Some interesting things about driving
and a little bit of game theory
Driving is the most dynamically involved thing we do on a regular basis. It’s the biggest machine most of us operate.
“A Book on the Game Theory of Driving” or “What You Can Learn about People By How They Drive” were ideas that I had many years ago. Before we get ahead of ourselves, let’s start with a post!
Now, I have no reason to believe that other people will find this interesting.
I also have no idea how good a driver I am compared to other drivers.
I am writing this down because some of this was hard-won enough for me that I want to remember what was interesting to me about it, or what seems not-obvious to me while it still seems not-obvious enough that I am thinking about it in case it is useful to future-me or anybody else.
When I started out driving, people said that it is very important to see what is going on.
My main question was “how do I see everything?”
The advice I got was,
“You turn your head and look”
“By seeing it.”
“By looking.”
This was both the correct advice and also incredibly unhelpful to me.
I very rarely see someone driving as if they are a chicken, constantly turning their head at every slight stimulus — and those people tend to be the less smooth and less safe drivers. These drivers seem to be over-depending on reflexes and physical dexterity.
Most drivers look straight most of the time, and still know what’s going on. And so “you look by looking” didn’t actually seem to be what everybody is doing.
It became a real question for me, for a long time. I’d get better through practice and experience, but the fundamental circularity of the question and answer stuck with me.
I’m actually surprised that over the years I’d managed to break down “looking” into a few different things.
You can crack open your driver side window.
This isn’t looking—this is hearing.
This is a simple trick that I didn’t learn until years into driving. I got a partner who vapes, and so the windows would generally remain open, even on highways even though everything gets louder and it’s harder to hear each other and you have to make the music louder.
But opening the windows, even just a crack, and even with louder music, means that you can hear when a car is right next to you.
You also are not in your own private little capsule anymore, and so are more energetically in tune with other things outside of your car.
Objects in existence do not vanish in or out of existence.
When I first started driving, I was taught to assume that people and therefore cars can do anything basically at any time. Someone could fall asleep, someone could stop suddenly, someone could not stop and keep going very fast.
Assume that cars are boulders and they won’t see you.
In general this is really solid advice, especially for pedestrians.
At the time I did appreciate it, but took it a bit too literally.
If your model is that anything could happen at any time, then you get a new problem, which is the problem of paying attention to too many things.
So we need some complexity reduction.
It’s really good to assume that cars won’t do the smart thing, and that people do not see things some percent of the time, and do fail to stop or fail to control their car.
It also did help me to remember, though, that though objects can slow down and speed up unpredictably, they don’t just appear or disappear.
This is only partially useful in city driving where people make sudden turns all the time.
But for the purpose of this point, let’s suppose you are on a long straight road. If there was not somebody behind you for 20 minutes, there is not all of a sudden going to be somebody behind you.
This is only true if you’d actually been checking your rear-view mirror pretty often in those 20 minutes.
But if you have, this means you can relax some of the time.
And being able to relax some of the time versus none of the time is an empty-set to one difference when it comes to energy expenditure.
It means intersections and hills are are far more unpredictable than average (the sites where you would have had zero prior visibility of if a car is there because there’s no way you could have possibly seen it), and straight roads are far more predictable than average.
This allows for better heuristics around when you actually have zero way to have seen something very important and need to use a lot of energy paying attention, versus where you have been keeping attention for a long time already and if something was there, you would have noticed it by now.
Generally other people don’t want to be in an accident, either.
This is a modifier on the base point that anyone can do anything at any time. But this is another complexity-reducer that lets you spend more energy on what actually leads to accidents.
If you are driving on a highway, for example, the car in front of you won’t suddenly make a U-Turn and start heading the wrong way on the highway.
(Or if it does, that’s actually fine if you had enough space in front of you for them to suddenly stop—which you always should. And now the erratic driver is….going the other way far away from you. Very good.)
But generally speaking, if you are driving relatively sensibly, other (crazy) drivers will speed and swerve around you assuming you are keeping a constant speed because they want to accomplish their goal of pulling off their stunt. And on the flip side, the sensible drivers will notice that you are driving sensibly and either keep pace with you or drive defensively themselves.
You are not invisible on the road, and people generally will not think it is fun to hit you.
They still might. People can randomly act not according to their own interests, but a lot of the time people will be acting as if they don’t want to be in an accident.
This means that the people who are acting unpredictably on the road, you can notice them and try to not be too close to them, even if they haven’t done anything scary yet. You can identify them and get away from them safely after having marked them as an erratic driver.
By keeping a model that most of the drivers will not be erratic drivers, you’ll be able to notice those that are.
This set of heuristics around people not wanting to be in an accident has helped me a lot with highway ramps.
Let’s suppose you’re on the highway and you see oncoming traffic in the merge lane. Generally you have three options: speed up, slow down, or switch lanes. Which one do you do in what situation?
Advice I’d gotten on this has often been,
do what makes most sense in the moment
or
do what you’re most comfortable with.
On digging deeper, I’d found more detailed descriptions for how to analyze the situation (enough detail for some nice charts!) based on which lane you are located in and what they’re doing.
But between the intense charts and “do what you want,” there is also a middle heuristic.
You see, the person merging wants to end up in a spot on the highway that is not the same spot where you are.
That’s it.
Knowing that this is what the merger is doing means that the calculus you need to be doing involves noticing where that car will end up, and not being there when it does.
That’s how you avoid crashes in merges.
It might not be everything one needs to know, but it certainly has helped me to reduce complexity down to the right level.
So these are three things I’d learned that are expansions on You just look by looking.
What I like about this list is that it shows that useful expansions can be possible.
In this case, I can see that I was expanding into what’s worth looking at, and how not to burn through attentional resources.
I do find that “what” questions can sometimes break a circularity between the question and the answer, when how and why questions fail to make a dent.
Perhaps guessing the “what” questions that your students/coachees/clients might be thinking is a way to become a better teacher.
It is great watching teachers who can adept to many levels.
One reason I like going to mixed-level dance classes sometimes is that an instructor confident enough to have mixed-level drop-in classes is skilled enough to show something to the class, and then make on-the-spot modifications for you when they notice what level you are at. It’s pretty cool to watch.