm not making the assumption. I'm talking about roads I drive everyday that are clearly on timers and I've even got good estimates on how long the cycles last.

This article is so horribly incorrect and fails right at the very beginning with the wrong assumptions and a idea that isn't thought through. What the author forgot is to take by far the most important variable into account: SPEED

So let's quote the author:

"The important fact: there is a limit to the number of cars that can pass by a given point on the highway in a given amount of time, and that limit is one car every 2 seconds, per lane. So imagine you are in heavy traffic during rush hour. There are a certain number of cars in line in front of you. Let’s pick a point on the road to call the front of the line — say, the point at which you plan to exit the highway. The line gets shorter by one car every 2 seconds. If there are 1,000 cars in front of you, it’s going to take a minimum of 2,000 seconds for you to get to the front of the line. It doesn’t matter whether people are kind and let cars merge in front of them, zipper-style. It doesn’t matter how much stop-and-go there is. The simple fact is that it takes 2 seconds per car for you to get to the front of the line, and there are some cars in front of you that have to get there before you do."

Ok this is correct. BUT: I don't care if it takes me 2,000s if my exit is 44 miles away (80mph) but I do if my exit is 10miles away.

So you drive along at 80mph and there is cars around you everywhere. Do you care? No, you're covering distance. It's the same as if you were all by yourself on the highway. The distance between cars is absolutely irrelevant and so is the time between cars if you still go 80mph. Nobody would call it a traffic jam.

Next quote:

"Leaving space in front of your car for people who are trying to merge won’t solve anything."

Yes it absolutely does. And it again solves one very very important piece of a traffic jam: Average speed. Leaving space is crucial. It allows people to do two things:

- Accelerate

- Not having to brake

These two are crucial to get speed up and thus resolve traffic jams. Incidentally he linked an article which visualizes exactly this:


(Section: Merging-lane Traffic Jams, A Simple Cure)

You can see on the right side how his "cars per second" is _exactly the same_ than on the left side. Yet, everybody will be much happier on the right side. Why? Because the speed's been increased and people in the cars cover ground.

Next quote:

"Suppose you’re on a 2-lane (each way) highway and one lane is closed up ahead due to construction. Now the flow rate of your lane is cut in half (or there are twice as many cars in line in front of you, depending on how you want to look at it). Road signs commonly ask you to use both lanes up to the point of the bottleneck. That’s reasonable advice, but it’s not going to get anyone home faster"

Yes it absolutely does. Why? We are all programmers here so let's talk in our lingo: This is our API. Our interface that we agree on. If people follow the interface (rule), then we have people agree on it and we are predictable and prevent braking. This is exactly what we want. Increase speed (with gaps), prevent breaking, be predicable. No surprising cars merging means no breaking and thus increase in average speed which in turn means no more traffic jam.

Please, disregard this article. The conclusions are wrong, the assumptions are wrong and the deductions and the terrible advice he gives at the end is wrong.

Having smoother traffic is good on many grounds: happier drivers, better fuel efficiency, and what not.

But it does not make people go home faster. Let's put a simple deductive thinking practice:

If you are the first car in the jam, you are stopped and slowed by traffic lights, police, accidents or whatever, but the traffic jam does not affect you, you are going as fast as you can.

For the second car, you will not get to your destination faster than the car you're following. So however you drive, you will not be faster than following with minimal distance.

The same applies to the third car, and on ward.

Basically, the fastest way to get through a traffic jam is to follow closely, and that means stop and go.

Thank you for that comment. You're 100% correct. However, it again shows that it's important to distinguish between theory and practice:

IF your model is correct, the traffic jam will clear since everybody will at some point reach their desired speed of approx ~80mph again. However, the reality is differnt. The model doesn't work since people break. That's the thing that makes your "deductive thinking practice" break down: People break. To prevent this....

Wait, let me just link this:


I think you're not considering the full problem.

You are describing a particular greedy algorithm wherein each individual car gets home as fast as they can. 1) Car #1 gets home as quick as they can by following closely. 2) Car #2 gets home as quick as they can by following closely, given #1. 3) Car #3 gets home as quick as they can by following closely, given #1+#2. ... Yes, given 1..N, the course of action which minimizes your own time is to follow closely. I agree. And people will probably act that way.

But that doesn't make it an optimal solution. Consider that car #1 introduces several seconds of delay due to braking and acceleration. Then car #2 does. Then car #3. By the end, you've got hours of delay built-in.

Optimal is obviously hard to define. But most people would agree that it's not every driver maximizing their personal interests to a globally shitty end. Everyone following closely is the prisoner's dilemma where both betray each other.

Consider the following alternate thought experiment. Flow of traffic is very much non-linear. That's what stop and go does -- it introduces cascading delay where there was none. Suppose roads are just 10-15% over capacity, and it results in a global average commute time of 1h instead of 45 minutes. This is not unrealistic. What if every day, 20% of drivers leave 1h later, thus preventing stop and go and keeping commutes to 45 minutes. Even if you counted their extra hour as commute time, the average is still less 1h.

I'm not saying this is a good idea (though strategies like this have been used in some cities). I'm just saying that are ways to get everyone home faster than everyone maximizing self-interest [co-operative self-driving cars please!].

Yes it does get people home faster if only because there is no latency on speeding up after breaking.

This. Smooth driving will make for a higher average / continuous speed, not to mention it'll cause less wear (brakes, clutch) and lower fuel consumption (accellerating costs fuel, braking is a waste of motion energy)

If you keep the same distance between you and the car in front of you regardless of your speed - which you seem to say -, I hope you are never behind me.

You're right. But in case of a traffic jam we operate at speeds that allow us to do this. Look, the Article's author actually mentions it too:


"The difference between these is negligible at high speeds, but at a low enough speed, it becomes difficult to maintain a 2 second following distance from the front bumper of the car in front of you without impinging on the rear bumper of the car in front of you, especially if said car is more than 0 feet long. So under these circumstances the flow rate of the highway decreases below 1 car every 2 seconds — maybe to 1 car every 5 seconds. So now you have to wait 5 seconds for every car in front of you in line."

There you have it. We have plenty of space/time to play with in case we're in a traffic jam since we approach very very slow speeds.

Obviously I'm not saying you should stay 10ft behind somebody when going 80mph.

Leaving space does not increase the average speed, because the average speed is dictated by the MAXIMUM speed in addition to the number of cars on the road.


First we must assume we live in the 3rd dimension of a physical realm, wherein objects like cars cannot pass through other objects. Cars can not drive through other cars, thus, if one car stops or slows down, all cars behind it stop or slow down.

Second, 5000 cars cannot travel 65mph over one mile of road, because it is physically impossible to fit 5000 cars inside a one lane 1-mile stretch of road. This is the bottleneck that causes the initial traffic jam.

In order to get all 5000 cars to travel the distance of a mile in the same TIME as 1 car, your only option is to increase their speed. But our speed is limited (by the speed limit of the road), so the only thing we can do is squish them together and hope for as fast a speed as possible given the dimensions.

If all the cars were linked and communicated in unison, they could react immediately and maintain a high degree of coordination in moving around each other and traveling as fast as the number of cars and speed limit will allow over a given area. Unfortunately, humans suck ass at driving, and their communication and latency make the movements of cars much much slower than is theoretically possible.


So what does "leaving space" actually do within our constraints of number of cars and max mph? It smooths out the errors humans make in driving. Specifically, it adds a time buffer for communication and decision-making; more time is given to driving decisions in order to prevent unnecessary stops, which would slow or stop the cars behind it.

What "leaving space" does NOT do is change the laws of our physical world. The given number of cars still have a maximum possible speed in a given distance, due to the traffic laws and physics.

Is it possible that the buffering from leaving space in the road will cause traffic to actually move faster than the starting-and-stopping of human error? Yes.

But it also uses up additional space, which is equivalent to more cars on the road, which as we know from above means a lower speed to travel the same distance.


You can not get away from it - until you reduce the number of cars or increase the max speed, your traffic jams will remain.

As for my personal opinion, I believe the one thing that could reduce traffic slowness is to speed up merges. If you want to help a traffic jam, merge at a speed faster than the cars around you and accelerate away from the merge area. It won't permit more cars on the road, but if a car is going 45mph and you move in front of it going 35mph, physics only allows for two results: the car behind you will hit you, or the car and all the cars behind it will have to slow down to 35mph.

The other thing that could be done is sensors in the road could detect the level of traffic (number of cars) and provide on a digital display the maximum speed humans could manage for that many cars on the road. If all cars then traveled at that speed, we would slowly but at a continuous pace travel down the highway without any stops. Until the first asshole that merges slower than that speed, which would make everyone slow down to his speed, defeating the whole thing....

Agreed 100%

I'm sorry if my post was confusing. I didn't really mean to go to infinity with my theory. I really only wanted to point out the flaws of the article. I didn't mean to say that traffic jams can be avoided. I simply disagree heavily with the author and agree with what's been out there in traffic theory for decades. Part of what you described well in your post!

Again, I'm not suggesting anything radical or want to emerge a new theory. The existing ones are good. I'm just trying to point out flaws from the OP.

There is a huge benefit related to your point about smoothing out human error:

People who pick the wrong lanes can change back immediately and travel the right way rather than either (a) slowing to a crawl with their indicator on hoping to be allowed into the correct lane by some kind soul (who also has to stop/slow down) or (b) looping back round on themselves and trying again.

Obviously, both of the latter solutions add to congestion, and the most common solution I see is an awkward combination of both (a) and (b).

EDIT: In addition, smoother-flowing traffic is desirable in-and-of itself, as fuel economy is greatly improved.

> You can see on the right side how his "cars per second" is _exactly the same_ than on the left side.

Is it my browser, or is this completely wrong? For me, the right side moves about double the amount of cars into the upper edge of the image. And I believe that this is the point of the section — it wouldn’t make any sense otherwise.

You are correct. And it's not just the images either, it's to illustrate the point that difficulty merging lowers the average flow-rate below that of "maximum lane occupancy".

I have to correct myself: Yes you're correct. I thought that the top left blinking arrow indicated that a car passed but that's very wrong. It's just displaying the road construction arrow and the rate or cars is indeed about twice as high.

There are considerably more cars on the Left GIF though. I guess if the number of cars in both lanes and in both GIFs was equal, the average speed of cars in the right GIF would be lower such that flow (throughput) would be the same.

Cars per second are the same. Open the two gifs next to each other. Each has 8 frames and the top right lamp is on for 4 and off for 4.

The lamp is irrelevant. Look at the cars.

> You can see on the right side how his "cars per second" is _exactly the same_ than on the left side.

Yes, but there are way more cars on the left side. That is the difference. Occupancy (density) determines speed when flow rate is fixed, and that graphic is actually a perfect illustration of this. You can't make the cars on the left travel at 80mph because it would not allow for a safe following distance.

One thing that is the same on both sides is that if you're the nth car back from the merge, it's going to take 2n seconds for you to get to the merge. Yes, the important thing is how many miles you can cover in that time, but this way of thinking places an upper limit on how many miles that can be, based on the occupancy of the road.

Also, pretty sure the advice at the end was tongue-in-cheek ;)

I'm sorry for the harsh critique. I think new radical ideas are great and people should question everything that's out there. But this way was not thought out well enough IMO.

I hope you keep it up and maybe have a follow up post where you can fix some of the flaws.

Some of your ideas are correct and noble. But always remember to take into account the real world driver and his/her behavioral flaws.

Few thoughts: Interesting would be a stochastic behavior model of drivers + driving strategies and in connection with red lights, accidents, rush hour to see what happens. Thought, that model would be quite some work in python.


Really appreciate the feedback, comes with the territory :)

I added an update to the article looking at what happens when you vary the average car length (corresponding to the % of trucks on the road), fwiw.

> I added an update to the article looking at what happens when you vary the average car length (corresponding to the % of trucks on the road), fwiw.

That's silly, at most reasonable traffic flow-rates the difference in a couple of feet between a short car and a long car (esp with how uncommon the outliers are) is insignificant compared to the constant overhead of the per-car padding. And below reasonable traffic flow you're already at a high enough density to cause catastrophic traffic speeds.

I just read some of your older blog posts. They're close to my research and well written and very informative. Keep up the good work!


Thanks! What's your research on?

Please allow me to unify the two theories, because both seem correct.

The free flow merge theory applies only in open areas with maybe temporary blocks that flow is unrestricted in any other way.

The no-merge theory applies on situations with traffic lights (and/or other special flow control such as a policeman or toll booths etc) where you suppose there are cut-off points that allow a specific number of vehicles per second.

I'm sorry. I didn't mean to lay out any "theory". I really just meant to criticize. I'd suggest to just stick with the theories that researchers have come up with and improve upon them.

Don’t let people merge in front of you, ever.

When you merge, you're merging in front of somebody, so a rule like this, universally followed, is effectively a ban on merging. How's that going to work, exactly?

Travel time isn't the only consideration. Things like safety, fuel consumption, and stress levels are important too.

I thought the whole point of these "fixed traffic jam" articles was to make your undoubtedly terrible commute go faster by giving you something to do, or at the very least providing a false sense of control over a situation (like a superstition). The science behind traffic is completely arbitrary and nonsensical no matter how many graphs or numbers you put behind it. There are simply too many different drivers out there to make any empirical evidence useful.

I think the idea is to give it a treatment analgous to how statistical mechanics works for properties of gas, for example. True, we can't easily say what individual atoms/molecules might do, but we can say something about the gas as an ensemble.

It's true that free will may make drivers tougher to predict than individual molecules. But molecules obey quantum mechanics and one could conceivibly construct probability distributions for how that particle might behave. In principle, probability distributions could also be constructed for the behavior of individual cars/drivers, so a statistical approximation for the behavior of the ensemble of drivers in a system is probably a reasonable aim.

I get that we want it to be consistent but having driven the same roads for years, in the same times of day, in varying but similar weather patterns the driving is completely sporadic so no amount of normalization well give you anything more than a crapshoot on any given day other than "it will probably take me anywhere from 20 to 50 minutes." Statistical analysis is great for many things but understanding the "now" of a traffic pattern is not one of them.

I think it depends on what one wants to model. For a general understanding of how traffic flows, a statistical model is okay. On the other hand, if you want to know why a specific traffic jam is happening and how it's affecting the flow, a statistical model won't be sufficient.

I wish more drivers would follow the driving etiquette that advises one to "stay right except to pass."

On every highway trip I see countless cars camped out in the left lane but moving at the same rate as the rest of traffic.

One should only use the left lane for passing. Otherwise, stay right.

Traffic flows smoothly when people follow this system.

I blame The Eagles. By popularizing the meme of "fast lane" in preference to that of "passing lane", they've badly damaged the laminar flow of highways, and likely contributed significantly to road rage.

It is the fast lane in California. According to the California Driver Handbook, the left lane is known as the "fast" lane (p32). They suggest if there are three lanes to use the left lane to drive faster, pass, or turn left. They also say "Do not drive slowly in the left (fast) lane." (p64)


The California fast-vs-passing lane pattern also emerges due to two things that California requires that other US states tend not to require: semi-trucks must go slower and keep right.

From the handbook cited above, or see also signage on many CA highways: The maximum speed limit on most California highways is 65 mph. You may drive 70 mph where posted. Unless otherwise posted, the maximum speed limit is 55 mph on two-lane undivided highways and for vehicles towing trailers.

... When you tow a vehicle or trailer, or drive a bus or three or more axle truck, you must drive in the right hand lane or in a lane specially marked for slower vehicles. [emphases added]

On divided highways of 2 or 3 lanes in each direction, such as I-5 in the Central Valley, these requirements strongly separate traffic into 2 sets - semi-trucks doing 55-ish in the right lane, and non-trucks doing 65-70-ish in the "fast" lane. (okay, yes, 85-ish, except in Kern County).

The results are that 1) over-the-road truckers dislike driving in CA, and 2) CA non-truck drivers quickly learn to avoid the right lane.

The other 49 US states tend to have more of a "keep right except to pass" pattern, depending of course on other local variations, especially population density, traffic volume, and similar.

Source: someone who drives (and tows) in 18+ states per year.

> I wish more drivers would follow the driving etiquette that advises one to "stay right except to pass."

In many places this is the law, rather than just ettequite. :)

Of course, that doesn't mean people actually stay right unless they're passing.

In practice this cannot work. "Except to pass" allows for any nonzero differential speed. Someone will be doing 55MPH in the rightmost lane and someone will pass them at 55.1MPH, and someone will pass them in the left lane at 55.2MPH.

This is fine as long as each driver migrates back to the right as soon as feasible.

It's evolved sufficiently here in California that I believe Californian drivers now prefer to pass on the right rather than the left, given a choice.

That was the clearest explanation of traffic I've come across. I'm inclined to buy into it (often it seems no one knows what they're talking about when it comes to traffic).

It also runs counter to what is commonly said about traffic jams. Most people focus on restoring smooth flow by doing things such as driving really slow to create space in front of you. This guy makes the case that "flow" is not really the important factor here.

For the record, I am "that guy" who runs all the way down to the end of the closed lane before merging over. Except I don't consider it an asshole move because as OP points out, I'm making the best use of the road ahead (while allowing fewer people to merge ahead of me)

For the record, I am "that guy" who runs all the way down to the end of the closed lane before merging over. Except I don't consider it an asshole move because as OP points out, I'm making the best use of the road ahead (while allowing fewer people to merge ahead of me)

In interstate travel I have noticed the increasing frequency of the "two trucks" maneuver, in which a big rig driven by someone philosophically opposed to your habit will drive in the "closed ahead" lane directly alongside another big rig in the "open but crawling" lane. Once the two trucks near the point of obstruction, the "open" driver is happy to let the "closed" one merge in, and there is no merging at the point of obstruction, which is both a safety and traffic-flow win.

I contend that the two truckers have done more to speed actual traffic flow (i.e. from a point before the obstacle to a point after it) than any number of assholes have. While at the beginning of the maneuver there is a solid line of fuming assholes behind the "closed" truck, by the time the obstacle has been reached all fuming asshole traffic has completely merged behind the "open" truck. Since it is only merging at the point of obstruction that slows traffic more than the obstruction itself already has, this is a net win.

Of course this maneuver must be repeated by successive pairs of truckers, and occasionally an asshole will "win" by racing forward between pairs, but that's minor.

I wonder if good zipper merges would be more likely to happen if, instead of having lane 1 merge into lane 2, both lanes got merged into lane 1.5.

Yes, that probably would improve things. I could totally see that working if we had electronic (or other quickly reconfigurable) lane markings.

Part of the problem is a lack of established convention. When a theme park opens its gates in the morning, we expect a large crowd to jostle in through a constricted entrance like a funnel. However, when we queue up at the bank, if the line extends past the cordon, nobody runs up to that point and stands shoulder to shoulder with other patrons; we maintain a single file.

Which of these rules applies to cars merging? There isn't a set standard, so people invent their own and perceive their rule is universally understood.

There is a protocol. Even in heavy traffic, if you observe the following distance that the DMV tells everyone to observe when taking a driving test there is always room to merge.

...following distance that the DMV tells everyone to observe...

Who on earth leaves such a distance in stop-and-go traffic? I certainly haven't seen it. You might as well just assume that every driver will be safe, considerate, and wise.

If traffic is stop-and-go, a zipper merge is trivial.

Ummm, not in my experience.

Smoother traffic flow (as opposed to stop-and-go) may get you to your destination slightly slower due to added distance between cars, but that is not the only consideration. Gas mileage and car wear-and-tear are improved (and I suspect safety as well) when the driver is not alternately stomping on the accelerator and brakes.

This is the one thing that really pisses me off. It gets worse when the lane that you are merging into has a line of cars going back almost a kilometer.

Doing this is like saying "Hey you idiots, look how smart I am" to everyone waiting for their way in the right lane.

I don't know why people get so mad about this. You use the term "right lane", but what does that mean? It only means what you want it to mean in a given situation.

The lane is open until it is not, and as long as both lanes are open they are both "right" (in the non-directional sense, obviously). Why rely on clearly nebulous (sometimes variable, sometimes non-existent) social norms rather than concrete traffic norms? Use the open lane, or don't complain when others do. It's not asshole behavior, it's common sense driving.

There is no reason to take offense while driving so long as people 1) follow the rules of the road and 2) do not endanger others with reckless driving. Using open lanes does neither, and it only inconveniences the people that refuse to use the roads as they are intended to be used.

There are known choke points where traffic is forced to merge. There is the occasional tourist, but rush hour happens every weekday. People know about these merge points, see traffic backing up in all lanes, and will still jump into the lane about to end trying to rush around a few extra cars.

It's rude, and it demonstrably increases traffic.

Okay, when do you think people should start merging? Thirty feet before the "choke point"? One hundred feet? One mile? Three miles?

It's an arbitrary game, and everyone makes up their own "rules of decency" and gets mad when someone follows the rules of the road rather than the rules of their head. Yes, people "know" about choke points, but that doesn't mean it's wrong (or rude) to use open lanes. So long as those lanes are open, they are going to be used, so everyone may as well make use of the space.

Is it my fault that others choose not to use those lanes? No. Is it your fault that it upsets you? Yes. It's not rude. I'm not being impolite. I wish everyone would just follow the same set of rules so we could stop being mad at each other. Which set of rules should we follow? I vote against the ones that change in with the person you're asking.

Having said that, I don't really like when people change lanes just to jump ahead of one or two cars. It doesn't upset me and I think it's fair game so long as it's done safely, but I empathize with the distaste for that a little more.

During "normal" periods where there are no choke points because there is no traffic these "smart" people do not have a problem staying in the "right" lane. However, when the traffic starts growing up almost everybody, except them, go into what looks like a queue, one car at a time, waiting for their turn to move forward/leave the road they start passing everybody on the "wrong" lane and merging at the end.

This is not only rude, but also blocks the other(s) lane(s) bringing traffic to where it wasn't before.

Its almost like say it's ok to pass in front of other people in the theater line.

Using the available lane-age/real-estate to attempt negotiating a merge at speed and minimize the problems caused by inertia is a good thing. This works in at least two ways: 1) The backup is less likely to extend past an off-ramp as pointed out. 2) Human-nature. If everyone in the backed up lane were paying attention and trying to get off the road as soon as possible, this benefit wouldn't exist. However, some people use the opportunity afforded by the congestion to check their email or a map or adjust their hair/makeup or turn around to yell at the kids or just day-dream. When the car in front of them moves, it takes them several seconds to snap out of whatever is distracting them. In these cases, it's possible to slip in front of them without adding to the number of inert vehicles.

This is true of unstopped vehicles as well. Not everyone is comfortable driving at the maximum safe speed. In fact, where I live, fines are enforced on all state roads from 6 to 13 MPH below engineering guidelines. Even so, many people choose to travel even slower than allowed by graceless statute. While these patient souls approach the bottleneck, several of the swift may de-queue themselves to the benefit of everyone who would otherwise wait behind them.

Oftentimes, the road hasn't actually reached saturation--it's just a compound case of live-lock and if people use the full measure of available lanes (instead of stopping at the first sign that says "Road Work Five Miles Ahead"), they have a greater opportunity of resolving the contention without promoting a cascaded slow down.

Of course, all of this is an edge-case. As the OP alludes, when the road is actually saturated, the real solution involves either decreasing demand for or increasing the supply of throughput at the chokepoint. Mere bickering and maneuvering will not avail anyone.

What pisses me off is having to wait in a kilometer-long line of people merging when what I really want to do is turn right down a side street.

Please keep that line as short as possible by using all the lanes available.

I thought we were talking about freeway/tollway/highway traffic? If there are side streets, that's a different situation.

What pisses me off is having to wait in a kilometer-long line of people blocking all available lanes when what I really want to do is turn right down a side street.

Please leave exit lanes open by merging as soon as possible.

If HelloMcFly originates from the "open lane", I am indifferent to where he decides to merge into the "right lane", as long as he doesn't force anyone to unnecessary slam their brake.

What really piss me off are drivers who move from the "right lane" into the "open lane", and then merge back after cutting perhaps 5, 6 cars.

Except I don't consider it an asshole move because as OP points out, I'm making the best use of the road ahead (while allowing fewer people to merge ahead of me)

Which solves nothing but to piss off everyone else behind you who has waited because ultimately someone will let you merge in (which they flat out shouldn't) and said behavior will only encourage others to do so in the future which inevitably will piss off even more people.

Yep. It's just like saying "I just jump to the front of the queue because I'm minimizing time spent waiting". Yeah, for you, whereas everyone else has to wait N more seconds.

Nobody is forcing you to stay in the congested lane. It was your choice and you have no reason to think everyone else has to do the same. If it's a legal lane, then drive in it.

That's fine if you can safely merge into the exit lane without slowing down traffic in the non-exit lane.

Here's when it's not OK: cars in the exit lane are traveling at 2mph. You drive up next to the line, slow down from 55mph to 2mph and wait with your blinker on for someone to let you in. Now you've slowed a second 55mph lane down to 2mph, blocking thru traffic and creating a dangerous stoppage in the non-exit lane.

Seriously, don't do this.

I see this all the time in Massachusetts - Route 2 West to Route 95 North, about a mile from my home. At rush hour there are always enough jerks playing the "merge at the last minute" game that only the left lane is still moving significantly, and sometimes not even then. Every once in a while there's an accident because people who were unaware of this "congest all the lanes" behavior didn't see a blocked left lane soon enough.

You're right, it's a major dick move. It's reckless and endangering; police should treat it accordingly.

Perhaps a more advanced camera technology that detects asshole driving behavior instead of speeding? Rich dudes pay to get ahead. Increased revenue for the police department. Improved traffic. Win-win for everyone.

Have you blocked through traffic? Only if the number of cars per second drops. Also as there's going to be a queue of, say 1000 cars behind the merge point, it's better than the queue doesn't extend backwards into some other area of traffic, causing people who are turning off before the merge point to become involved in the slow traffic

Its a legal lane to no where. By using it you're then relying on the good graces of one of those chumps you just passed. They don't need to let you back in. They're in the legal lane to somewhere, you just got no where really quickly.

You don't have to let me in. I am sure someone in front of you will. Then you have the right to throw a little hissy-fit because of having to live with the repercussions of your poor life decisions.

My response isn't to throw a hissy-fit.

I catch public transport to work. Mostly because I'd rather have a nap on the way to and from work than deal with wankers like you.

> For the record, I am "that guy" who runs all the way down to the end of the closed lane before merging over. Except I don't consider it an asshole move because as OP points out, I'm making the best use of the road ahead (while allowing fewer people to merge ahead of me)

Asshole move? I don't understand. I don't live in the US, but here it is encouraged to act like this! It's called a zipper merge. I remember public messages on the TV encouraging this behaviour, and there's road signs near merging lanes.

What's so asshole about this?

I've tested this theory and others on my commute. Whether I speed up and never let anyone in (nearly impossible since someone won't care and will merge knowing you'll most likely not hit them) or keep a good 7-10 car distance, it doesn't matter. My commute on the freeway with either of these two ways or on side streets is always 45 minutes +/- 1 minute.

The difference is I'm using less gas and putting less wear on my clutch, brakes, etc. by leaving a big gap between myself and the cars in front of me.

Furthermore I find that my demeanor is much more calm and relaxed when I leave a big gap because I'm not tensing up to avoid rear ending someone every time I stop.

Great. I really hope that like 100% of jerk drivers do not find out about this. Never let people merge in front of you...geez man, really?

That was a joke ;) Onramp regulators are the real answer, letting only 1 car merge into traffic at a time at a rate well under 1 every 2 seconds.

> That was a joke ;)

I found your article interesting in light of the popular traffic piece you mentioned, but ending it with jokey suggestions after a fairly serious analysis doesn't really make it clear if you're joking or not.

But hey, maybe @themodelplumber and I are the only two who thought you were serious.

Yeah, I failed there. The whole article started out jokey but I ended up cramming a bunch more "serious" analysis in.

Doesn't this simply displace some of the wait to the onramp?

I used to be skeptical of ramp metering until the day I drove several miles below the speed limit to 'celebrate' the anniversary of a ticket I'd received. Instead of driving 14 MPH over the limit, I drove 14 MPH under the limit (legally that time) as a form of radical civil obedience* . As expected, this caused quite a bit of unnecessary congestion.

It actually turned out to be harder to sustain a backup* * once I entered the metered areas. The reason? Metering limits the number of idiots on the road at a given time. I.e., my attempt at being intentionally stupid was mitigated because the meters prevented the road from nearing saturation. The little bit of wiggle room on the road made it relatively straight-forward to just route around me.

So yes, in one sense meters merely move the wait from one point to another but they also remove some of the interactive effects of human stupidity.

* https://www.youtube.com/watch?v=OoETMCosULQ "I know no method to secure the repeal of bad or obnoxious laws so effectual as their strict construction." -- President Ulysses S. Grant's First Inaugural Address

* * No I did not intentionally prevent people from passing me. My beef is with the revenue centers that are the pre-emptive enforcement of low velocities--not with my fellow passer-motorists.

"At the risk of being helpful, here are some things YOU can do that are actually guaranteed to improve commute times for everyone: ... 2. Don’t let people merge in front of you, ever."

How will the commute times of those who sit there waiting to merge be improved? If traffic flow was heavy enough they would never merge.

If they didn't pack a sandwich they'd starve!

Smooth flow doesn't get you there any faster than highly variant turbulent flow. This is obvious. Now given that there will be little difference in overall throughput would you rather a) have a relaxing smooth drive or b) a stressful highly variant aggressive drive? The answer should be a, because the increased of risk of a fatal crash vs. the miniscule return of a few car lengths is not worth the effort you put in.

You can verify this yourself. Drive as smooth as possible. Smoothly accelerate and decelerate. Leave large gaps. Try to use both your accelerator and brakes as little as possible. Now observe aggressive drivers and memorize their car's make, model and license plate. Observe the amount of risk and effort they undertake to get ahead. Then observe them only getting about 5-10 car lengths for their trouble. Finally observe them stopped behind you two traffic lights down.

Laminar flow = turbulent flow, but laminar is so much better.

This is besides the fact that laminar flow reduces traffic waves, reducing the variance for all those behind you, leading to a reduced risk of death and injury for everyone in your lane. My driving style has probably saved a life, or two.

I honestly believe it's the quick decrease in speed by drivers that causes non-accident traffic jams during rush hour.

Don't hit the brakes before merging and when someone merges in front of you, you should have kept enough space so you do not need to hit the brakes. Really, I think traffic would move much smoother if drivers leave at minimum 2 car lengths of open space between the vehicles in front of them and only hit the brakes on the highway when it is absolutely necessary.

If all the cars are put to a halt because of quick and hard braking cars ahead of them, its going to cause a chain reaction that is way more time consuming to drivers behind you than the author's "2 seconds slower to destination" factor due to a merge in front of you.

The purpose of making traffic flow smoother is that it'll push out the point where flow breaks down and you can extend the zone of (relatively) uncongested operation with higher vehicle throughput. Smoother flows leads to fewer abrupt stops and rear-end accidents that make the wait even worse. (Cars in a 'beautiful line at 35mph' behave differently from stop-and-go and for instance a fleet of self-driving cars would make the highest throughout, and extreme example)

As for late merging, evidence shows that it's more efficient in congested areas and by reducing the line, you help prevent spillover to other areas and importantly accidents due to conflicts. I can see one super suboptimal situation where a car completely stops to change lanes (into a stopped/slow one) due to bad flow, resulting in more than just a 2 second impact on one lane. MN signs say "use both lanes / take turns" and dynamically adjust based on traffic conditions. In LA, the "zipper merge" seems to be standard for most construction zones in traffic jams, with the only distinction being where you merge in the last 200 feet

http://www.dot.state.mn.us/trafficeng/workzone/doc/2004DLMS-... http://www.dot.state.mn.us/trafficeng/workzone/doc/When-late...

Thanks for explaining this for us. What is missing from this compared to what I see on the roads is trucks. The truckers always leave large gaps and it's a sure bet that jerks will cut in front of these rigs. I don't think the trucks are leaving that much space to "smooth out" the traffic but it's more a matter of safety and efficiency (saves fuel) when all the nimble passenger vehicles are stop and go and zipping all over the place. Truck drivers are so patient, they have to be, especially in rush hour I've seen cars jump in front constantly and the trucks keep making more space for them. The other half of people seem to just follow at the same pace instead of cutting in front, and every time some one does cut in front it slows them down another 2 seconds. So my theory is aggressive drivers aren't much affected by trucks but easy drivers are slowed down because they won't get in front... But if nobody gets in front of the trucks then everyone makes it at the same time. Basically the trucks are blockers for the aggressive drivers, so if you use them they'll get you home ahead of others. I don't imagine any truckers are on HN?

If ever I hit a jam on the motorway, I always pull to the left lane (UK), join the trucks, and imitate their driving style. Almost always, over the length of the jam, I beat my neighbours in the middle lane without doing any lane hopping.

Hmm. Interesting. I have always contemplated this problem. First, I want to simplify the problem a bit and hope it'll apply to traffic jams. I think the problem is human reaction time. I must be missing something but hardly people talk about this. (although my solution might not be feasible). Let me explain.

So, lets say you are stopped behind 30 cars and the first car is waiting for the red to turn green. Now, as soon as it turns green, it takes a second (or two) for the first car to start moving. After that second, second car adds another second before starting to move. Based on this, you won't move your car until 30 seconds have passed. Isn't this one of the big issue? If all of 50 cars instantly start driving once it hits green, wouldn't it reduce traffic jam greatly. Now, I do realize this is not feasible but if computers were controlling the car, 50th car can move instantly after green.

On a side note, I realize the bottleneck situation can't be solved easily BUT again, I think human reaction (response) time is a huge part of it in most of these scenarios. If computers were driving, I wonder if we can have 10 times the car we have today without traffic jams.

I'm wondering what if I'm missing something here:

"If anyone tries to tell you that if only drivers left space in front of them and took turns merging, traffic would flow smoothly, and it’s only because of jerks that there are any traffic jams at all, just ask them what’s going to happen at the next merge. Where is that extra space going to come from? You cannot keep 2 seconds back from the car that has just merged in front of you without, um, slowing down. If the car in front of you is also slowing down for the same reason, you have to slow down even more. This is basically the definition of a traffic jam."

However, if you have two lanes, with one lane blocked, then wouldn't not allowing for another car to safely merge into your lane prevent them from merging? Then you have one lane moving quite fast, but the other lane would be going more slowly...

Like I say, there may be something I'm missing here, I'm happy to be enlightened!

Given that ants don't have this problem, won't driverless cars be able to adopt ant-like traffic protocols to eliminate congestion?

cf. http://www.phschool.com/science/science_news/articles/ant_tr...

Ants have no momentum, to a first approximation. It's not at all clear their algorithm would scale.

> Ants have no momentum

And collisions between ants are not catastrophic.

Isn't the real problem with traffic more perception than reality? I'd be much more interested in how much additional stress/unpleasantness is caused by stop-and-go traffic that would be alleviated with the concepts mentioned in the "wrong" article.

As a matter of general policy, if you're going to declaim another article as wrong, you really should bring clearly stronger evidence—generally in the form of citations—to the table. Otherwise it's just "he said she said" as in this instance.

Has anyone tried to use genetic algorithms with traffic simulations to try to determine if there's a better way to manage traffic, especially in regards to the design of merge lanes, on/off ramps, etc?

I've thought for a long time that it would be fascinating to build a complex traffic simulator that could be used by people designing freeways, neighbors, shopping malls, etc to plan for traffic. Right now those designs are just based on best guesses (or terrible guesses) by architects and not simulations.

"... there’s not much benefit to trying to 'cancel out' a traffic wave by leaving a ton of space in front of you."

What do you mean by a ton of space?. What do you mean by 'not much'?"

The real question we should try to answer is "What is the optimum distance all cars should try to keep between them? And that can't be answered qualitatively, only quantitatively.

This article accounts for some of what causes traffic jams, but have a look at this video. It shows a study done where people tried to drive on a circular track at a constant speed.

Things start out okay, but it quickly devolves into the back propagating wave.


I don't understand the claim that front bumper to front bumper is a bigger problem the faster you go. When you're going fast, the 4 meters of a car are much shorter (in terms of time), because 4 meters at 4 m/s is a much smaller amount of time than 4 meters at 0.2 m/s.

It says it is a bigger problem at low speeds.

...oh. Wow, major reading comprehension fail there. Thanks.

I suspect that a critical mass of self-driving cars will improve traffic flow for everyone.

I don't have a clue what the percentage to achieve critical mass is, but my wild-ass guess would be 20-50%.

My two favorite parts of this article: the author lives in New York and "Don’t let people merge in front of you, ever." Now I understand New York drivers!

Shockwave effect is a big contributor to traffic. Even moreso than accidents/road construction.

How to avoid traffic: work from home.

"Don't let people merge in front of you, ever."

Californians have this one covered.

Everyone has their own pet theory of how people should behave in traffic. The problem isn't even that most of them are wrong -- most of them are "correct" in the sense that, if everyone behaved in this way, traffic jams would be less common and have shorter durations.

The problem is that you don't know which theories other drivers have, so you don't know how they will react to different situations. If you don't know how other drivers will react, you need to drive conservatively -- which usually means slowing down.

Take a construction merge down to a single lane. It might be the case that it's best to use both lanes up to the merge, so that the traffic jam takes up less space. It might be the case that it's best to merge as soon as you figure out what's going on, to provide more time for the merge negotiation and reduce the risk of accidents at the merge point.

In reality, it doesn't matter either way. If I try to use both lanes, I don't know that I'll be allowed to merge at the lane closure. If I merge early, I don't know if someone is going to use the now-unoccupied lane to zoom up and cut someone off, making both lanes slower and more accident-prone. I can't do either one and confidently say that I've made the correct choice, because I don't know how other drivers will react. What I do is just pick one, and slow down when something unexpected happens.

This article also generalizes some things incorrectly (it seems based mainly on German data, presumably because they have better infrastructure for measuring such). For example, I live in the northeastern U.S. We are notoriously bad drivers up here, even among other east coasters. Here, a two-second following distance at 65MPH is almost unheard of -- actually, a two-second front-to-rear distance probably qualifies as a "large space", as some people here seem to think such a space can fit multiple vehicles. That's about 100ft or 30m, so I suppose it technically can, but it is not safe.

I can react to this either by leaving a very large space (5-6 seconds seems sufficiently large that people get impatient and leave before it fills up) or by leaving a small space (.75-1.25 seconds is the largest I can leave and have a good chance of nobody trying to use it). The former is much safer, so that's what I do.

The key is that, over a period of about two minutes, I need to maintain this following distance relative to the same vehicle. Usually this is fine: at most three cars tend to use that space at a time, and if anything they're more likely to tailgate the vehicle I'm following than to stay back near me. Occasionally I need to slow down in a situation where many vehicles fill that space, or someone moves in and immediately slows down. This is the exception, rather than the rule.

Yes, I know this behavior can cause traffic jams under certain conditions and with certain other drivers -- but so could the alternative, under other conditions. Given the choice between two bad options, I optimize for my personal safety.

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