Traffic jams are not caused by flaws in road design but by flaws in human nature. | Hacker News Reader

Traffic jams are not caused by flaws in road design but by flaws in human nature.

nytimes.com

31 pointsrobg36 comments

"Adding lanes or roads is a short-lived fix. Widen one highway, and drivers from another will defect."

Thereby making both highways have less traffic. The fact that people come when capacity is improved shows that you're lessening the load on other roads!

"Traffic jams are not, by and large, caused by flaws in road design but by flaws in human nature."

Not accounting for human nature is a flaw in road design. Maybe it's just too expensive to account for it, but traffic design is an exercise in understanding driving habits.

Most of the facts in this article are right on and I bet the book is worthwhile, but the writer of the article didn't seem to really think through what he's saying. It's sad when good concepts get ruined by weenie-thinking.

>Thereby making both highways have less traffic. The fact >that people come when capacity is improved shows that you're >lessening the load on other roads!

Yeah: on other roads whose traffic problem was not as bad as the one you just expanded.

Other roads with traffic problems bad enough that drivers avoid them if they believe a new alternative has appeared.

Right. For example, say you have a street network with a main street at capacity (IE, stabilized at crowded, but endurably so), and four outlying streets that eventually lead to the same destination, but have a less appealing route (slower, toll stops, plagued with potholes, pick one). The outlying routes may not be crowded at all, but if the main road is widened, it will probably absorb even more traffic from the outlying streets.

You could make this into a graph problem, if you like, or draw analogies to finding parking.

Paradoxically, adding routes can actually make traffic worse for everyone. Suppose you have two equal-time routes between A and B. If a shortcut is added connecting halfway on route A to halfway on route B (obviously these are not direct-as-the-crow-flies routes with identical speed limits or this wouldn't work) Travel time for everyone increases by 50% (whether you take A, B, or A-B).

I read an article before (tried to find it but couldn't) about a mathematician and [his/her] grad student who had calculated the theoretical maximum cost of this sort of thing and that 50% is the worst it gets, but that's still pretty bad.

This is called "Braess' Paradox"

http://en.wikipedia.org/wiki/Braess'_paradox

Maybe I'm not understanding the article here, but could this apply to the problem with aggressive lane-changers on a highway?

When people (I find this in myself) get on the freeway, it seems like they tend to get clear over to the fast lane right away. Shortly thereafter traffic starts to back up in that lane and the next slower lane looks like it's winning -- so a bunch of people move over, causing that lane to slow and the original lane to speed up. The resultant pattern is an oscillation of stop and go traffic in the lanes in question.

It seems like if this doesn't exactly apply that it at least speaks to what might be going on. I wonder if there has been any research in disallowing lane changes in areas prone to stop and go traffic.

The article about the 50% maximal cost used that exact situation as the attention getter.

I've always wanted highway response crews to come equipped with unfoldable response screens. Something like a nice 20' screen which would block off the view of the accident from the opposing lane. I've always thought a considerable about of time, money, and perhaps life could be saved by such a policy.

It was nice to finally see an article that explained why I don't see such a thing.

I thought about this, and I don't think it would work. From the response crew's perspective, they want a massive traffic jam, because the cars are moving slower and paying more attention. So the response crews who would have to put up the screen would have an incentive to oppose it.

I sorta agree, and I sorta disagree...

Here in Seattle I've been held up for over an hour on the 520 bridge because of an accident THE OTHER DIRECTION. Having screens would prevent rubber necking on the other side of the highway (with a large cement barrier in between).

Also, I think you can do a lot to the screen itself to increase driver safety. Make it 15' high and entirely yellow/black diagonal stripes and it'll catch driver's attention and get them to slow down. TV and radio educational ads can help too.

What about for opposing lanes? A wreck going the other way is still a sight and both directions will slow to sneak a peek.

The article also mentioned that because of the spectacle, drivers are paying less attention to the road and looking at the wreck, causing secondary accidents.

Then again, a tarp with emergency vehicles is another reason for drivers to slow down and look at something, according to the article. The trick is to find something so mundane that drivers won't bother slowing down.

I think an advertisement printed on the tarp would help to avoid the rubbernecking. We're already great at ignoring billboards and other ads.

Then again... because so many people take the time to slow down and look at the wrecks, why not take advantage of the extra eyeballs and put up signs with ads on them?

That may be the best idea I've ever heard. Bravo.

"The fact that people come when capacity is improved shows that you're lessening the load on other roads!"

False. There's a concept called Induced Demand. (http://en.wikipedia.org/wiki/Induced_demand)

There are many drivers who stay off the roads because the cost of congestion is too high. For instance, people who won't take a given trip if it takes longer than 20 minutes. By increasing road capacity the trip time goes down to 15 minutes. This induces demand from the crowd that desires <20-minute trips. As they join the crowds on the road, they bring congestion back to its equilibrium level of a 20-minute trip.

Even if people bring the equilibrium back up, the extra capacity made their lives better for those whose trips become worthwhile. The idea is not to minimize traffic, but to maximize welfare.

Interestingly the core problem of highway congestion is that drivers decelerate faster than they accelerate.

An example: Person A slows down a little on the highway for some reason - maybe she sees something on the side of the road, maybe her cellphone goes off, it doesn't really matter. Person B behind her slows down a little as well to keep her clearance, person C does the same and so on. Now person A speeds back up to her original speed (when she is done talking on her phone, watching stuff on the side of the road, or whatever) but she doesn't accelerate quite as fast as she decelerated - this is perfectly normal behaviour - notice it next time you take a trip in your car. Person B does the same thing when she sees person A accelerate and so on. Before long you have a whole row of cars that need to brake because the cars in front of them have decelerated quickly and accelerated slowly. This is what causes most traffic jams on highways.

This is also why a traffic jam moves backwards on the road as time passes.

Could it be any other way? Could you design a road to induce acceleration and slow deceleration without affecting safety?

You could make all roads downhill.

In both directions! Where's Escher when you need him? :)

A simple and effective way of dissolving a traffic jam is to keep going at the same speed - let me explain:

You see congestion up ahead, and you slow down to, say, 40mph. This creates some room in front of you in anticipation of the inevitable braking. You then drive through the traffic jam at a constant speed with no acceleration or deceleration, and the people behind you will have no choice but to follow (or change lanes which is the weak point of the solution) What you will see at the other end (if you could go back and look) is that you have dissolved the jam because you have prevented people from decellerating quickly and accelerating slowly.

Several people have tried this successfully and it works.

It's really fun when 3 or 4 people all know and do the same thing. It solve the 'weak point' and once or twice I've been part of a 'team' that cleared a jam that way. It's a weird sort of rush.

This reminds me of an old episode of Scientific American Frontiers.

http://www.pbs.org/saf/transcripts/transcript904.htm#5

Thanks for the link. I believe it was this episode that influenced a lot of how I drive (merging, staying slow but consistent in stop-and-go traffic, &c). It should be required watching for new drivers.

Many of the research projects into automated highways concluded essentially the same thing and that average journey times could be significantly reduced, and propagation delays effectively removed, by everyone travelling at near similar constant speed.

The problem however is not everyone will do this e.g. some people simply love to speed, and changing behaviour in people is harder than changing the rules they should follow.

Some of the projects successfully proved the concept in real world usage of the technology, but issues due to insurance (who will be liable in case of accidents) and infrastructure cost seem to be the main barriers at present.

Yes, people are risk-averse.

Yeah, you're pretty right about that. That's one thing I notice driving to school every day. The average speed of cars traveling on the freeway after a highway distraction is significantly lower than the average speed coming into it. I've never understood it. It's as if people have forgotten how fast they were initially traveling. Very frustrating.

Traffic jams may have complicated causes for any fixed number of cars on the road, but it's worth noting that they don't often have traffic jams in rural Kansas. The fundamental problem of traffic is that there are too many cars on the road, and that's because road usage is priced too low. The predictable result is queueing, i.e., traffic. Most of the fancy investigations into the psychology of drivers, etc., are simply higher-order corrections to this basic 0th-order result.

Public transportation is the answer to traffic jams. If the government stopped renting out street-side parking for artificially low prices then the cost of parking would become a bigger incentive to use public transportation.

Agreed. Pricing parking correctly would be one big step. A potentially even bigger step would be to repeal the medallion laws that currently give taxis weird cartel status in most cities. This would allow anyone to offer transport for money, i.e., you could form jitneys (http://en.wikipedia.org/wiki/Jitney). I secretly want to find a large city with bad public transport (Los Angeles? San Jose?) and start a jitney company---which, on its way to making billions, would solve the problem of urban mass transit. (Talk about "Be good" and "Make something people want"...)

He's touched on some reasons, but definitely not all of them.

A big part of traffic is that communication doesn't happen - because it can't. I can't tell the driver behind me "I'm going to be slowing down in 10 seconds, get ready." Likewise, it isn't easy to convey to someone merging onto a freeway that I'm going to let them in ahead of me so we both end up slowing down so that we're both still in contention with each other until someone takes the initiative.

Now, if only we had some way that cars could communicate with each other and the road so that they could regulate speed and allow operations such as merging as efficiently as possible. . .

People tend to act more selfishly in cars than they do outside of them, much like many people do with the relative anonymity of the internet. I suspect it is because drivers are in semi-protected pockets of their own space, and they're essentially sitting still - there is not much they can do to discharge physical frustration caused by traffic besides yelling, giving people the finger, or driving more aggressively.

Isn't the real problem that a lot of people drive to, and then back from, work at the same time?

The real problem is that a lot of people drive on the same roads at the same time. What they're doing on the roads is irrelevant (although work is probably the biggest cause of the demand for road usage. I assume you're attempting to make a case for staggering work hours so "rush hour" is spread over a longer time).

Roads are an interesting economic model. There are basically 3 primary factors: current demand, current supply, and price. Price is especially interesting because price isn't necessarily dollars (or your currency of choice).

Price = Demand/Supply

So if demand goes up (rush hour, popular events, etc.) or supply goes down (construction, accidents blocking lanes) then the price goes up. Part of the price is paid in taxes that goes to build and maintain the road, the other component of price is time. How much time will you spend driving. If it's rush hour and there was a collision blocking the left lane and you're on your way to a baseball game then the demand is pretty high, and the supply is pretty low, so the price will go up. If this isn't a toll road, then the additional price you're paying is 100% time. Of course the beauty is that you could in theory take a toll road and pay in currency instead of time, although there aren't always toll roads as options.

<sarcasm> I suppose the best solution is A: don't go anywhere, or B: get a transporter like in Star Trek.</sarcasm>

Actually, I wonder if anyone out there has tried making toll-roads where you pay more the more time you spend on the toll-road, as opposed to the more distance you travel. That would certainly provide a nice economic signal to avoid congestion.

Of course, you'd also need to install a large number of speed cameras, to stop the obvious method of gaming the system...

I think that would lead to dangerous situations. You would see people constantly changing lanes to try to maximize their speed. Accidents would happen more frequently.

Reply on news.ycombinator.com