Hacker News Reader

Comment by imgabe | Hacker News Reader

> Whereas, when a trail is overcrowded, the ants restrained themselves and avoided joining until things thinned out.

This sounds a lot like how traffic waves [1] occur. When there's a traffic jam ahead, you can often help clear it out just by slowing down to the average speed it's moving. Going a steady 10mph is much better than going 30mph for 1/3 of the time and then being stopped for 2/3 of the time. If everyone can start going smoothly at 10mph it will eventually speed up.

Traffic jams occur because cars are arriving at some point on the road faster than they can leave that point. If the cars coming towards it just slow down to allow the cars there time to leave, the jam will clear.

[1] http://trafficwaves.org/

View on HN

I'm no traffic jam expert, but driving manuals in west coast traffic (Highway/Interstate) has taught me its best to find a large semi-truck, ideally with an empty flatbed trailer, and just follow it. The semi driver doesn't want to ever completely stop (if they can avoid it), they have a much better view since they're somewhat higher up, and they accelerate/decelerate much slower, so its easier to predict their movements.

The result is they tend to continue moving forward without the stop/start of normal cars. I just sit behind them in 1st gear or whatever and since no one wants to "get stuck behind a semi" people don't usually care if you leave the larger gap, so you don't have to deal with drivers sliding in front of you and eating up your cushion of space. My left leg isn't exactly going to atrophy from the lack of clutching, but it certainly does save it a fair bit of work.

Took me a minute to realize by "driving manuals" you meant driving a manual transmission car. At first I thought you meant the behavior you described was recommended by the drivers manual that the licensing test is based off of. :)

Seems like good advice.

> I'm no traffic jam expert, but driving manuals in west coast traffic ...

From what I can gather, I'd rather ride a motorcycle for my commute and lane split.

In Toronto I can ride for seven months of the year (Apr-Oct), and with the more moderate climate of the Pacific Ocean, it'd be possible to do it year-round. The occasional downpour during the wet season can easily be handled with Goretex(r). (I also pedal cycle some days of the week for commuting just to get some cardio.)

I use this method to an extent, but the unfortunate result often becomes people zipping in in front of you when you inevitably have a bit of extra space compared to those around you. This seems to leave you with two problems which are hard to solve simultaneously. The first is the one you mention where you want to keep a mostly constant speed compared to the whole group. The second is the effective need to locally keep up with the cars around you. Solving both problems simultaneously seems essentially impossible. I guess removing self-interest is the only real way (which appears to be what the ants do).

The slow-n-steady method really only works with a critical mass of participation. Unfortunately most people either don't know that it helps or don't care and are happy to just keep stomping the gas and breaks repeatedly.

Slow and steady only works up to a certain vehicle density. Once that has been exceeded, slow and steady simply serves to cause congestion further back in the road than necessary.

There is no way around stop and go traffic once the maximum capacity of flowing cars has been exceeded. After this, leaving extra space is just waste, which is why in more densely populated areas, people "cut" into the space in front, because the only optimization left is to use as much road as possible.

There are other benefits, enjoyment and reducing wear and tear. Increasing delay by a few % in a huge jam is not material.

If everyone did it, it would be material. But everywhere I've seen, other people usually slip into whatever space there is. Which technically does maximize the rate of vehicles passing through a point in the road, as you can take advantage of the various rates of acceleration.

Whether or not the time saved is worth the wear and tear, environmental damage, increased risk of collision, I can't say. I'd say probably not, but it's going to happen anyway.

Not really, as mentioned in a sibling thread. If you’re in a big hurry, simply leave two minutes earlier you’ll be ahead of all the folks you’re worried about.

There is an efficiency gained through having space to change lanes as well, which is ine cause of the slowdown.

Someone always responds with this, but the reality is that it doesn’t matter. Will you arrive later if one car jumps ahead, or ten? Might delay you an extra minute. If that is a problem simply leave one minute earlier, haha.

It's not that. It just makes the thing harder to do. If someone cuts in front of you, they frequently will brake too, and your previous speed will bring you up to them.

Just drop from 7 mph to 6 temporarily. It is possible because I’ve done it weekly for the last two decades, after reading about it on Slashdot.

> I guess removing self-interest is the only real way (which appears to be what the ants do).

This is, in my utopian imagination, the best argument for automated cars. Which works great until a manufacturer, or enthusiast, decides to make an "aggro" AI which takes advantage of the selfless AI...

They can be stable for a long time!

I saw one of these early one Saturday morning on I280 in San Jose. I presume it started with congestion during Friday night commute. It was still there at 7AM - cars adding to the back, slowing to a stop for an instant then starting up again and zooming away from the front. All for no apparent reason (no accident/no other congestion on the freeway anywhere).

Freeways tend to clear out at 3 AM. So I would presume that it started that morning. Perhaps with a minor fender bender an hour earlier.

That said, I try to clear those up when I see them.

The other problem in the Bay Area is that cars don't know how to merge and no one wants to let another driver in front of them.

Packs of 5 or more cars driving close to each other will try to merge in to the freeway with cars on the freeway, who also have little space between them. So eventually everyone needs to brake and it slows everyone else down.

And turn signals are seen as a sign of weakness -- cars will actually speed up to close a gap when you signal.

you can often help clear it out just by slowing down to the average speed it's moving

I'm of a similar belief. However, I've run into self-proclaimed professional traffic engineers who insist the best thing is for you to tailgate the car in front of you as closely as possible. The provided justification was that the more tightly traffic is packed, the sooner people can reach their exit & free up space on the road. I still don't buy it.

Does the steady-speed method work, even with cooperation?

The problem is usually people merging in front, requiring that everyone else slow down. I don't see how steady speed is even really possible there, let alone how it helps.

Let's say everyone has a two-second following distance. Then each lane moves one car by every 2 seconds, and that's that.

As a motorcyclist I help by zipping around, lane splitting at safe speeds, and generally getting the hell outta there. It’s hard job, but someone’s gotta be the free radical.

Before folks complain: I only lane split up to 30mph and at speed differentials no more than 5. Those bikers you see zooming past and scaring the shit out of you are idiots.

I checking out recommend http://amasci.com/ by the same author: Bill Beaty.

Site was made in 1994, and looks like it. My friends and I spend hours on here in high school reading about all his interesting science experiments, and trying some out ourselves.

> Traffic Jams occur because cause are arriving [...] faster than they can leave

This is certainly true, people often falsely start the car instantly and continue to drive just to stop a few meters later. The safe distance not also is a safety measurement but also a good way to at least help to declutter the jam.

>The safe distance not also is a safety measurement but also a good way to at least help to declutter the jam.

After a certain point, frequently exceeded in most cities, using more road capacity can't possibly declutter the jam. It will push the jam further back.

I often think that when exiting a jam (the point where cars begin to speed up) that the best thing one can do for everyone behind you is speed up as fast as you can. I've never modeled it or anything, but does this seem true to the folks here commenting on HN?

Don't speed up faster than the person in front of you, or you might end up having to touch the brakes and create a new traffic wave that propagates backward again.

If I can see the next section of stopped cars, I will slowly accelerate and time it so that by the time I reach them, I don't have to stop and can continue driving at a somewhat constant rate.

I'm not sure speeding up as fast as you can helps because usually there is another jam around the corner.

No.

The barrier is the time it takes people to start to move, not how fast they do once they do move. Therefore start up immediately, but there is no point in zooming ahead. In fact it is counter-productive!

What really matters is reducing how many cars are stopped. Once that gets to zero, the jam can evaporate. Therefore when you see a jam ahead, slow down. Create space. And then try to just barely not have to stop. And when you come away, if there is any sign of a second jam create space and do the same.

The result is you save on brakes, create immediate frustration for any stupid drivers who happen to be behind you, and make life a lot better for cars 20+ behind you.

It's impossible to do this as each and every car has different acceleration and deceleration. The stop and go flow is inevitable. Trying to keep your car rolling is good, especially for wear and tear, but leaving excess space in front is also bad due to increasing the length of the congestion behind. The only solution is for cars to enter the congestion at a lower rate than they exit the congestion.

We rarely accelerate/decelerate to the limit of the car, and therefore should choose.

If just a few percent of drivers did as I suggest, stop and go disappears. In effect slowing in advance moves the congestion back earlier, reduces its size, and makes it easy to have it evaporate.

The reason that traffic jams happen is that freely moving traffic carries more cars per minute than stop and go. So there is a traffic volume where there are two states - everyone moving or traffic jam. But when everyone is moving, even minor slowdowns can cascade and switch states. And once switched, it is impossible to fix until traffic volume dies down and the hard stops are eliminated.

We can't easily fix traffic volume. But the strategy that I suggest eliminates hard stops without reducing cars per minute. And therefore causes the jams to go away faster.

> If the cars coming towards it just slow down to allow the cars there time to leave, the jam will clear.

Yes, but, which problem are you trying to solve? The thing most drivers want to optimise is door-to-door time: slowing down as a solution is a tough sell!

I used to do my bit for this but now my car has adaptive cruise so I let it stop and go as it likes with my foot on the floor.

Reply on news.ycombinator.com