V
erlyn Klinkenborg, who regularly contributes interesting and well-written little essays to the New York Times Op-Ed page, writes today about four-way stops and what a surprisingly successful tidbit of human cooperation they are:
What a four-way stop expresses is the equality of the drivers who meet there. It doesn’t matter what you drive. For it to work, no deference is required, no self-denial. Precedence is all that matters, like a water right in Wyoming. Except that at a four-way stop on the streets of Rancho Cucamonga everyone gets to take a turn being first.
The underlying theme here is nothing less than the importance of rational games playing to a functioning society. As poker players, we tend to focus on game theory’s competitive applications in zero-sum situations, but game theory is equally as applicable to cooperative interactions that realize non-zero-sum benefits. (For more on this subject, see my review of Robert Wright’s Nonzero.)
In fact, I just record a video for Poker Savvy Plus yesterday in which I used traffic lights as an example of a Nash Equilibrium. The really remarkable thing about the four-way stop is that it is largely self-governing, as opposed to the traffic light, where drivers obey orders from (literally) on high. Yes, there are rules for how to behave at a four-way stop, but their application in a particular situation are almost always left to the individual judgment of the drivers. There is no flashing light or other signal to tell you when it is your turn and when you must defer.
Yet it works, and traffic flows. Contrast this with the roundabout, which is nasty and brutish, if not solitary or short.
But then again consider the case of two lanes of traffic that have to merge into one. It’s well known that the most efficient method for traffic flow as a whole is the zipper technique, and yet this doesn’t stop the pesky individual from stubbornly plunging ahead, when he should be letting his neighbor in, to the greater disadvantage of all. It’s tough to say whether the chief motive is self-interest or rather some perverse pleasure in unruliness for its own sake.
If you haven’t read the book Traffic yet you should, gets into this and the many other idiosyncrasies of how things play out on roads, very interesting/ terrifying.
I would argue the zipper merge isn’t actually an equilibrium, which is to say that one player can unilaterally improve his outcome by not complying. From a sort of Rawlsian original position, it makes sense that we all agree we all want everyone to comply with the zipper, but the only consequence to not complying is the scorn of other drivers. While this could have implications in an iterated game (they might cut you off in future iterations of the game), real world driving situations usually don’t entail a likelihood of “playing” again with the same drivers.
Running a stop sign, on the other hand, usually entails an irrational level of risk.
Poker isn’t always zero-sum either…there are many situations, especially in tournaments, even totally within the rules, when coalitions form between certain players to the detriment of other players.
True. To be honest, though, I don’t think I’ve ever actually seen game theory used to analyze those situations.
And those are actually still zero-sum situations, right? If you and I are big stacks with three players remaining at the final table, and we avoid playing big pots with each other, we are both gaining EV that is lost by the third player.
Cooperation at a four-way stop is truly positive-sum. No one wins in a car accident.
> Contrast this with the roundabout, which is nasty and brutish, if not solitary or short.
Do you not drive when you travel? As someone born/raised in england and living in the US for a long time now, I find that while 4 way stops are much more “co-operative” than roundabouts … roundabouts are _much_ more efficient.
Merging two lanes into one sucks everywhere though :).
You could also look at the zipper like a prisoner’s dilemma. If I can cooperate or not and “villains” has the same options, non-cooperation will be the best option for me (assuming the goal is to move as fast as possible) if I’m not sure what villains will do. So while for all the drivers the best possible outcome is when everyone cooperates, for me as an individual, non-cooperation is the more profitable option in the sense that I can move faster than if I would cooperate.
I find this stuff really interesting cause the rational choice for a single person (in the sense that every person maximizes his return) leads to non-optimal outcomes for the community.