43 comments

[ 3.2 ms ] story [ 89.8 ms ] thread
Traffic in India looks similar to ants: https://m.youtube.com/watch?v=RavLjmWdMK4

But really it's each driver executing their own internal 'greedy algorithm' to get themselves off the road as soon as possible.

When traffic jams occour, it's the blokes who break the rules (the ones drive on the wrong side of the road or drive aggressively) who get to their destination first.

Kind of similar to the traffic in Saigon, Vietnam https://www.youtube.com/watch?v=gKLWZjBu2iQ
I was in both of those places (Saigon and all over India) over the last few month. That's the first thing I thought of when I read the headline.
Wow, there are a lot of scooters/bikes in Saigon! If everyone would drive instead only cars I guess traffic throughput would collapse.
It definitely is dangerous, but the video seems to be running at 2x, perhaps to make it seem more dangerous than it is actually. The pedestrians walking around are a give away.
Just to clarify, this video is sped up. But yea, I get the point.
That doesn't actually look noticably crazier than traffic at some times and places in NYC. The main difference is NYC has more traffic lights and human traffic enforcement officers -- but only 4 wheeled vehicles are highly likely to obey those -- pedestrians mostly don't and 2 wheeled vehicles vary, so you can certainly see reminiscent scenes here. I think I might actually prefer to traverse that intersection than say, biking in a hurry from 14th to 57th on 8th ave. during evening rush hour. Most of the conflict looks like just the usual "through traffic versus left turners" which happens anywhere there aren't signals with separate left turn phases
In the Netherlands and Germany I've seen traffic jams develop due to drivers slowing down to observe an accident on the other lane. I know this is illegal to do in both countries, but it doesn't keep anyone doing so. I've never been able to not slow down myself because the laws of physics didn't allow me to pass through the observers. Stricter and more expensive ticketing might help.
> Stricter and more expensive ticketing might help.

I doubt it, and… please no. Demerit points maybe. Fines disproportionately punish the less affluent and incentivise the wrong attitude to enforcement.

Day fines could work.
For those who don't know the term: it is using fines proportional to the amount of money one has available for spending on a day, as they do in Finland, where speeding tickets can go over $100,000 for the very affluent (http://www.theatlantic.com/business/archive/2015/03/finland-...)

Yes, you cannot hope to reach complete equality of the impact a fine has on the drivers, but you cannot do that with demerit points, either. A driver who needs his car to make a living will likely be affected stronger than a driver who has a decent alternative.

Sure, but there have to be lower and upper limits to it. Or would you want a rich guy to pay $750k? Hmm, maybe there shouldn't be an upper limit.
I agree with what the sibling comments said regarding fines, and it would be interesting to see statistics before/after Germany introduced hefty fines/penalties for not keeping the required distance relative to speed.

One problem I've observed with uncoordinated speeds on German Autobahns is drivers who seemingly switch to their left lane without noticing a faster vehicle approaching which then has to brake suddenly which in itself is a danger to everyone. Out of anything, I feel like autonomous vehicles with coordination will make lives less stressful.

I would prefer a more general change: studies show that traffic jams and accidents happen more frequently if cars are going at widely different speeds. So the consequence should be not only to ticket drivers who go too fast but also drivers who go too slow. So instead of setting a speed limit we need a speed range and deviation from it in either direction should be punished. Obviously this would only be for freeways and certain highways.
I've been in one such jam in the Netherlands where there is an upper limit at which everyone consistently (for the most part) drives their cars on the highway. So I don't know if that would help.
This isn't really pertinent to the article. It's not about "stream smoothing" through more judicious use of gas and break.
Please don't ticket people for driving at a safer speed while next to an accident.
I assumed he was talking about an accident in the opposing lanes. I've seen that a lot in the US. Divided highway with a good 30 ft of grass between opposing lanes and people slow down because of an accident on the other side. Nothing but rubbernecking.
Gazing at some accident twenty meters away with no influence on your side of the Autobahn instead of looking at the road in front of you is the opposite of safe driving.
> In the Netherlands and Germany I've seen traffic jams develop due to drivers slowing down to observe an accident on the other lane.

That's nothing out of the ordinary. In the US the behavior is known as "rubbernecking".

>Stricter and more expensive ticketing might help.

I wouldn't hold my breath. After all, you are attempting to change what appears to be fundamental human behaviour, curiosity.

Wouldn't a surprise "lane reduction" due to an accident alone be enough to account for the traffic jam?

Imagine that there's an accident in a single lane. The remaining lanes now have to take in all the other drivers from the accident lane, while these drivers decelerate and attempt to take the remaining lanes at speeds lower than the average speed. Yet single car cutting in line could trigger the deceleration of the entire remaining lane (like in "traffic wave" models). The cars in this lane then try to cut into another lane, and so on, such that the whole traffic is impaired. And this keeps going on for every cohort of cars reaching the accident area, producing the traffic jam.

This is my intuition of it, at least.

You've accurately observed what everyone in rush hour sees firsthand; how the poison distribution occurs in real life.

Worse, if you're the one in that vanishing lane, behaving in those optimum for you ways, you're the winner.

Automated driving, particularly if the automated driving lanes only obey 'safe driving speed' (not an actual limit) is the only real hope aside from a reduction in the traveling population.

Speaking of that, the best solution to this issue and that of skyrocketing prices for housing of any sort is build-baby-build (in all the cities).

Nitpick: it's spelled Poisson distribution. We are probably looking at a Poisson process, actually.
You're right, but this is something different. The OP talks about accidents on highways with a hard shoulder in the middle of the road, in a lane going in the reverse direction (I'm fairly sure that interpretation is correct, but it requires some reading between the lines)
That's right.
To be clear, I'm talking about highways with a solid divider and no way to change lanes except taking an exit. Also, I've been in one such jam in the Netherlands where everybody drives at a consistent 130 km/h max.

Reading the responses here, autonomous vehicles going in a coordinated fashion might be the only viable solution.

Yet another reason why self-driving cars will be an improvement over human drivers.
Yet another irrelevant analogue being trumped up as "evidence" of something yet to happen. I'll convert once I see actual vehicles on actual roads.
I've seen actual self-driving cars on actual roads. But they are haven't hit the market yet.

What do you mean by convert? Are you unsure whether they will happen? Skeptical about whether they will improve on humans, or whether it's a good idea in the first place?

>>> I've seen actual self-driving cars on actual roads.

No you haven't. You've seen supervised cars with pilots ready to take over when needed. No manufacturer yet has dared release a truly hand-and-eyes-free vehicle onto public roads.

OK, if that's your definition, that's fine. (It differs a bit from consensus definition, of course.)

Could you answer the spirit of the questions I asked about your `conversion'? Thanks!

Ants aren't constrained to a few lanes and can climb over each other if necessary and even build bridges out of their bodies.

Ants do not have to obey a two second rule to prevent hitting the ant ahead of them.

Ants don't have to wait for a good sized gap when merging into a busy lane of ants.

Please, no more of these antics. :)

Ants also cooperate.

Ants share the road only with their own kind.

Ants are blind, mostly.

Ants don't really care if they live or die.

Despite all this, they still collide with each other constantly. If they were the size of cars, these collisions would cause constant injury. They are not an appropriate analogue.

Yeah, gee. Too bad I'm not using my car to forage for food, and follow scent trails.

Oh, if only...

Ants are not perfect though.

There's a phenomenon called "ant circle" in which ants form a self-reinforcing circle of pheromone trails, causing their entire colony to collapse.

https://www.youtube.com/watch?v=mA37cb10WMU

Or in cooler terminology, the circle of death. It can also happen with sheep and larger herd animals, although they're at least smart enough to (usually) stop and take a break when they get tired.
Instead of lemmings?