The Prisoners’ Dilemma and Bus Stops

I started riding the bus in Boston in my second year at Harvard – in the first year, I was too afraid of making a fool of myself to do it. I knew that I’d have to pay for it somehow, and I imagined myself holding up an entire bus of commuters as I tried desperately to shove soggy dollar bills into a slot.1 My change of heart about the bus came after I spent three months in India in the summer of 2015. Hailing autos and trying to explain where I needed to go without speaking any Hindi made the MBTA incredibly tame by comparison.

In any case: I started riding the bus in my second year, but my route is a popular one between 7:30 and 9:00 am. During the winter, when there’s too much snow to ride a bike without some Masshole yelling at you (a story for another post), the bus looks pretty appealing.

So, sometimes you’ll be standing, waiting for it, shivering, and it’ll be so full that it passes right by you.

You have one way that you can guarantee that you get on the bus: you can walk four or five blocks down to an earlier stop. I’ve done this a few times, but every time I have I’ve felt deeply ashamed if the bus does end up skipping the stop closest to where I live.

Here’s my thinking about why it isn’t right to walk to an earlier stop:

Suppose there are just two people, and they’re getting on the back of a motorcycle, rather than a bus. Only one can sit on the motorcycle, so if they’re at the same stop, they’ll flip a coin to determine who gets to take that transportation. For the sake of simplicity, let’s say that each player receives utility 1 if they make it to work on time, and 0 if not. Then, the game they face looks like this:

Bus Stops 1

As this is, I don’t see anything inherently wrong with walking to an earlier stop – if we both do it, we end up just flipping the coin the same as we would before. The only thing that’s really wrong about this is that we only benefit by taking advantage of the other player’s naivety, but they could easily join us at the earlier bus stop if they thought of it.

However, if they can’t walk that far – if they have a disability or if they’ve got kids in tow (suppose you can stick the kids on the motorcycle) or if they’ve just got a ton of bags, the set of choices looks like this:

Bus Stops 2Now I think it’s clear that you’re not justified in moving the earlier stop, since your movement makes them late with certainty. This probably happens pretty often in real life; it isn’t unrealistic to think that some proportion of people taking the bus would find it painful or impossible to travel another few blocks. Since we can’t know whether our companions at the bus stop are in this situation, maybe we should not move to the earlier stop just in case.

There’s a third issue. Consider now a game with 3 stops, where traveling an additional stop costs c. Player 1 chooses first whether to move or stay, and then player 2 chooses. Assume that 1-c>.5, so that it is worth it to move if the other player stays. Then, 1-2c>.5-c. It must also be that .5-2c>-c, since we could add .5 to both sides of that inequality and get back to our previous conclusion that 1-2c>.5-c.

dynamic bus stops

The green path indicates what happens if 0>.5-c: player 1 moves and gets on the bus, and player 2 doesn’t move and doesn’t get on the bus. However, the interesting outcome is the red path – in this equilibrium, both players move until they run out of bus stops, and then they both flip the coin to decide who gets to ride on the bus. [Look up “backward induction” and “dynamic games” if you’re not sure how I came to this conclusion.]

What’s interesting about this? Well, at the end of our dynamic game, for some values of c, both players are left worse off than they would’ve been if they could’ve just cooperated. It’s the prisoners’ dilemma.

Kant’s categorical imperative goes: “Act only according to that maxim whereby you can, at the same time, will that it should become a universal law.” I may be misunderstanding this, but it sounds like “choose no action which would make everyone worse off if everyone chose it.” So, stay at the first bus stop. (This site seems to agree with my interpretation, but this article doesn’t.)

Let’s suppose for a second that Kant says we should cooperate in “Prisoners’ Dilemma” style situations. Public goods provision is one such situation that crops up over and over – why should I pay for a road if I know you’ll pay for a road? If we all think this way, no road will be built without government intervention. However, if we all followed the categorical imperative, we wouldn’t need government public goods provision. I wonder whether Kant concludes that there is little need for government in an ideal world?


1. In the two years I’ve been riding the bus now, this scenario has happened at least five times to me. It is just as mortifying as I imagined it would be, but what I didn’t know is that if you struggle long enough, the driver will either seize your money and feed it to the bus themselves, or they’ll tell you to sit down. This to say: any suffering on the bus is temporary. Back

Advertisements

Charlottesville

I have a slurry of thoughts about Charlottesville, VA.

First Thought: I’ve seen a few tweets from Alt-Right accounts saying that the hate of the Left is just as dangerous and worthy of condemnation as the hate of the Alt-Right. A few posts from liberals quoting religious figures who say that we can’t fight hate with hate. There’s an MLK Jr. quote: “Hate cannot drive out hate; only love can do that.”

If we’re defining “hate” as “hostile actions motivated by intense dislike or prejudice” (from Google), then I’m inclined to agree with this idea. The decisive losers of a violent conflict don’t automatically start agreeing with the winners – if they did, we would never have put Confederate statues up in the first place.

But what if we think about “hate” as just meaning “intense or passionate dislike”?

It seems to me that we cannot get anywhere without this. We need the capacity to label ideas as morally repugnant. You can’t say “Being a white nationalist means that you aren’t a good person” without this kind of hate. (I mean, maybe you’re able to make moral judgments without attaching emotion to them? That’s probably a pretty rare ability to have…)

To say “I am entitled to my beliefs – if you judge me for them, you’re being hateful” is sort of an appeal to moral relativism. Which is weird, since usually liberals are thought of as being the moral relativists, the “anything goes” gang.

Second Thought: I’ve seen quite a few tweets about publicizing the names of the White Supremacists in pictures. If you search “ruin his life” on Twitter, you’ll get a bit of a taste of this.

Eichenwald tweet
One example of a “ruin his life” tweet.

I’m of several minds. The first mind is that if you attend a Nazi rally and you’re photographed, you are an automatic liability to your workplace – they’re completely justified in firing you.

The second mind is that this explanation looks at whether the firm is justified in firing you, but it doesn’t really make a claim about whether you deserve to lose your job. Suppose you attended a Nazi rally and you wore a hood, and your supervisor somehow found out that you’d attended, but no one else knew and there was no chance that anyone would find out. You might still be a liability – your supervisor knows that you have beliefs that make equal treatment of your coworkers and customers unlikely.

What if, in talking to your supervisor, you disavowed your beliefs, promised never to attend a rally again, and generally swore to not be a dick either at work or out. Would your supervisor still be justified in firing you?

I guess the question I’m trying ask is: where does the punishment of “deserving to lose your job” come from? Does it come from a company’s profit maximizing choice, or does it exist outside of that? Do we care about the consequences of you losing your job? We probably don’t care about the consequences to you, since you attended a Nazi rally, so we’re not feeling super charitable, but do we care about the consequences to your family, your firm, etc?

Would we think it was “okay” to keep an avowed Nazi on-board at the company if that company could show that firing him would sink an entire division, leading to more job loss? Would we think a firm was justified in keeping him employed if he was the sole provider to three kids?

And if these kinds of circumstances matter, aren’t we just kind of doing a “maximize the general welfare,” utilitarian calculation? Seems like we should be able to reach a more uniform judgment than this.

Third Thought: “Ruin his life.” Okay, yes, he loses his job. But then when should he be able to get a job again? Should he never be able to get a job again?

I think this isn’t so different from “Ban the Box”. If it is right that someone convicted of second-degree murder who has served their time should be able to get a job, then it must also be right that someone convicted (in the court of public opinion) of bigotry after some time should be able to get a job. Maybe we have different standards for the convict because they actually went to jail and so we can be certain that “justice was served,” whereas the Nazi may or may not have suffered by his association?

I’ll have to think about this one more – I don’t think you can favor banning the box and favor ruining someone’s life for his involvement in a Nazi rally and still be logically consistent, but maybe I’m missing something.

Competing Obligations

I’ve just finished reading Justice by Michael Sandel.

Near the end of the book, Sandel talks about reparations. In a Rawlsian framework of justice, which allows for universal obligations and particular obligations created by consent, reparations cannot be justified. As far as I understand it (and this is the first time I’ve encountered all of the political theories that Sandel is covering, so I may very well be misunderstanding), a system is just if all participants would have chosen it, if they’d been in a contracting situation where no one knew their role in society, what talents they were endowed with, etc. People probably wouldn’t agree to reparations in this kind of society, since they’d know that there’s some chance they’d end up paying for something their parents did, and so reparations cannot be part of a just system.

Sandel’s response to this is that it points out a flaw in Rawls’ political philosophy – we do think that there’s a third category of obligation – we think there’s universal obligation, obligation because you consented or made some contract, and then there’s a type of particular obligation that’s situational. Like, I’m obligated to vote, because I’m a citizen. Or I should call my mother, because I’m her daughter – not because I need to call everyone in the world, and not because I made some kind of contract with her to call her, but simply by virtue of our relationship.

I might be obligated to pay reparations or to make apologies for the sins of my ancestors because I recognize that I’m a member of this society. I recognize that the government from which I now benefit was built on the backs of slaves, and I recognize that many of the prejudices from that time are still held, making it difficult for black Americans to receive equal treatment in society.

So, let’s talk about taking statues honoring Confederate generals down. Sometimes people make the argument that we need to keep these sorts of monuments up in order to remember history. These are clearly people who don’t understand the basic purpose of Wikipedia. Or a public library. Or a museum.

But what about the folks who are more open about keeping such monuments? These are the ones who will tell you that they are proud of legacy of their ancestors. They want to keep these monuments up to honor them. [Peek on a Facebook comment thread about monument removal if you don’t believe that this opinion exists.]

However, I will bet anyone $500 that I don’t have that the very same people who want to honor and preserve the legacy of their ancestors are the ones who would not want the United States government to issue an official apology for enslaving Africans and their descendants. And I’ll bet you another $500 that if you asked one of those people why the government doesn’t owe black Americans an apology for slavery, they would say, “I didn’t own slaves.”

Look – I haven’t fully worked out this screed yet, so I’m not 100% sure that this is entirely logically consistent, but it seems to me: You Can’t Have It Both Ways. You cannot say that you are honoring the legacy of your region and your ancestors without accepting some modicum of responsibility for their actions. What, then, are you honoring? If you completely disavow their actions, then you wouldn’t want to preserve their statues. If you do not completely disavow their actions, then you probably do owe an apology to black Americans, because you’re actively contributing to a prejudiced society.

Now, I think you might be able to get around this conundrum if you were super specific about what it was that you wanted to honor. If you said, I want to honor my great-great-grandfather’s commitment to his children, then it’d be acceptable to have a statue of him, even if he was a slave-holder. But Confederate monuments aren’t honoring some aspect of men unrelated to slavery – they are honoring this man’s willingness to die in defense of slavery. So either you support that or you don’t.

Is Specialization the Point of Life?

(Note: this post might not make a ton of sense – I am ravenously hungry and struggling to make my thoughts connect in a coherent-ish way.)

When I was deciding to take a leave of absence, one of the professors I spoke to said, in the course of our conversation, that the point of life is to specialize. My gut response, then and now, is that this isn’t true. But I haven’t really pinned down why I don’t think this is true.

I guess we need to define some things to determine whether this claim is all the time true, sometimes true, or never true. I’ll define “the point of life” as “an action without which a life is ‘worse.’” This is a lower bar than we could set – I think “the point of life” sounds strong enough that we could define it as “an action without which a life may as well have never been lived.” But if specialization doesn’t pass the threshold set by the first definition, it certainly won’t pass the second.

Then, what do we mean by “specialization,” here? In the context of the conversation, “specialization” meant content expertise. But I think we can be more general with it – you’ve specialized if you have one subject that you know more about than others. Translated: your life will be less worth living if you do not have some subject that you favor over others.

I can’t evaluate this, because there’s no practical way to know exactly the same amount about everything – you’re always “specializing” in something, if only by choosing not to know something else. That is, the only way to not specialize at all is by either knowing everything or knowing nothing.

So, I clearly screwed up on defining specialization. Specialization definitely means knowing more about one thing than others, so let’s say that there’s some threshold. For instance, if you know twice as much about cheese as you know about the next thing you know a lot about, then you are specialized in cheese. But then, we’re saying that if you increased your knowledge of the next category (the thing you know the second most about), all of a sudden you will not be specialized, and your life will be “worse.”

This definition of specialization also leads to crazy outcomes. It should not be the case that learning more makes my life “worse” than it was if I’d known just a little less. Clearly thresholds won’t work. But we’re close – we could say that you’re worse off spending that additional hour working on something that isn’t cheese than you would’ve been if you’d spent that hour learning more about cheese. I think this is probably what my professor was referring to (not about cheese, obvs, but about the marginal benefit of an hour doing something you’re expert in versus something you’re not).

I’ll need to think about this about more, but my feeling is that this would imply kind of weird preferences. You certainly wouldn’t have Cobb-Douglas preferences about your knowledge – you’d prefer to invest in one thing, rather than a mix. You’d also need to see increasing returns to knowledge. We’d be able to recreate it with a utility that’s equal to the maximum of time spent across every possible field. So, I guess technically there’s some preference for which this claim is true, but it certainly isn’t all the time true. And this is all assuming that personal utility is the best way to assess whether a life is better or worse.