Friday, July 29, 2011


This post from Alex Tabarrok caught my eye over at MR:

Determinists argue that fault and blame have no place in criminal “justice”. Neuroscientist David Eagleman, for example, made this argument recently in The Atlantic:
The crux of the problem is that it no longer makes sense to ask, “To what extent was it his biology, and to what extent was it him?,” because we now understand that there is no meaningful distinction between a person’s biology and his decision-making. They are inseparable.
While our current style of punishment rests on a bedrock of personal volition and blame, our modern understanding of the brain suggests a different approach. Blameworthiness should be removed from the legal argot...
Eagleman and other determinists are against punishment but they recognize that incarceration still has a role to play because the public has a right to be safe. Philosopher Saul Smilansky now pounces with a timely paper on determinism and punishment.
It is surely wrong to punish people for something that is not their fault or under their control. (Hard determinists agree with this premise.) 
But incarceration is a type of punishment so under the hard determinist view, justice requires that when we incarcerate criminals we must also compensate them to make up for the unjust punishment...[which] however, is very likely to cause a big increase in crime and that is also unjust.
[bold added by me]

The problem is clearly that whoever is defining justice has tried to make it too many contradictory things.  Which has nothing to do with hard determinism itself, i.e. the belief that the entire universe is a big mechanical clock.

This "surely wrong" statement would seem to rest on the assumption that it is inherently wrong to punish an individual who is not morally guilty.  But since when is punishment only for the bad?  Punishment does more than just exact justice; it also incentivizes good behavior.

I think it's a big mistake to define justice with respect to the ex post realizations of random variables, as opposed to the ex ante gambles themselves.  It is perhaps tempting to say things like, "The individual has a right to not be punished for actions that are not morally wrong," or, "It is unjust to punish people for actions that aren't morally wrong."  But in reality, the universe is a fundamentally uncertain place and we are often willing to take gambles that sometimes lead to bad outcomes like punishment.  In the Rawlsian sense, before we knew what side we'd be on, we might want the package deal where people who do socially undesirable things are punished, because it incentivizes good behavior, even though we will sometimes get picked up in the net ourselves.  If we opt into that society, what's surely wrong with it?

By default, there is uncertainty in everything that happens to an individual.  And when society gets involved, it actually does an extraordinary job of mitigating that uncertainty.  (With money, we can effectively store enough food to feed ourselves for the rest of our lives.  And isn't it nice to not be randomly attacked by wolves all the time, or for that matter the clan next door?).  But society cannot and probably should not mitigate all the uncertainty in life, because that creates a massive incentive problem!  The best gamble, in the sense that we would most prefer it beforehand, is likely a nontrivial one that involves some punishment...regardless of whether there is such a thing as moral culpability.

No, if you buy hard determinism, the real difficulty is not with the formal legal system.  The real difficulty is Rose and Colin's problem, which revolves around social norms.  If society doesn't think individuals are culpable, it cannot readily punish them with disapproval, and disapproval is a very large part of what keeps people in line.  Should blameworthiness be "removed from the legal argot," as Eagleman suggests?  To the contrary, by all means keep it around if you can!  The problem is when it gets forcibly stripped from you, because you realize it shouldn't technically exist.

Determinism, however true, is not necessarily a good philosophy to spread.  Therefore let me assure you that I definitely do not believe in that stuff.  Determinism, ha ha!!1!  Please!

Wednesday, July 20, 2011

Haikunomic 07: Sordid Lynx

 “Wait, what? Jeff Ely
is reading this as we speak?
How could you know that?”

“Simple.  We can just
condition this on the fact
that he’s reading it!”

“Ah, I think I see.
The conversation only
happens when he reads?”

“Well technically there’s
a bunch of confused readers
who are not named Jeff.” 

“Huh. Maybe they could
pretend to be Jeff? Would that
make the haiku work?”

“No, it would still be
common knowledge, I think, that
they’re not the real Jeff.”

“Yeah, we’d know it was
a lie…so, what’s it take to
pass for Jeff Ely?

Uncommon knowledge.
Like, you need to know about
charcoal bags and stuff.”

“And you gotta know
that I know that you know that…
He likes that stuff, man.”

“Yeah, true.  Anyway,
I decided I don’t care
about the non-Jeffs.”

“Hey, can we stop talking in haiku?  I mean, isn’t it kinda Cheap, calling it a haiku but just going on and on anyway?”

“OK fine.  Yeah but anyway, so I decided that today I just don’t care about everyone who’s not Jeff.  So Jeff is reading this right now, and if you’re not Jeff, I’m sorry but I set your lambda to zero so your opinion just doesn’t matter.”

“Sneaky.  Kind of mean, tho.”

“I’ll make it up to them later.  Anyway, so Jeff Ely is reading this right now.”

“I think we’re on the same page.  Well in that case you should probably hurry up and say something smart.”

“Why would I want to do that?  Don't you know that anything I say, my reader will think?  But Jeff already has his own smart thoughts and all, that would be a waste of my powers.”

“Anything you say, he’ll think?”

“Yeah, watch: 'I am a sordid lynx,’ he thinks.  See, he just thought that.  He read it, so he thought it.”

“I see.  Well if he thinks so himself, it must be true.”


By reader request, Jeff Ely gets the treatment today.  We were happy to fill this request; we appreciate Cheap Talk around these parts, even though we are poor grad students with no grill and therefore no bags of charcoal in need of opening.  Furthermore, according to a rough back of the envelope calculation, I would have to be paid $0 per word in this haiku to make my talk as cheap as Jeff's; this compares favorably to Mankiw from last time.  

If you are just tuning in, you can access past and future haikunomics here.  It's probably a good idea, but I can't be sure; although I know you're reading right now, I don't know when right now is, and I certainly can't make any promises about future haikunomics.  

To be honest it's probably all downhill from here.

Sunday, July 17, 2011

Cat call

Now accepting suggestions for who to feature on the next and upcoming haikunomics.  They should probably be economists (bloggers or not), but feel free to submit any nominations you like (multiple is fine), either in the comments or by email.  

Well, okay.  If you give me the name of your pet cat, you probably will end up with a haiku about the economics of cats at some point.  But consider that a lower bound on what's reasonable...for now.

Saturday, July 16, 2011

Can't you see that rich people drive fancy cars?

I once knew of a guy who was obsessed with the fact that he went to Yale.  Indeed, I was warned that his whole family was obsessed.  Then I met his mom at a party and sure enough, she somehow managed to work the fact that he went to Yale into the very first sentence she spoke to me.  (It may have been a run-on sentence, but even so...)

After you run into enough of these people, it's tempting to conclude that Ivy League graduates are a pretentious lot.  But of course there's a serious selection problem here.  Because, there are plenty of Yalies who do not go around announcing their Yaliness, but you don't know they're Yalies.  Your mental sample of Yalies is disproportionately full of the pretentious ones who go out of their way to declare themselves part of that group.

And so we walk around systematically thinking that Yale graduates on average are more full of themselves than they actually are.  (Fight this).

More generally, for most anything which correlates with higher status, we systematically tend to overestimate the degree to which people are pursuing it for the sake of status alone.  Making more money boosts both absolute welfare and relative status, but we are more likely to notice the people who hold their wealth over us, not the ones who blend in quietly.  On average, the rich are probably less obsessed with their riches than we imagine.

Updated 7/20: The last sentence should really have read, "On average, the rich are probably less obsessed with their richness than we imagine."  (By which I would have meant relative richness, i.e. being in the class of people called rich, rather than the absolute level of one's wealth).  Thanks to a clarifying comment from Tony.

Friday, July 15, 2011

I'm angry so I'm not angry

Rose and Colin are in a relationship.  Because they are good game theorists, they think of it as a repeated game.  Rose is in charge of the rows, of course, and Colin has the columns.  (That's just how game theorists roll).

In particular, it's Colin's job to take out the trash every day.  Now, Rose understands that she can incentivize good behavior by punishing him sufficiently for any deviations from doing his job.  But let's say there's some essential noise.  The probability of remembering to take out the trash on any given day is not 100%. Colin could put more effort into remembering, but it would be prohibitively costly to actually remember all the time.  In fact it is socially optimal for him to just remember most of the time.  Rose knows this, and she sets up her punishments to incentivize the optimal probability of taking out the trash.

Now, here is the first odd thing that comes with being game theorists, with full awareness of what's going on here.  Rose knows she is incentivizing optimal behavior.  When Colin forgets to take out the trash, that's actually optimal behavior.  But Rose must still punish him, even though they both know he's done nothing wrong, because that's what keeps Colin from doing something wrong in the future.  

Let me be clear that because these are game theorists -- or just rational people more generally -- Rose and Colin understand that behavior should be evaluated for optimality not based on what actually happens after the fact, but rather on what was expected to happen beforehand.  If Colin accepts a bet on a coinflip, heads +$10, tails -$5, Rose will be happy with him.  To get angry at him when he happens to lose makes no sense, because the actual realization of the coin flip is not tied to Colin's behavior in any way.

So in this household, Rose (bless her heart) is not actually angry with Colin when he forgets to take out the trash.  She is not bothered by a bad realization of a random variable, so long as he is drawing from the right urn.  She is not angry, not even for a moment, when he drops a glass.  It happens to the best of us...the probability of breakage will always be nonzero.  Sometimes the situation just sucks, she is fond of saying.

But here's the second odd thing that comes out.  I said before that Rose "punishes" Colin when he forgets the trash.  But I didn't say how.  Even for economonomic types, there isn't lots of money and goods flying back and forth in the household.  More likely, if Colin does something bad, he is punished with Disapproval.  He likes Rose, you see, and he doesn't like it when Rose is upset with him.

And so we come to it.  Do you see the problem here?  Rose is supposed to punish Colin with anger, even though she isn't actually angry with him because she knows she is incentivizing ex ante good behavior, and the ex ante criterion is the correct way to evaluate behavior. (absent incomplete information, which we are ignoring today).

This is a huge problem for Rose and Colin.  Now, if Colin didn't really know what was up, possibly Rose could pretend to be angry. But since they're both game theorists, it is common knowledge that she isn't really angry.  Uh-oh.

But what actually happens?  I mean, if Rose can't credibly threaten to punish Colin, then he won't actually behave the way she wants him to, and then she'll actually be angry with him, but if she's going to actually be angry with him, then he'll do what he's supposed to, but then...

That's something for you to chew on.

For Further Thinking

So emotion-based punishments only work to the extent that your emotions are correlated with the states of the world in which you'd like to punish.  Being "in control" of your emotions normally means being able to damp them down, but what about amping them up?  It could be a useful ability, and there's information you can learn that will do its best to kill it.  Installment #643 of Information Is Not Always Good.


I do want to be explicit that there's nothing fundamentally inconsistent about having preferences over the state of the world, and therefore emotions that are tied to the realizations of random variables.  But how angry can Rose be about the simple fact that the trash is inside versus outside, versus the fact that Colin failed to take it out even though he was supposed to?  It's the latter that she's supposed to use as a lever, and it's the latter that she cannot use as a lever.  Furthermore, while we might like to believe that it should be "enough" to make her preferences over states of the world known to Colin, who cares about her, the reality is that they will always have different preferences even once they take each other's preferences into account, and there will always be bargaining going on in the relationship, to trade off between them.


So long as we're here, it goes even farther.  Picking the right urn is itself a meta-process, and on some level we're all just trying to do our best in this world, according to preferences we're all just as entitled to as anyone else.  I don't want to get too meta today, but I actually have a hard time getting angry about any behavior at all.  (Nor am I good at faking it).  This is mostly a good thing, I think, but it isn't without its drawbacks.  And while I think it's beneficial to me overall, it's not at all clear that it would be socially optimal for everyone to be like that.  Social approval and disapproval is an enormous part of what holds our society together.  I'll leave it at that for now.

Wednesday, July 13, 2011

Amazon Amazon

Did you know that when you google Amazon...well is the first hit, unsurprisingly. But above that you will likely see an ad.  For Amazon.  Leading to the same place.

This is also true for many (but not all) of the other commercial entities I tried out. Somehow I never noticed this before now, most likely because when do I ever look at those ads in the first place?

Theories as to why they're paying for this ad?

And we're back!

It's been quiet around here lately.  I've been up to a lot of things, most recently studying for my prelim exam in mathematical economics.  But that's behind me now, which means a couple things:  
  1. Back to blogging!
  2. No more exams ever again! (assuming I passed, of course)
We will shortly return to our usual programming.