Quantcast

«

»

Jun 08 2007

Highway merging and the theory of evolution

Some time ago, I wrote about the best way for traffic to merge on a highway, say when a lane is closed up ahead. There are those drivers who begin to merge as soon as the signs warning of impending closure appear, thus making their lanes clear. Others take advantage of this lane opening up to drive fast right up to the merge point and then try to squeeze into the other lane.

I said that although people who followed the latter strategy were looked upon disapprovingly as queue jumpers, it seemed to me like the most efficient thing to do to optimize traffic flow was to follow the lead of the seemingly anti-social people and stay in the closed lane until the last moment since that had the effect of minimizing the length of the restricted road. To merge earlier meant that one had effectively made the restricted portion longer.

Some commenters (Gregory Szorc, another Greg, and Jeremy Smith) disagreed with me, saying that what is important is not the length of the restricted road section but the ability of traffic to maintain speed. After all, a single lane of cars can travel quite smoothly at 60 mph for quite a distance, even if there is a lot of traffic. They said that the best thing to do is to merge into the other lane whenever you can do so without significantly losing speed. Clearly this means merging as soon as possible, when traffic is still light, rather than following my suggestion of waiting until the latest moment when traffic is heavier and merging has to be done at a low speed.

The last month I have been doing a lot of highway driving and have been observing this again and realize that I was wrong and the commenters right. When traffic is light, people can merge at any point and not back up traffic because the speed at which they merge is close to the normal speed. So it seems that the key feature is the ability to maintain speed and to merge when you can do so, which means when the traffic flow is light, which usually is well before the actual lane closing. In fact, I think that highway workers should post signs many, many miles ahead of the restriction and recommend that people merge as soon as possible.

But highway signs alone are not going to be enough to have the desired effect. What is needed is widespread public awareness of the benefits of merging well before you actually have to.

Of course, there will always be people who ‘cheat’ and try to go as far as possible along the closed lane and thus end up slowing traffic at the merge point and destroying the benefits for all. What can be done about this?

Interestingly, this phenomenon parallels the problem of explaining altruistic behavior using evolution by natural selection. It is easy to argue that a group benefits if all its members practice some particular trait, say by sharing food equally all the time so that everyone survives in both good times and bad. But the catch is that evolution by natural selection works on the basis of what is good for a single organism, not for groups, because it is an organism that has genes and propagates it. And that means that a cheater (i.e., someone who, when he has plenty, hides some of his food without being caught) benefits more than the others and is more likely to survive. If this tendency to cheat is an inherited trait, then over time cheaters will come to dominate in the population. Evolutionary biologists have developed theories on how to explain the evolution of altruistic behavior in the face of this seeming advantage for cheating.

In the case of highway merging, if everyone, without exception, follows the early highway merging rule, then long bottlenecks could be a thing of the past, unless traffic is so heavy that merging at normal speed is just impossible. But the occasional cheater will get a short-term benefit of getting a long stretch of open road, while the people behind him get the negative effects of having him slow down traffic at the merge point. So he gets the benefit of others merging early while others bear the cost of his cheating, making cheating an advantageous option to that single organism.

Of course, I am not suggesting that selfish and inconsiderate highway driving habits are inherited traits that will spread in the population by being passed down to the inconsiderate driver’s children via his or her genes. But they could be like a ‘meme’, a mental virus that, like a gene, is a replicator that seeks to propagate and increase its incidence in the population, which in this case consists of the minds of people. This meme would encourage people to benefit themselves in the short-term at the expense of others, even though in the long term they too lose when someone else practicing the same behavior slows down traffic ahead of them.

7 comments

Skip to comment form

  1. 1
    Thought Shaman

    Actually, natural selection can also explain some aspects of altruistic behavior. If the survival of the genes depends on a group’s survival rather than on each individual member, then altruistic behavior confers an advantage. For instance, it is harder for a carnivore predator to fight off a group of herbivores acting as a unit than a single member.

  2. 2
    Ben

    The traffic analogy is interesting Mano, and I remember one evolution writer claiming that altruism and cheating were essentially a zero-sum game, and really depend on the individuals being cheated.

    For example, let’s assume a merge point where everything is bumper-to-bumper, and the traffic is moving very slowly. The merging traffic can do the “right” or altruistic thing and jump in early, thus taking their proper place in line, or they can speed to the end and jump in. Whether this succeeds or not really depends on the drivers already in traffic. Do they close ranks to let the cheater in, or do they not? Since drivers are for the most part unrelated and uncommunicative to each other, there is no real group pressure to shut the cheater out, so he wedges his way in and shaves 10, maybe even 15 minutes off of his commute.

    Like the pack of herbivores though, if this same behavior is attempted on a funeral procession, the cheater will never be allowed to enter. The cohesiveness of the group of drivers is too strong, and they won’t stand for him/her “cheating”, and the cheater gets hosed as all the cars behind him zoom by and he has to accelerate from a stop.

    Different situations, different success. In driving, let’s be honest, the cheaters and aggressive are favored because little exists to counter them. As an aggressive driver, I realized this and do it all the time. It irritates the hell out of my fellow drivers, but I win and they lose, and that’s OK with me. If I try the same tactics on a test though, where the other people have more to lose by my cheating than a couple minutes on their commute, I will undoubtedly lose, and lose big. So I follow the rules because doing so is ultimately more to my advantage than not.

    As environmental conditions change, so does the “fitness” of any given strategy. People like us in the industrialized world are rewarded for specializing, where tribal or more primitive societies are rewarded by having jacks-of-all-trades. They need to minimize the effects of loss; a nation of 300 million people does not.

    Right now it just pays to cheat in traffic. =]

  3. 3
    Mano

    Ben,

    You are quite right that if left alone cheaters will win out and that it depends on how others respond. Thought Shaman in the earlier comment gives an example of how it might work in one case.

    But rather than being a zero-sum game, the development of altruism is more like an arms race. The rest try to foil cheaters by creating defenses, then cheaters improve their cheating methods, the others respond with better methods of foiling them, and so on.

    I am planning a more detailed post in the future discussing how this works because altruism is an important quality that deserves an explanation as to its origins.

  4. 4
    Jon

    I still agree with your first theory, Mano. A merge point in traffic is essentially a bottleneck and everyone merging too soon actually restricts the flow and backs traffic up further back along the highway. To quote the DMV website on this subject: “To keep a good pace in heavy traffic, merging should work like the teeth on a zipper. One vehicle merges; a vehicle already on the highway passes; another vehicle merges; a vehicle already on the highway passes, and so on. Of course, in this day and age of aggressive driving, it may not always work out this way.” Using reasonable and safe driving methods, I’d say use the extra lane until the cars become 1:1. If you don’t and merge back when it’s 1:2 or higher= crappy zipper.

  5. 5
    Defensive driving

    I would at least partially second Jon on this. Or your original position rather.

    My take is that if everybody starts merging directly at the sign, this will just move the point of congestion start from actual lane end point to the sign.

    “In the wild” traffic does not merge freely usually, you already have a congestion at the sign more often than not, and merging at that point does not give any “altruistic” advantage. There is really no difference to almost standing traffic at what point you merge to it. But there is a big difference to you. And I don’t call this cheating, or aggressive, or any other word with negative connotation – it is just smart to do so…

    Just my 2 cents :)

    Regards, Misha

  6. 6
    Los Angeles Personal Trainer Leon Lavigne

    Getting them to put up the road signs sooner would be the hardest part.

  7. 7
    Mike

    Excellent article. As a long haul trucker, merging is one of my pet peeves. Yes, when traffic is still moving, merge early. If there is already a long backup, waiting until the merge point and taking turns is the best option. Nice write up.

    Here’s a page with some additional driving tips: http://www.drive-safely.net

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>