Hacker News new | past | comments | ask | show | jobs | submit login
Systemic traffic jams caused by small groups of drivers, study says (mercurynews.com)
148 points by freejoe76 on Jan 8, 2013 | hide | past | favorite | 121 comments



This is a very interesting report. It was based on anonymized tracking of cars with cell phone GPS signals to better understand the whole traffic network of the Bay Area. Identifying choke points for traffic led to a policy proposal.

"John Goodwin, a spokesman for the Metropolitan Transportation Commission, the region's transportation planning agency, says the best way to spread out traffic coming from these neighborhoods is to install metering lights at their freeway onramps, which spaces out the traffic to help both those drivers and everyone else get to their destinations quicker. Though many important Bay Area freeways already have metering lights, such as Interstate 280 in the South Bay and Interstate 680 in the East Bay, others don't."

The Twin Cities metropolitan area, where I live, was the first place in the United States to gain special federal permission to put on-ramp metered signal lights on federal Interstate Highways used heavily by commuters. They empirically help a lot in smoothing traffic. One reason we know that is that for a while a doofus state legislator shut down the freeway metered ramp program, until traffic here became so unbearable that the meters were put back in use. I have heard from friends who travel here from other parts of the country that the ramp meters (implemented as red-yellow-green traffic signals just before a car gets onto the freeway) are confusing to people who usually drive where on-ramps are just unimpeded paths onto the freeway. But they definitely speed up traffic.

The Twin Cities has one federal highway, Interstate 394, with a pair of reversible lanes, usually eastbound (into Minneapolis from the suburbs where I live) in the morning, and westbound (out of the city into the suburbs) in the evenings. That helps with rush hour commuter traffic, except that a lot of cars are eastbound for evening appointments even as commuters are leaving the city, so the reversal of lanes still leaves the regular, nonreversible lanes badly congested each evening. I wonder if a traffic study like the one reported in the interesting article submitted here could identify how to smooth out the traffic problems we still have here.


I'm currently making regular morning and evening roundtrips on I-395 near DC. My northbound segment in the morning is in the same direction as most traffic (into the city), and the entrance is metered. When I do that same segment in the evening, it's the less traveled direction, and there is no metering. This is interesting because the evening drive is often just as slow as the morning drive, even though there's less traffic, something I blame entirely on the lack of entrance metering.


For those of you who want to read a bit more about the twin cities traffic metering program there's an entertaining chapter in Numbers Rule Your World that goes through the history and (very lightly) the stats behind it. http://junkcharts.typepad.com/numbersruleyourworld/


We do have metering lights (mostly just red and green, one car per cycle) on most of the on ramps in the Bay Area, but a lot of them do not get turned on, even when the freeway is congested and those that do turn on tend to turn off at the end of the official high traffic time, instead of at the end of high traffic (although they usually are adaptive to unscheduled low traffic).

In the South Bay Area, there is also usually a major shortage of space to queue behind a traffic meter or to allow for acceleration after a meter. CalTrans knows how to properly design freeways, they do it in Southern California, but up here, they don't show it.


Those damn lights are terrible. We have them in Auckland where one authority controls motorway a (freeway) an another the local roads. The lights cause banking up at on ramps, blocking local roads. The motorway flow is improved, sure, but local single lane roads become impassable for hours. The lights just shift the problem here in Auckland.


That's an annoying aspect of trying to retrofit old roads with new tech. Roads designed for ramp metering (in Australia at least) have much longer ramps, both before and after the signal (enough space after for the car to accelerate to the speed of the motorway from being stopped, and before to stop the queue clogging surface streets).

But, it could also be a problem with the algorithm they use. The more simple algorithms generally don't take queue length into account, but ALINEA/HERO, which we use in Victoria has queue control.


The lights just shift the problem here in Auckland.

This is exactly their intention, and it's exactly what it does in the Minneapolis area -- it adds a natural saturation point to every onramp, eventually leading to lots of traffic simply using surface roads.


"It was based on anonymized tracking of cars"

If it was anonymized, they could in theory release the data, right? However the article states "all the data gathered in the study will be kept confidential". Don't these 2 statements contradict each other? Why would they need to keep anonymous data confidential?


The two statements are consistent. In order to do the analysis they need to keep track of the full paths that the individual drivers make. Given the paths, particularly over a period of time, it is easy to identify individuals, even if you dont have their names to begin with.


Maybe they can delete the first 100m of every trip, so it would be harder to poinpoint locations w/out losing that much


"On the Anonymity of Home/Work Location Pairs"

Abstract. Many applications benefit from user location data, but lo- cation data raises privacy concerns. Anonymization can protect privacy, but identities can sometimes be inferred from supposedly anonymous data. This paper studies a new attack on the anonymity of location data. We show that if the approximate locations of an individual’s home and workplace can both be deduced from a location trace, then the median size of the individual’s anonymity set in the U.S. working population is 1, 21 and 34,980, for locations known at the granularity of a census block, census track and county respectively. The location data of people who live and work in different regions can be re-identified even more easily. Our results show that the threat of re-identification for location data is much greater when the individual’s home and work locations can both be deduced from the data. To preserve anonymity, we offer guidance for obfuscating location traces before they are disclosed.

http://xenon.stanford.edu/~pgolle/papers/commute.pdf


100m radius of two end points on a simple trip would narrow it down a lot and would in many cases allow you to identify an individual.

If additional journeys are also linked to the same phone identifying individuals can get even easier.


Then it sounds like they have data tracking drivers even off the main roads: homes, driveways, small residential streets, parking lots, office buildings. IOW it sounds like they could have done a better job at anonymizing it by truncating driver paths that are off the main roads and highways to only keep data relevant to their study.


How do you know what data is relevant? Ey did the right thing by not sharing the data.


They study high traffic density, so by definition, irrelevant data is data where traffic is under a certain density, which would automatically exclude private areas (homes, driveways, etc).


Ah, unexplained downvotes on what I believe is a reasonable point I make...

At the very least, if data cannot be made anonymous, and can so easily be associated to persons, then this is an argument that they should have never collected it without my consent in the first place. This would be an invasion of my privacy.


There's a limit to how much such data can be truly anonymized, especially when it can be combined with other data sources.

Would you be ok with someone relating an "anonymized" database of fingerprints? Of sequenced human genomes? Of credit card transactions? Anonymization is tricky.


That's the problem AOL ran into when it released their anonymous web search data several years ago.

I think their hearts were in the right place, but it turned out, even though AOL had no interest in attaching real names, that there were enough artifacts in there that other people could. Especially if it spilled hot soup on AOL.


It is nontrivial to make data so anonymous NO ONE can deanonymize it. It's considerably easier to make data anonymous in a way that people acting in good faith to not try to deanonymize it won't realize who is who


It is nontrivial to make data so anonymous NO ONE can deanonymize it and still preserve the salient information.


It's surprisingly easy to de-anonymize an anonymized dataset. http://arstechnica.com/tech-policy/2009/09/your-secrets-live... Just because the researchers decided to keep their subject's privacy doesn't mean they should publish the data for any troll to play with.


Because other parties could work out a person's identity from their regular point of origin/journey route, and thereby invade their privacy.


I remember reading about a study where your commute was nearly good enough to completely out you, or something like that.


Well since the researchers had to identify the culprits that were slowing down others, it's not truly anonymous.

See other's comments about de-anon.

I wonder how many of these "anonymous" drivers knew that their data was being used this way and would consent to doing so if they knew it.

EDIT: Enh? Why the downvotes? Please explain if you are going to do so.


The research did not care about identifying the culprits more than by where their starts and destinations and at which times and speeds they travel. Knowing the identity is rather useless to the research.


We have those quite a lot in the UK (along various motorways), although they're typically only turned on when congestion starts to build up.


The lights in the Twin Cities are the same for the on ramps. Or were rather but same difference.


Reminds me of an interesting site[1] I came across a while ago, in which the guy argues very convincingly that traffic jams occur as waves, which are started when even a single car does something to cause the car behind them to slow down (avoiding something, stopping to merge, etc). If it's non-trivial, then that will begin a chain reaction - a wave propagating through a medium of cars. The severity of that effect is directly proportional to the density of the medium/traffic.

Given that perspective, metering lights to reduce density seem like exactly the right solution.

[1] http://trafficwaves.org


My father, in grad school, worked in a lab with a guy who pioneered a lot of this research. He showed the coil like action of these waves and how they created wrecks. Here are some other things I thought were cool:

He calculated the risk of changing lanes on a highway and estimated how much people valued their lives in dollars (usually fractions thereof) based on this one criteria.

He drove vehicles into lakes to study the escape from such a vehicle.

He tracked the eyes of truckers and found that their eyes would fall to the front of the hood and dart up a few times about 30-45s before they fell asleep and wrecked.

Another guy tried to make a talking highway by varying the distance of groves and marks on the road.

Just thought some might find this stuff interesting or amusing.


Honda made a road that plays part of the William Tell Overture[1]. It doesn't sound right, but the effect was still fun when I tried it.

[1] http://en.wikipedia.org/wiki/Musical_road#Civic_Musical_Road


I'd be curious if you have data on the risk of lane-changes. (Knowing how dangerous they are would probably cause me to do fewer of them.)


I would guess the risk is not all that significant in the scheme of things. Just another one of those, "Look, you say human lives are priceless, but here I will prove they do have a price (whatever that might be)" type deals.


The first time I heard about traffic waves and the use of active feedback to eliminate traffic jams was from [Prof. John Mahoney](http://www.asap.unimelb.edu.au/bsparcs/aasmemoirs/mahony.htm) in 1984, he gave two 45 minute lectures on the subject and later in the year posed two related unsolved problems in the area on our exams, we were undergraduates at the time but that was typical of him to make up exams at the last minute and pack them with current problems from either current literature or his in-tray.


Yuki Sugiyama and co did some lovely research on this phenomena and were able to reproduce it on a circular road. A rather splendid video of said experiment: http://www.youtube.com/watch?v=7wm-pZp_mi0


Awesome! I just saw their paper from the trafficwaves site a little bit ago and was curious what this would really look like. I like their point that a car suddenly slowing down is not really to blame for the jam, but rather an inevitable trigger that happens after a certain critical density is reached.


This is exactly what I was thinking of when I saw the article. I was impressed by how obvious it seemed after having read about it.


I'd like to see a website where I can enter my home address and work address and it will generate a chart of leave and arrival times for each day of the week based on prevailing and constantly measured and updated typical traffic patterns with accident-jams removed. In my experience traffic patterns, even without accidents, are highly non-linear and leaving 10 minutes early or later can drastically reduce your drive time. You could even have it rank different routes so that on any given day of the week you might take a different route and leave at a different time to minimize your drive time and increase the probability of arriving on time.

Metering lights have their use but are limited in how long they can hold up a car whereas if I am at home or work its no problem to adjust my schedule +/- 10 minutes or more to avoid a jam. As more people signed up for such a site it would also become more efficient since it could take into account where and when people actually want to go in its planning calcs.


I wanted an app to do this, and just let me know when it was going to take less than x minutes to get to work/home. The simplest workaround was to bookmark Google Maps directions, so a single click would tell me the expected time with traffic. I saved it as an Android bookmark as well, so I could single tap it from there, too.

This works ok if you're interested in waiting until after traffic, and not "beating" traffic. In my situation, under x minutes meant leaving at 5:30am or after 9;30am, and going home before 3:00pm or after 7:00pm, so that was fine. YMMV.

Google Now is somewhere close to this, but I've found it finnicky. Maybe because of my more flexible schedule was throwing off its ability to predict my commutes. I would have preferred something more explicit for work hours, than it guessing based on my GPS'd habits.


I was hoping for a solution that can solve (or at least minimize) the problem between 6-9am and 4-7pm. I think there are a lot of short-lived local effects that can be avoided. For instance near my office is a government agency that quits work right at 5pm on the nose. If I leave one minute before 5pm it is smooth sailing, one minute after 5pm and traffic is snarled until 5:30pm or so. Also, as far as general commuting, the school start times are one major effect since these are local and hard set arrival times for the parents. A smart database could calculate these effects and other cross-interactions that really impact jams so a small change with x% of drivers can prevent avoid the log-jams or backups that take time to dissipate once they form. Perhaps the data could even be used to get schools with interfering traffic patterns to stagger their start time 30 minutes which could have a huge impact.


Have you seen the Nokia Transit app? I don't use it, so I'm not sure. But from the last time I've played with a demo of it, it seems it might be helpful to you. I think it only runs on Windows Phone though (no website).


The problem with any such site is that if it became sufficiently popular it would result in feedback effects that might destroy its predictions.


Or it could do better -- dispense predictions/instructions so as to ensure an even distribution of traffic (e.g. by randomization).


An unethical freemium model for such a site: dispense instructions to ensure an even distribution of traffic to free users, but provide a purely optimal path for paying users, perhaps even making suggestions to free users to move them out of the way.


I wouldn't call that unethical - even free users would benefit from such a system, in the same way that non-users would also benefit, but paying users would benefit even more.

The system actually requires itself to be ethical. The ability to make things better for the paying users is dependant on the system having a lot of free users. Those free users have to be getting some utility out of the site, otherwise they won't use it. And without large numbers of free users, the paying services can't be offered!


True. But if it got that popular it could then be used to optimize the schedule road use of all the participants taking everyone's situation into account. It would probably be even MORE efficient at finding semi-optimal travel schedules for most users. It couldn't be a static database it would have to constantly revised as road usage changed even in response to its own calcs.


I've always felt that driving in traffic is a sort of prisoner's dilemma.

Consider a two-lane highway that splits into two directions, a very popular direction and an unpopular direction. Inevitably, we will see a group of people who will wait to the last second to cut into the popular lane. These late-comers cause two problems: 1) people to hit the brakes who get cut off in the popular lane, and 2) people to hit the brakes in the unpopular lane who are forced to slow down to allow the selfish drivers to merge.

When I take the generous approach and wait in line, I hate people who cut in. On the other hand, I find myself clever for cutting in at the last second and saving time when I take that approach. I'm still not sure which approach is 'right'. All I know is that the latter will increase overall delays while doing the former will have little to no impact on overall delays but a major impact on my own.

Traffic is a pretty interesting subject.


> " I'm still not sure which approach is 'right'"

If you have a circumstance where 2 lanes are merging into 1 lane, merge as late as possible. It sounds counterproductive, but it's actually best for traffic as a whole.

See http://trafficwaves.org/seatraf.html and the HN discussion at http://news.ycombinator.com/item?id=785259 . http://drmomentum.com/aces/archives/003800.html also has a number of nice links.


My interpretation is that gbadman did not mean the situation where two lanes merge into one, but rather where you have two lanes traveling in a single direction approaching an interchange where the lanes split such that the left lane goes one direction and the right lane goes another.

In gbadman's example one lane (let's say the right hand lane) goes to a road that does not back up, whereas the left hand lane goes to a road that will back up through the interchange. In this situation, you have a left lane congested with people waiting to go on the busy road, and a right lane that should be almost free flowing. By most understandings (including the execellent book, Traffic) people wanting to go to the busy road should wait in line in the left lane, allowing people going to the less busy road to pass freely. However, many drivers will stay in the right lane until the very last minute, then attempt to move into the left lane before the right lane splits off, backing up both lanes.


You are right in your interpretation of what I meant.

I think the lane split is the better example of the prisoner's dilemma. As others have mentioned that in the two-lane merge situation the global optimum is the selfish approach.


Yes. It's paradoxical because merging early seems like the polite thing to do -- but actually, if you do that, you're allowing people who were behind you to get in front of you, which makes the merge unfair.


>If you have a circumstance where 2 lanes are merging into 1 lane, merge as late as possible. It sounds counterproductive, but it's actually best for traffic as a whole.

I think it's important to stress the caveats of that recommendation.

It applies to heavy, congested (as opposed to flowing or merely partially congested) traffic with a merge that is orderly, zipper-like.

So, don't just merge as late as possible. Rather, when faced with heavy, congested traffic perform an orderly merge, as late as possible.


except when presented with a chance, someone would attempt to disrupt this order in order to get ahead themselves. The pure selfish (or greedy?) method doesn't lead to the optimal result.

As i mentioned in some other comment, self-driving cars taht you "hail" are the best compromise between public transport and private transport.


I've debated that sort of thing several times with my dad. He argues that if everyone waited until the last possible point to merge, things would be better, because there'd only be one point of congestion. I.e., nobody could "cut" the line because everyone would be merging at the end, instead of the people in the slow lane being slowed down not just by people merging in ahead of time but also by people merging at the last possible second.

I can sort of see his point, but the downside to me seems to be that if you wait until the very end, and don't get to merge in cleanly, you have to slam on the brakes, and nothing kills traffic like someone having to merge in from a complete stop. So my tentative position is that it would be best if everyone merged exactly as soon as they could do so without significantly slowing down (speeding up to merge seems like it would be a good thing, in comparison?). But good luck with that, heh.


According to the book traffic, your dad is right. But the effeciency only works if everybody waits.

After reading Traffic, I always merge at the last minute (and I have never had to "slam the brakes" from this behavior)


When you say major impact, do you mean something like 45 seconds, or something else?

I'm asking in all seriousness, I have no idea what sort of traffic you are driving in.


Not sure where gbadman is, but a similiar situation I often see at the circle interchange in Chicago waiting in line (as you should do in that case) can easily cost you several minutes.


I'm in Montreal. We have a serious traffic problem during morning and evening rush hours. In fact, TomTom has put our city at #4 in North America for worst traffic and it seems #1 for both morning and evening traffic peaks [1].

Two places that I can think of this being brutal are taking Decarie Blvd exit from Decarie North and taking the 20W exit off of the 13S. I'm sure that are some other pretty good examples out there.

1: http://www.tomtom.com/lib/doc/congestionindex/2012-1003-TomT...


Consider a two-lane highway that splits into two directions, a very popular direction and an unpopular direction.

Heh, years ago I wrote a comment here about a situation like that, but with three lanes (and one more for the exit), that I suffered.

Then, as if they were reading, in a few weeks the output was changed as I wanted: two lanes in each direction. The results were excellent, no more jams in that exit.


This research was published with supporting information in an open access journal (Scientific Reports [Part of Nature Publishing Group]). See the full text for more.

http://www.nature.com/srep/2012/121220/srep01001/full/srep01...


One of the things I really like about this study is that it captures not only road congestion but where the drivers enter into, and exit, the freeway system. And it reconfirmed for me yet again that starting/ending in Sunnyvale is reasonably optimized for working in the 'valley' part of Silicon Valley.


The problem I have is the study didn't fully factor in road layout/design. Highway 85 here in the Valley spans from the south Valley up to 101 in Mountain View. Within months of opening, it was a recurring traffic jam. Boom or bust in the valley, traffic let up a little in the bust, but the road is congested pretty much from 6:30am until 9:30/10am in the morning. Intersections of 85/280/101/237 at the north end impact traffic northbound in the morning. 85/280 impacts southbound in the evening.

Again - patterns persisted during both the last two booms and busts.

It is more than just a "small group of drivers".


Isn't this a strong argument for better buses?

I guess that's not really the thing if one works in the valley with a huge salary...

Surely the geek bus would be a cool place to be though?


the problem with public transport isn't that its uncool, but that you still ahve the "last mile" problem. Also, you have to plan your life around the transport, where as if you owned your own car, you dont need to do so.

THe best solution isn't better public transport, but self driving cars that you can get ondemand (but don't own).


THe best solution isn't better public transport

I disagree.

- If public transport arrives and departs often enough, you don't have to plan your life around the transport.

- If the public transport network is large enough, you can eliminate the "last mile" problem, either by making it short enough to walk quickly or by connecting backbones to local distribution. Classic example, train + bus.

Of course the Bay Area has nothing close to this kind of network, but you said "the best solution", and I'm pretty convinced an idealized public transport system is much better than driverless cars. Superior passenger density, superior fuel efficiency, superior materials usage, etc.


what you have described - big enough a network, running often enough, etc - are all the properties of the driverless car. In effect, the public transport system no longer consists of trains/buses, but a huge pool of driverless cars you hail.


The difference is the cost. Public transport has superior fuel, space, and materials usage.

Another difference is that public transport is a known working solution which has been implemented with great success in many cities since the 19th century. While driverless cars is something we just now is getting the technology to build so it is untested.


The difference between (driverless) cars and trams/buses is that the latter require a _lot_ less road space per passenger.

Take a look at the picture here[1].

[1] http://farm9.staticflickr.com/8456/7999510360_8e46299621_b.j...


Not necessarily. Cars still use a large amount of land area per person, even if they're self driving and don't need many car parks (I disagree with that last part though, considering commute patterns).

Take a look at http://carfree.com/ for a good overview of what could be done with public transport, as well as all of the unseen issues with car use.


>you still ahve the "last mile" problem

Here in europe we just walk the last mile ;) (although it's never as much as a mile.)

Like all these things it's about urban planning.


The last mile problem is mostly caused by poor urban planning, where the urban planners did not consider future building of public transport.

And in Sweden this poor urban planning was basically caused by people thinking public transport was uncool.


New York City does public transit remarkably. I've heard that London, England does as well. In both cases I would take public transit over a car even if the roadways had little congestion.


The top of Upper Manhattan to Brooklyn can be an hour and a half by public transit vs. 45 minutes by car. Queens to Brooklyn can be worse. New York City has transit that is good enough, meaning you don't have to have a car, but it's still atrocious. If you have to do a commute like that, even if you want to use public transit rather than a car, you have to be prepared to give up an extra hour and a half a day roundtrip.


The first two paragraphs of this article are crap, typical of the "journalism" of today.

  A groundbreaking study ... has pinpointed a small group of
  drivers making Bay Area freeways miserable for the rest of
  us...
You mean the other drivers, incremental vehicles coming from other sources, are simply not contributing to traffic misery? Who knew?! At least the solution is both obvious and simple: the "small group of drivers" should clearly be "gotten rid of" so the misery of the virtuous is removed. Line 'em up against the wall...

  ...they come from a few outlying neighborhoods and travel
  long distances together in the same direction like schools
  of fish -- clogging up not only the roads they drive on, 
  but also everyone else's.
You mean, they're "everyone else's" roads, and these cheeky interlopers are taking what rightfully belongs (only) to "everyone else"? And the other commuter vehicles are not traveling together "like schools of fish"?

Why do the (government) planning departments bear NO responsibility for the results of _their_ decisions? Specifically, they have the final say regarding where houses, apartments, roads, shopping centers, ad infinitum, are built. Yet the ultimate fault is with people who simply chose to live in certain locations, rather than bad capacity planning prior to their homes being approved to be built?


The rest of the article discuss government solutions and metered lights though. Misleading, populist headlines and first paragraphs are staple of modern journalism.


Driverless cars can't come to market fast enough.


One of the purported benefits of driverless cars is that they can drive together in packs. This article would seem to contradict the benefits of that claim. OTOH they would be less prone to the erratic behavior that can create standing waves of congestion.


They can drive together in packs because they can brake and accelerate in a synchronized way, instead of creating waves of accelerating and braking cars.


Driving on lengthy highways in Australia is interesting, at least in the state of Victoria, where enforcement of the speed limit is very strict. The limits of 100kmh and 110kmh are enforced within a mandated speedometer error limit that's very tight for new cars. Older ones have looser accuracy, predating the legislation that enforces the error limit.

One if the consequences is that on long stretches of road, cars that are very slightly quicker catch up with others. Then, because there's safety in numbers, and because other cars normalise speed through a kind of meaning speed check, clusters form. Overtaking is tantamount to breaking the law, though some attempt to creep by, leading to reduced distance between vehicles, and increased stress for drivers due to reduced reaction time.

Under these kind of circumstances, a strong argument could be made for driverless cars, if only to avoid the increased potential for collision due to driver error. And it demonstrates for one embedded in the packs that form, the kind of experience that we can expect.


I've been on roads where this happens in Britain, when a police car is driving at the speed limit (generally quite loosely enforced on motorways) with people behind forming a cluster which won't overtake out of fear. Quite frightening, and surely combining the worst of all worlds in terms of accident risk (both a high risk of an accident and a high risk of an accident causing fatal or serious injury).

I asked my aunt, who is a police officer, about this, and she said in general the police try and drive slowly (well below the speed limit) on the motorways to prevent this phenomenon from occurring. I wonder whether the guy I saw was doing it deliberately or had just forgotten..


Volvo has been testing a "road train".

(http://green.autoblog.com/2012/09/18/volvo-finishes-satre-pr...)

(http://www.bbc.co.uk/news/technology-18248841)

It's an interesting idea, even though not as advanced as Google's car.

I'd like to see results of testing, especially for the standing wave phenomenon, but also how a truck driver copes with driving "a vehicle" that is many times longer than what they're used to driving.


Pretty interesting. I wonder if any non-technical approaches have been tried - like educating drivers to move to a right-side lane if they're slow. At least on the 101, it seems like so much of the congestion would be alleviated if slow people stayed out of the left-side lanes. It'd be safer too since it'd remove the incentive for drivers to weave through traffic.


Many commuters shouldn't have to be on the roads at all. Fiber is cheaper than freeways.


An alternative are variable speed limit systems: http://en.wikipedia.org/wiki/Speed_limit#Variable_speed_limi...

They can be used to slow down cars heading into a traffic jam to reduce the arrival rate so the traffic jam can dissolve. Roads with a lower speed limit also fit more cars (need less safety distance) - driving slower means less traffic, so traffic jams can be avoided by reducing the speed limit.

These systems are very wide spread in Germany, covering most Autobahns around major cities.


Dynamic pricing might be a good solutions (with reasonable maximal prices). However, it would be hard to introduce them because of public pressure.


We have dynamically price toll roads and HOV lanes in Houston. I don't recall many objections when they were implemented.


why would dynamically pricing tolls cause less traffic? Especially in the morning rush hour - you gotta get to some place by a specific time, you have to pay the toll no matter what it was (with reasonal maximal prices of course). its only when there is an unreasonable maximal price (or no maximal price) that people's behaviour would change - imagine a toll of $500 during peak. It would put off people from going on the road.


Well that's true, but the article already points that taking a few % drivers out of roads could make a huge difference.

See: http://www.ted.com/talks/jonas_eliasson_how_to_solve_traffic... You don't need anywhere 500$, most likely 2-5$ toll is good enough.


In Seattle I noticed that traffic is always light on Monday, but some places get jammed other days of the week. Apparently, there are many people who don't have to be somewhere on Monday, and those same people chose to drive during rush our on other days of the week. What's up with them anyway? If you don't have to go to work, why drive anywhere during rush hour?


It could be that people with somewhat flexible schedules end up going into work later or earlier on Monday than average, spreading out the traffic enough to keep it moving fast.

Also, it could be that there are people who don't normally work on Monday, which is actually fairly common in many service industries. Many restaurants are often closed on Mondays, for example.


Moreover, more Mondays and Fridays are taken off either for "sick" leave or vacation. It may only take a small percentage to make a noticeable impact on traffic flow.


"These commuters .. travel long distances together in the same direction like schools of fish"

Sounds like what they need is a train, tram or bus.


Reactionary boomers in Marin county have been slow-rolling the regional rail plan (SMART) for decades. Don't hold your breath.

If SMART ever comes into any kind of decent service it should relieve a significant demand among north bay-to-sf weekday commuters. Note, however, that SMART service to Larkspur Ferry is "planned" and "phase 2" which means probably not in my lifetime.


"This has enormous potential," said study co-author Alex Bayen, a UC Berkeley professor of electrical engineering and computer science. "These findings are going to come into practice in the near future. This is not just a scientific study."

``... not just a scientific study.'' Reinforce engineering v. science stereotypes much?


I yearn for the day when driver-less cars become the norm. But then, I'm sure the same people will find ways to jailbreak the software and enable reckless driving behavior :(


Why wait for driverless cars when there's public transport which lets you enjoy the same benefits?

Isn't it sad how pefectly fine we're with hauling 2000+ lbs of hardware just to get 150 lbs of flesh from point A to point B in the morning and back in the afternoon.


the single biggest downside of public transport is that you don't control the schedule (nor where it goes - but lets ignore that for now).

A driverless car is a taxi that's cheap and effective. You don't have to park it, you don't have to figure out the route. You just hail one, get in and type in the destination. Its sane public transport.


Actually the schedule part rarely matter for commuting to work since those routes usually have a high density of buses/trains. In Stockholm during rush hour the subway arrive every 1 to 5 minutes, commuter trains every 7 to 15 minutes, buses usually every 3 to 15 minutes.

Not controlling the destination usually matters more since that means you may have to change train or bus several times.


the "same benefits", so long as you, y'know, ignore all the benefits of cars.


So, basically, the suburbs are the problem?


No, it's the workdays, especially 9:00 AM.

Companies should get tax breaks for randomizing the start time of their workdays.


It is the combination of both, plus the lack of good public transport.


The title of the submission is deeply misleading, though it follows some rather ridiculous claims in the article-

Instead, they come from a few outlying neighborhoods and travel long distances together in the same direction like schools of fish -- clogging up not only the roads they drive on, but also everyone else's.

The notion that they are to blame is asinine. A highway is being utilized beyond capacity, but picking out any group on the highway adds dramatic narrative yet little insight. The suggestion that you remove "just 1%" is surprisingly naive as well, reminding me a bit of this Onion piece - http://www.theonion.com/articles/report-98-percent-of-us-com...

If the highway had less utilization, it would cause many who use mass transit (like light rail) to choose the highway. Rinse repeat. It is how highways in virtually every metropolitan area eventually reach a point of saturation, that new point becoming the natural balance.


Not necessarily. When a highway is close to capacity, it only needs to be blocked for a short period of time for it to back up traffic for miles. So if there's a rush from a particular suburb, it could very well cause traffic problems at a specific time.


When a highway is close to capacity, it only needs to be blocked for a short period of time for it to back up traffic for miles. So if there's a rush from a particular suburb, it could very well cause traffic problems at a specific time.

No traffic is more natural, per se, than any other traffic. Again you've repeated the claim that you can single out users and attribute, essentially, blame.

This study is most certainly bunk. Someone thought it would be novel to try the sort of traffic analysis that every single traffic department in the world does, and of course they're going to try to discern some profound finding, like "lots of people come from lots of people to hit the city core at 8:30am".


> Again you've repeated the claim that you can single out users and attribute, essentially, blame.

Where did I claim that? You seem to be missing something, somewhere.

The gist of the article is that large bunches/waves/schools of traffic travel in from outlying suburbs. I imagine that where that pack travels, jams ensue, and anything that you can do to smooth out that peak will help traffic flow more smoothly.

And they're using cell phone signals/GPS to track drivers down to the individual level. Pretty sure not many traffic departments do that sort of analysis, even if you type it in italics. Mostly it's just pressure sensors to measure average flow of particular roads.


Right. Which is why the only sane way to deal with traffic is to ramp up the tolls until the utilization falls below 90% or whatever the threshold is. I don't care if the rush hour toll on the Bay Bridge is $50. It will cost less, overall, than having thousands of people stuck in traffic for hours.


I think you're exaggerating somewhat disingenuously. I commute over the bay bridge and it generally flows well, "stuck for hours" just isn't realistic. It's already $5-$7 which seems quite enough given that I already pay local taxes which are supposed to be for infrastructure.


Or, you know, investing in infrastructure where it is needed.


As there's no alternative in public transport (that I'm aware of), if you charge $50 to cross the bridge, you will make it impossible to work in SF for people living North and earning below a certain threshold.

So these people will either have to move, or find a different job. That seems like a somewhat extreme outcome.

Offering attractive alternatives to commuting by car might be a better way to reduce congestion on the roads.


Public transport (BART + AC Transit) already carries the majority of transbay traffic, so I don't understand your claim that there is no alternative to driving.


You've not refuted the article but merely described it as naive, ridiculous, etc. Whilst it's true that a quiet highway attracts drivers, that doesn't mean that reducing the drivers this article discusses won't deliver benefits.


>You've not refuted the article

I clearly indicated the issue with the article. If there are 50,001 people at a 50,000 person stadium, you can't point to one guy and say he is the reason it is over capacity. Further, there is no one who doesn't understand that less drivers on the highway = a faster commute. But therein lies the classic unintended consequence: Whatever you do to improve the commute leads to more traffic. Twin that highway and you'll see a massive boom in suburban building, for instance.

Or we can select 1% of the drivers on the highway, call them the problem, and call it a day.


I think that the article is saying they have a disproportionate impact on traffic. They are more directionally-biased, spend more time on the road, and have a greater spatial and temporal overlap with other drivers' commutes.

All of those things apparently contribute to traffic more than people who travel in random directions at random times.

Also, it means that it is (potentially) more of a fixable problem. If you can target those specific sources and destinations of problem drivers, you can try to offer them alternatives. Even if they are replaced by other drivers, many of those other drivers may behave more randomly, and the roads will have better utilization.


You're missing the point, which is that the the maximum carrying capacity of a road is a function of the skill exercised by the drivers.

The issue revealed here is that even a small number of unskilled drivers can do terrible things to the flow of traffic on busy roads. In other words, crowded roads have a very low tolerance for bad drivers.

Since the number of disruptively bad drivers is so remarkably low, and their negative effect is so stratospherically high, a program designed to improve their skills could save astonishing sums of time and money.

So it's not about "calling them the problem and calling it a day." It's about actually fixing the problem. And you don't need to single people out to do this. Rather, you set the bar for licensing at a level that they'll need to do additional work to cross. Given how few of them there are, this can be done with no disruption to 99% of the drivers on the road, and major benefit to 100% of them.


The article specifically mentions that it's not driver skill which is the issue:

  These commuters aren't necessarily slow or bad drivers. 
  Instead, they come from a few outlying neighborhoods and 
  travel long distances together in the same direction like 
  schools of fish -- clogging up not only the roads they 
  drive on, but also everyone else's.


It may be time we start introducing reverse traffic cameras. Ones that issue fines for not following a minimum speed in peak commute hours for the vehicle that is unquestionably the first in a pack responsible for the jam.

Proximity calculations, leading vehicle frequency, trailing vehicle average speed, etc., using existing technology already in place that can be leveraged and innovated on to accurately determine culprits.


Did you read the article? There is no single car that causes a jam. In fact according to previous research http://math.mit.edu/projects/traffic/ (see "Conclusions" at the bottom), even if all cars follow the same rules a jam can still occur. They can also occur in very simple situations like in this video https://www.youtube.com/watch?v=Suugn-p5C1M


The article has nothing to say about needlessly slow drivers, just about traffic volume overall.


You can see how the title of the post might give the idea that it was about specific drivers. My first thought when I read it was "I bet it's those god damn hippies hyper-mileing in their priuses", but then I clicked through to the article and it failed to confirm my preconceived opinion.


Sure, so did I, but then I read the article.


Oh yeah you mean women?




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

Search: