When Defector-Punishers meet Cooperator-Punishers in the Streets of Paris

ETA: I’ve got to find a new source for the video.

The Prisoner’s Dilemma is a game theory experiment that explores the conditions under which people cooperate or defect against each other. I assume you are already familiar with the details.

In a single game of the Prisoner’s Dilemma, defection or cooperation depends a lot on the folks involved’s individual personalities, but in multi-iteration games (games where people play multiple times against each other,) cooperators generally punish defectors, which eventually leads to mutual cooperation–the best outcome.

One of the implications of this finding is that people will cooperate more with people they’ve had (and will have) repeated interactions with than with strangers. I learned this the hard way when I went from playing board games with my highschool friends to playing with a group of strangers, and promptly got defected on in a plan to split the profits from Broadway and Park Place. The whole business sounds silly in retrospect, but believe me, if I had ever encountered this person again, I would have defected–hard–against them.

Punishing defectors leads to a stable system of mutual, beneficial cooperation.

But in when experimenters took the multi-iteration prisoner’s dilemma abroad, they discovered an unexpected (to them) behavior: cooperation-punishing. These are people who defect against cooperators, leading to mutual defection. Mutual defection is also stable, but shitty.

Cynically, we might say that this is less about “punishment” as that the cooperation-punishers smelled a sucker and decided to benefit themselves. They may also have been unable to realize that their opponent would probably change their behavior in response to the initial defection due to insufficient ability to model other people’s thought processes, and so simply continued doing the thing that had worked once, even once it stopped working. (IE, they were dumb.) As a practical matter, though, we can refer to this as “punishing cooperators,” since that is the result.

Societies with smart people should converge on mutual cooperation; societies with dumb people converge on mutual defection.

What happen when these two styles meet, and people from societies where defectors-get-punished meet people from societies where cooperators-get-punished?

In an actual prisoner’s dilemma experiment, it is of course obvious whether you cooperated or not, but let’s think about this in the much fuzzier terms of normal human human interactions, where there is far more debate and uncertainty about intentions (and effects.) If, in the normal course of your daily life, most people cooperate and defectors are defected against, and then suddenly someone starts defecting against you, your first response may be to soul-searchingly examine whether you did something to cause the defection. For example, suppose you are part of a social group that normally eats dinner at each other’s houses once a week, and suddenly one week, someone doesn’t invite you to their gathering. A reasonable response would be to ask yourself, “Did I do something to piss them off?”

Many of the most liberal people I know seem completely incapable of figuring out, on an instinctual level, whether or not they are being taken advantage. They get hurt and say something like, “I don’t want to abandon my faith in humanity,” or they try to “examine their privilege” even harder. It is painful to watch; sometimes I just want to yell, “It is okay to hate people who have hurt you!”

There’s a saying that a conservative is a liberal who’s been mugged. What’s the word for a person who’s been mugged and is still a liberal?

Last weekend, my housemate and I were mugged at gunpoint while walking home from Dupont Circle. T

… when a reporter asked whether I was surprised that this happened in Georgetown, I immediately answered: “Not at all.” It was so clear to me that we live in the most privileged neighborhood within a city that has historically been, and continues to be, harshly unequal. …

What has been most startling to me, even more so than the incident itself, have been the reactions I’ve gotten. I kept hearing “thugs,” “criminals” and “bad people.” While I understand why one might jump to that conclusion, I don’t think this is fair.

Not once did I consider our attackers to be “bad people.” I trust that they weren’t trying to hurt me. In fact, if they knew me, I bet they’d think I was okay.

One of my friends was homeless for 20 years and never mugged anyone. I know people who have been reduced to shoplifting food because they did not have any, but still never pulled a gun on anyone, broke into their house, threatened someone, or stole from them.

Poverty does not make good people rob others at gunpoint. This is bullshit, and an insult to all of the people who have endured poverty without hurting others.

I have not been able to write about the Paris Attacks and their fallout since they happened, mostly because I try not to write posts that look like this: DJGGGYWEEERRRRRRRRK!!!111!!

But I came upon this graph today, of French attitudes toward Muslims before and after the Charlie Hebdo attacks:

From Pew Research Center, "Ratings of Muslims rise in France..."
From Pew Research Center, “Ratings of Muslims rise in France after Charlie Hebdo…”

I regret that I do not have more recent data from France, but I do have some from America:

From Pew Research Center, "Ratings of Muslims rise in France..."
From Pew Research Center, “Ratings of Muslims rise in France after Charlie Hebdo Attack…”

Take a look at those conservatives!

Forget about “It’s better to be feared than loved.” Apparently being feared makes you loved.

To be fair, I have noticed a habit among certain people to delicately start a sentence, “Now, I like Muslims, but…” or “I like blacks, but…” which may be driving some of this. Anti-racism has become such a dominant value that even conservatives cannot express the pain and horror they felt from 9-11 without first throwing out an anti-racist disclaimer (not that it works, of course. They are always guilty of racism, no matter what they say.)

When the Japanese bombed Pearl Harbor, did anyone feel the need to stand around, declaring, “I’m not anti-Japanese, but…” ?

The mere idea of having a single blanket encapsulation of 1.6 billion people–1/5th of the world’s population–is idiotic. I reject the question. I have no opinion of Nigerians that applies equally to Bosnians, nor of Kazakhs that applies to Indonesians. Likewise, I have no opinion of Christians that covers Haitians, Norwegians, and Ugandans; Mormons and Eastern Orthodox; no single coherent opinion of Hindus, Buddhists, or Jews. But I have very strong opinions about the people I consider my enemies.

According to Newsweek (8-26-14):

One in six French citizens sympathises with the Islamist militant group ISIS, also known as Islamic State, a poll released this week found.

The poll of European attitudes towards the group, carried out by ICM for Russian news agency Rossiya Segodnya, revealed that 16% of French citizens have a positive opinion of ISIS. This percentage increases among younger respondents, spiking at 27% for those aged 18-24. …

Newsweek’s France Correspondent, Anne-Elizabeth Moutet, was unsurprised by the news. “This is the ideology of young French Muslims from immigrant backgrounds,” she said, “unemployed to the tune of 40%, who’ve been deluged by satellite TV and internet propaganda.” She pointed to a correlation between support for ISIS and rising anti-Semitism in France, adding that “these are the same people who torch synagogues”.

In lieu of the video that was supposed to be here, let’s just say that I think ISIS is pretty darn evil.

 

The Police

The flipside of criminals is, obviously, the police. No consideration of criminality can be complete without some consideration of the folks making the technical determination.

That involves more people than just the police, of course.

Much of the time, the police do a decent job. But the Legal/Justice System is out of whack. Has been for as long as I’ve been paying attention.

Some major issues:
1. Major corporations use lawsuits to destroy their competition. Maybe good for them; definitely bad for humanity.
2. Corporations and the wealthy pass laws that benefit themselves at the expense of everyone else.
3. The wealthy have way more ability to use the system to their advantage.
4. Politicians pass a lot of laws just to sound good, with terrible effects.

5. Prosecutors pursue convictions even when they know they probably have the wrong guy; judges are complicit. (This is a biggie.)
6. Plea bargaining.
7. Prisons are shitty.

It is fairly easy to imagine the police (and related folks,) after dealing day in and day out with criminals, begin to think less in terms of specific crimes, and more in terms of making sure that dangerous “criminal types” stay behind bars for a good, long time. So what if the guy didn’t commit this particular crime? He looks like a criminal; he probably did something.

Statistically speaking, they’d probably be correct. Take recidivism rates; knowing that 84% of carjackers will commit another crime, how eager would you be to let a carjacker out of prison? Would you try to find some reason to keep them in?

But what about the 16% who never commit another crime again in their lives? (Or at least, don’t get caught). It is unfair to imprison them for the rest of their days for the sins of others, after all. Next we’ll be imprisoning people for pre-crime.

The Justice System involves a lot of people who have to deal with a lot of very dangerous people in stressful situations; mistakes will be made. To be clear: this is a hard job, and one mistake can have quite disastrous consequences. In a nation of >300 million people, you will hear horrible stories of things gone horribly wrong no matter how great things are overall. Thus the importance of determining whether we are just hearing about tragic but basically random accidents, or regular, systematic abuses that we can actually do something about.

Unfortunately, I suspect we tend to focus on the former, rather than the latter–probably because the latter involves reading things like DOJ statistics, which appeals to most people like moldy pie.

In a good world, people can trust the police. Our justice system, unfortunately, does not inspire trust. This really needs to be changed, or else I just don’t see how the country can function.

Man arrested more than 70 times; released again

Nicholas Limpert from Spokane has been arrested more than 70 times. Now charged again, prosecutors are working to find a way to keep a career criminal off the streets. 

His first offense? Murder, at the age of 15.

“Prosecutors told KREM 2 News on Wednesday that they plan to get tough on Limpert with his latest trial.”

“Larry Haskell, the Spokane County prosecutor, said Limpert’s history of crimes could final[ly] be catching up to him.”

70th time’s the charm, right?

“”I think every day what I could’ve done to prevent myself from getting into trouble,” Limpert told a judge.”

Not get caught; clearly this man needs to work on his skills at not getting caught.

It is possible that Limpert actually has some desire to not get into trouble, but simply lacks any ability to control his actions and act in a non-criminal manner. It is also possible that Limpert just doesn’t want to go to jail for his criminal activities. Either way, nothing’s going to change.

Sorry, Les Mis: Criminals gonna Criminal

“3 in 4 former prisoners in 30 states arrested within 5 years of release” (from the Bureau of Justice Statistics press release, April 22, 2014.)Inspired by my recent musings, I thought I would refresh my memory on recidivism stats–I have a vague memory that murderers tend not to recidivate, (murderers tend to stay in prison for a very long time) and that car jackers do, but it’s a bad idea to make claims based on vague memories of old data.

So here’s what the press release has to say:

“An estimated two-thirds (68 percent) of 405,000 prisoners released in 30 states in 2005 were arrested for a new crime within three years of release from prison, and three-quarters (77 percent) were arrested within five years…

More than a third (37 percent) of prisoners who were arrested within five years of release were arrested within the first six months after release, with more than half (57 percent) arrested by the end of the first year.”

We could probably save some time and effort if we could effectively identify those third before releasing them. HOWEVER, I don’t know what percent of these people are being re-arrested on parole violations that the rest of us might not really consider “crimes”, like missing a meeting with one’s parole officer or forgetting to register one’s address.

“Recidivism rates varied with the attributes of the inmate. Prisoners released after serving time for a property offense were the most likely to recidivate. Within five years of release, 82 percent of property offenders were arrested for a new crime, compared to 77 percent of drug offenders, 74 percent of public order offenders and 71 percent of violent offenders.”

I’m guessing violent offenders spent longer in prison, and thus were older when released.

“Recidivism was highest among males, blacks and young adults. By the end of the fifth year after release, more than three-quarters (78 percent) of males and two-thirds (68 percent) of females were arrested, a 10 percentage point difference that remained relatively stable during the entire 5-year follow-up period.

Five years after release from prison, black offenders had the highest recidivism rate (81 percent), compared to Hispanic (75 percent) and white (73 percent) offenders.”

So, while while the chances of being a criminal vary widely between groups, criminals from all the groups recidivate at fairly similar rates. This suggests that we are probably actually arresting the subset of people who are criminals most of the time.

“Within five years of release, 61 percent of released inmates with four or fewer arrests in their prior criminal history were arrested, compared to 86 percent of those who had 10 or more prior arrests.”

Maybe guys with 10 prior arrests shouldn’t be released until they’re well over 40?

Some finer grain on recidivism by specific crime, after five years (note: this does not tell us the new offense,) from the PDF:

Violent: 71.3%
Homicide: 51.2
Murder: 47.9
Nonnegligent manslaughter: 55.7
Negligent manslaughter: 53.0
Rape/sexual assault: 60.1
Robbery: 77.0
Assault: 77.1
Other: 70.4
Property: 82.1%
Burglary: 81.8
Carjacking: 84.1
Fraud/forgery: 77.0
Drug: 76.9%
Possession: 78.3
Trafficking: 75.4
Public order: 73.6%
Weapons: 79.5
Driving under the influence: 59.9

Looks like my vague memories were correct. Murderers are the least likely to recidivate, probably due to the personal nature of many murders (you’ve got to really hate that guy,) and murderers being older when released, but they are still folks who aren’t great at solving inter-personal problems or running their lives. Rapist probably figure out non-illegal ways to have sex, or else get old enough to be less interested in it. Drunks probably learn to call a cab when drunk.

Relatively speaking, of course. A 50 or 60% recidivism rate still isn’t something that inspires great confidence. To be clear, again, this is not data on how many released murderers commit another murder or how many released rapists commit another rape–this is arrest for any crime. A further breakdown of re-arrest by new crime vs. old crime would be interesting.Carjacking, by contrast, looks like the Xtreme sports of crime–people attracted to this form of violent thrill-seeking seem unlikely to change their spots or find more legal alternatives.

On a related note, The role of parenting in the prediction of criminal involvement: findings from a nationally representative sample of youth and a sample of adopted youth.

From the abstract: The role of parenting in the development of criminal behavior has been the source of a vast amount of research, with the majority of studies detecting statistically significant associations between dimensions of parenting and measures of criminal involvement. An emerging group of scholars, however, has drawn attention to the methodological limitations-mainly genetic confounding-of the parental socialization literature. The current study addressed this limitation by analyzing a sample of adoptees to assess the association between 8 parenting measures and 4 criminal justice outcome measures. The results revealed very little evidence of parental socialization effects on criminal behavior before controlling for genetic confounding and no evidence of parental socialization effects on criminal involvement after controlling for genetic confounding.

In other words, looks like my basic thesis is holding up. Overall, I suspect it is far easier to fuck up a kid so they don’t meet their full potential (say, by abusing/neglecting) than to get rid of the effects of negative traits. It’s probably best to try to work with people’s inclinations by finding them life-paths that work for them, rather than trying to mold them into something they aren’t.

Lotteries

Oh look, W. Hunter posted about Lotteries.
“Lotteries can be useful natural experiments; we can use them to test the accuracy of standard sociological theories, in which rich people buy their kids extra smarts, bigger brains, better health, etc.

David Cesarini, who I met at that Chicago meeting, has looked at the effect of winning the lottery in Sweden. He found that the “effects of parental wealth on infant health, drug consumption, scholastic performance and cognitive and non-cognitive skills can be bounded to a tight interval around zero.” “

I count this as evidence in favor of my theory that winning the lottery does not have a significant effect on a person’s likelihood of committing crimes (eg, drug consumption,) and that the converse, becoming suddenly poor, probably also has no major effect.

There’s also a somewhat garbled reference in the article to an interesting 1800s land-lottery in Georgia; I recall the longer post on the subject and recommend it if you can find it.

Criminality–a WIP; your thoughts appreciated

I’ve been thinking about criminality, inspired by a friend’s musings on why didn’t he turn to crime during his decades of homelessness and schizophrenia. My answer was relatively simple: I think my friend just isn’t a criminal sort of person.

To clarify what I mean: let’s assume, similar to IQ, that each person has a “criminality quotient,” or CQ. Like IQ, one’s relative CQ is assumed to basically hold steady over time–that is, we assume that a person who rates “Low CQ” at 20 will also rate “Low CQ” at 30 and 60 and 10 years old, though the particular activities people do obviously change with age. Absolute CQ decreases for everyone past 35 or so.

A low CQ person has very little inclination to criminal behavior–they come to a full stop at stop signs, return excess change if a cashier gives them too much, don’t litter, and always cooperate in the Prisoner’s Dilemma. They are a bit dull, but they make good neighbors and employees.

A mildly CQ person is okay with a few forms of petty crime, like shoplifting, underage drinking, pot smoking, or yelling at people. They make fun friends but they litter and their party guests vomit in your bushes, making bad neighbors. You generally wouldn’t arrest these people, even though they do break the law.

A moderately CQ person purposefully does things that actually hurt people. They mug people or hold up conbinis; they get in fights. They mistreat animals, women, and children. They defect on the Prisoner’s Dilemma. They make shitty friends and shitty neighbors, because they steal your stuff.

A high CQ person is a murderer; they have no respect for human life.

A common explanation for criminality is that poverty causes it, hence my formerly homeless friend’s confusion. Obviously poverty can cause people to commit crimes they wouldn’t otherwise, like stealing food or sleeping in public parks. But in general, I suspect the causal arrow points the other way: criminality involves certain traits–like aggression and impulsivity–that make it hard to keep jobs, which makes criminal people poor.

Good people, reduced to poverty, remain good people. Bad people, suddenly given a bunch of money, remain bad people.

I’m not sure how one would test the first half of this without massive confounders or terrible ethics,  but the latter half seems relatively easy, if you can just find enough petty criminals who’ve won the lottery and aren’t in prison–although now that I think about it, it seems like you could look at before and during data for people affected by essentially government-induced famines or poverty events. Just a friendly wager, but I bet Jews during the Holocaust had crime rates lower than American inner-city-lottery winners.

But “criminality” is a complex trait, so let’s unpack that a little. What exactly is it about criminality that makes it correlate with poverty?

Subtraits: aggression, impulsivity, low intelligence, lack of empathy, low risk aversion, high temporal discount.

Any of these traits by themselves wouldn’t necessarily induce criminality–people with Down’s Syndrome, for example, have low IQs but are very kind and have no inclination toward criminality (that I have ever heard of, anyway.) Many autistic people are supposed to be low in empathy, but do not desire to hurt others, and often have rather strong moral compasses. Low risk-aversion people can just do xtreme sports, and high-time preference people can be bad at saving money but otherwise harmless. Even aggressive people can channel their aggression into something useful if they are intelligent. Impulsive people might just eat too many cookies or dye their hair wacky colors.

But people who have more than one of these traits are highly likely to engage in criminal behavior.

However, these traits do not appear to be randomly distributed (thus, criminalitty is not randomly distributed.) Rather, they seem to belong to a complex or archetype, of which “criminality” is one manifestation.

This complex has probably been more or less the human default for most of human history. After all, chimps are not especially known for not tearing each other’s faces off. And saving up wealth for tomorrow instead of eating it today doesn’t make sense if the tribe next door can just come in and steal it. In a violent, chaotic, pre-state tribal world, “criminality” is survival.

Over at Evo and Proud, Frost has been talking about his paper on the genetic pacification of Europe via executing lots of criminals, and various counter arguments, ie, In the wrong place at the wrong time? and How many were already fathers?

To summarize, briefly, Frost proposes that the precipitous drop in W. European crime levels over the past thousand years or so has been due to states executing criminals, thus removing “criminal” genes from the genepool. The sticky questions are whether the drop in crime actually happened when and where his theory suggests, and if enough people were actually killed to make a dent in criminality.

I suspect that Frost is at least partially right–many people who might have had children were executed instead–but there is another factor to consider:

A land where criminals are executed is a land where criminals are already useless or less than useless. They have gone from assets to nuisances (horrible ones, but nuisances nonetheless), to be swatted like flies.

In a land where criminals are useful, we do not call them criminals; we call them heroes. Is Che Guevara a murderer or a freedom fighter? Depends on who you ask. Is the man who crushes enemies, drives off their cattle and hears the lamentations of their women a hero or a butcher? In Mongolia, there are statues of Genghis Khan and he is regarded as the father of Mongolia. Vlad Tepes is a here in Romania.

In a land where marauding tribes are no longer a concern, you have no need for violent tribesmen of your own. In a land where long term saving is technically possible, people who do can get ahead. In these places, the criminality complex is no longer favored, and even mildly CQ people–too mild to get executed–get out-competed by people with lower CQs.

However, I do caution that recent data suggests this trend may have reversed, and criminals may now have more children than non-criminals. I wouldn’t count on anything being eternal.

Looking back over my own thought on the subject over the years, I think this is essentially reversal of sorts. Our legal system is built on the Enlightenment (I think) idea of redeemability–that criminals can be changed; that we punish the individual criminal act, not the “criminality” of the offender. This may not be so in the death penalty or for certain egregiously heinous acts like child rape, but in general, there are principles like “no double jeopardy” and “people who have served their time should be allowed to re-integrate into society and not be punished anymore.” The idea of CQ basically implies that some people should be imprisoned irrespective of whatever crimes they’ve been convicted of, simply because they’re going to commit more crimes.

There’s a conflict here, and it’s easy to see how either view, taken to extremes, could go horribly wrong. Thus it is probably best to maintain a moderate approach to imprisonment, while trying to ensure that society is set up to encourage lawful behavior and not reward criminality.

Your thoughts and reflections are encouraged/appreciated.