Trust Among Thieves

(Photo: .v1ctor Casale.)

A new paper by Federico Varese and Paolo Campana looks at police-intercepted phone data on the Italian and Russian mafia to study how criminals cooperate. In  “Cooperation in Criminal Organizations: Kinship and Violence as Credible Commitments” (abstract; PDF), they find that sharing information on violent acts increased cooperation. Varese writes to us in an email: “The idea is that criminals might trust each other more after they have shared compromising information on themselves and especially have used violence together, an insight from Thomas Schelling that we test and found to be correct.”

The abstract:

The paper argues that kinship ties and sharing information on violent acts can be interpreted as forms of ‘hostage-taking’ likely to increase cooperation among co-offenders. The paper tests this hypothesis among members of two criminal groups, a Camorra clan based just outside Naples, and a Russian Mafia group that moved to Romein the mid-1990s. The data consist of the transcripts of phone intercepts conducted on both groups by the Italian police over several months. After turning the data into a series of network matrices, we use Multivariate Quadratic Assignment Procedure to test the hypothesis. We conclude that the likelihood of cooperation is higher among members who have shared information about violent acts. Violence has a stronger effect than kinship in predicting tie formation and thus cooperation. When non-kinship-based mechanisms fostering cooperation exist, criminal groups are likely to resort to them.

Shane L

Fascinating, makes sense.

I had thought something like this might be true also for some of the ways societies used to punish crime or deviance. Consider the stoning of criminals by crowds: a very public act of violence in which the entire community became complicit in the violent suppression of non-conformity. The medieval stocks might be similar, as well as the public executions that members of the public would flock to as entertainment in the early modern period. Bringing the public into the punishment perhaps made the public complicit, giving a perception of legitimacy to the punishment.

A dark example might be the Cultural Revolution in China, when people were forced to denounce their own family members or friends. By dragging everyone into the violence, the radicals were making everyone complicit, which perhaps weakened the public's ability to resist or take a moral stance against the violence.



Really fascinating,

May I be the first to beg that you please make this a freakonomics podcast episode? Brings a whole new level of fascination to "prisoners dilemma" situations.


These results can be explained in the following game theoretic model example,which I found in a TTC video course :

One way to make a promise credible is to allow another player to retaliate if you break the promise-effectively, to encourage blackmail.
A. Three candidates, Dennis, Rebecca, and Indira, are running in a mayoral election.
1. Polls show Dennis with 40% of the vote, Rebecca with 45% and Indira with 15%.
2. Indira likes Rebecca a little more than Dennis, but she cares much more about having influence over zoning-a position Dennis could offer her if he wins with her support.

B. Because Dennis has no reason to keep his promise after Indira's move, he needs to find a way to make his promise credible.
C. Dennis could give Indira information that could cause a scandal if a she revealed it while he was in office. (Rolling back the tree shows that ) Indira will reveal the scandal only if Dennis refuses her the planning position, (In equilibrium,) Dennis wins and Indira gets the position.
D. Allowing himself to be blackmailed actually won Dennis the election!



This seems intuitive to me. It's mutually assured destruction. I would imagine that it's similarly easier to trust a partner in an extra-marital affair, if the other person is also married. If you both have something to lose, neither wants to be the one to initiate the 'exposure' battle, meaning you both have motivation to trust each other.

Additionally, I wonder if the violence is actually causal, or correlative. Is it possible that from the point that you decide you trust somebody enough to give out compromising information, or engage in compromising activities together, everything from that point forward is going to reflect greater trust than the time period prior to being able to trust them that much? Put it on a timeline, and the trust gains are probably going to be linear, with the actual violence/information sharing just being a landmark demonstrating the established trust.

Food for thought...


El Conquistador

Not a great analogy - I heven't had lunch - but it made me think about the old firing squad system. Everybody shoots at the same time (less chances to renege) and each person have the opportunity to deal with their own feelings of guilty (if any) afterwards by telling himself that maybe my shot was not the deadly one.

I also wonder if there isn't an intimidation mechanism in place to. Do we know if the sharing is reciprocal or one way? Maybe I tell somebody else what I have done so he/she knows how far I am willing to go if they do not cooperate?