Given that if each person acts in a self-interested way, both could end up seriously injured or dead, it would
appear
(as was the case in our original prisoner’s dilemma scenario) that the rational thing to do would be for each person to behave in the way that
doesn’t
lead to the best-case scenario from his perspective: that is, each should put down his gun.
Once again, things are not so simple. Notice that no matter what White does, Pink is better off if he does not lower his weapon. Consider the case in which White doesn’t lower his weapon. If White doesn’t lower his weapon and Pink does, then Pink will likely get shot. If White doesn’t lower his weapon and Pink doesn’t either, then Pink will likely get shot (since both will likely get shot in that scenario) but may not end up as badly off (because Pink may be the first to shoot and as we’ve seen the first to shoot typically ends up better off than the second). Now consider the case in which White does lower his weapon. If White lowers his weapon and Pink does as well, then Pink survives the scenario (as does White). If White lowers his gun and Pink does not, then Pink survives the scenario and is in a position to have his desire (to get White to act in a more “professional” manner) satisfied. So regardless of what White does, Pink is better off not lowering his gun. The same goes for White. The same exact paradox has resulted—doing the “rational” thing leads to all parties being worse off.
Solving the Prisoner’s Dilemma
There have been a number of approaches to resolving the prisoner’s dilemma. Most are designed (roughly) to show how the term “rational” can be applied to the co-operative course of action (such as neither Ordell nor Louis ratting on the other, or both Mr. Pink and Mr. White lowering their firearms) despite the fact that any participant in such a scenario will be better off not taking that co-operative course of action. Some strategies involve multiple iterations of the scenario. So, for example, persons put in these situations over and over will
eventually co-operate. These solutions work well under controlled circumstances, such as in a game theorist’s laboratory. They also work well in real life situations that tend to be recurrent, such as traffic-jam scenarios. Eventually people learn to cooperate (for example, by letting others merge in) so as to minimize the badness experienced by all. It should be obvious that these approaches to solving the prisoner’s dilemma will not be applicable in a sufficiently wide variety of cases. Specifically, for our purposes, they won’t be applicable to ECDS’s, because such scenarios tend to not be recurrent (one would have to be extremely lucky to survive more than a few of these in one’s lifetime).
A second class of solutions, which I’ll call “group rationality solutions,” appears a bit more promising. These tend to involve denying the claim that it is always rational for an agent not to co-operate in prisoner’s dilemma scenarios. Proponents of this view argue that a rational person would take into consideration the rationality of others in the scenario. Ordell, for example, would think through things as follows: “I’m better off ratting on Louis, no matter what Louis does; but Louis will also realize that he’ll be better off ratting on me, no matter what
I
do. Louis, in turn, will realize that I’m also aware of this. Since we are both rational, we’ll both know to cooperate.” So, in short, once someone in a prisoner’s dilemma scenario takes into consideration the rationality of others, then it is rational to choose the co-operative course of action.
One virtue of this solution is that it shares the main advantage of the previous solution discussed—that it results in cooperation. Here, rather than eventually being a learned behavior, co-operation results from persons’ abilities to reason in advance that co-operation in such circumstances is optimal. Given that the group rationality solution seems to have the virtues of the first solution and is wider in scope (in that it is not limited in application to recurring events), it is clearly the preferable solution to the prisoner’s dilemma.
There’s a drawback to the group rationality solution: it is overly optimistic about human behavior. One could easily imagine Ordell going through the thought process described above and then thinking, “Fuck it! I’m still better off if I screw Louis over, so that’s what I’m gonna do.” Once Ordell believes that Louis will choose co-operation, then it is no longer rational for
him to do so as well. (We could complicate matters further by observing that Ordell is far more likely than Louis—who is as dumb as a brick—to be able to work all of this out rationally at all, and that Ordell would have to work this fact into his reasoning.)
This worry aside, there’s a deeper problem with the group rationality solution. It’s not applicable to the ex-convict’s dilemma. The group rationality solution relies on the fact that the persons in the scenario have common goals. Neither prisoner wants to spend time in jail, so their goals of minimizing jail time dovetail nicely. Persons who find themselves in ECDS’s typically are at odds with one another. Causing serious injury (or death) to the other may often be a desirable state of affairs from such persons’ point of view. This is a huge impediment to group rationality leading to co-operation.
Moreover, persons sometimes put themselves in ex-convict’s dilemma scenarios intentionally. Consider the ECDS in
Kill Bill Volume 1
. A couple of trained assassins, Beatrix Kiddo (Uma Thurman) and Vernita Green (Vivica A. Fox), are fighting to the death and wind up with knives directly in each other’s face. The co-operative solution wouldn’t satisfy either, given that each strongly desires to kill the other. If given the opportunity, each would willingly get back into the situation, if they thought that it would further their ultimate goal. So if there’s a general solution to the prisoner’s dilemma that covers cases of all types, it lies elsewhere.
Tarantino’s Treatment of ECDS’s
I’m not optimistic that a satisfactory general solution to the prisoner’s dilemma exists. It may well be the case that everybody behaving “rationally” ultimately makes things worse for all involved. One might hope that most persons would choose not to behave in obviously self-interested ways most of the time, but experience (at least the experience of someone who lived through the yuppie era, for example) indicates otherwise.
Despite my pessimism regarding a general solution to the prisoner’sdilemma there is still much to be said about getting into and out of ECDS’s. Here Tarantino’s treatment of ECDS’s should be of great help. I’ll make use of a simple taxonomy: ECDS’s in which people die and ECDS’s in which people don’t die.
Given that Tarantino makes frequent use of ECDS’s in his films (we’ll look at six!) and more often than not people come out of them unscathed, one might be tempted to conclude that he’s a kind of optimist about the role reason plays in helping one negotiate one’s way through such precarious situations. This, I’ll argue, is not borne out. The further implication will be that rationality-based solutions to the prisoner’s dilemma are unsuccessful.
Tarantino presents us with two ECDS’s in which things go very badly for those involved and attempts to employ reason fail miserably. Consider the second ECDS from
Reservoir Dogs
. In this scene Mr. White, Joe Cabot (Lawrence Tierney), and Joe’s son Nice Guy Eddie (Chris Penn) are arguing about what went wrong with the heist (and whether Mr. Orange tipped off the cops). Emotions are running high. Eventually, White points his gun at Joe, Joe points his gun at Orange (who is lying on the ground in a pool of blood), and Eddie has his gun pointed at White. Eddie suggests that they put their guns down and settle things with a conversation. White warns Joe that if he shoots Orange, he will be shot. Eddie’s attempt at providing a rational way out of the situation has proven very short-lived: he then screams at White and all three fire their guns simultaneously. The “rational” solution in this case fails, at least in part, because the three participants have different goals. Eddie wants to protect his father, White wants to protect Orange, and Joe wants to exact revenge on Orange. If getting out of the situation were a priority for each, perhaps things would go differently.
Now consider the ECDS from
True Romance
.
116
This one is pretty much the mother of all ex-convict dilemmas. In this scene Clarence Worley (Christian Slater), Alabama Whitman (Patricia Arquette), and Dick Ritchie (Michael Rapaport) are involved in a drug deal with movie producer Lee Donowitz (Saul Rubinek) and his assistant Elliot Blitzer (Bronson Pinchot). Elliot is a police informant, and is wearing a wire. Once the transaction occurs the police bust in and point their guns at the producer and his bodyguards. The bodyguards have machine guns pointed at the police. Matters get complicated when four mobsters
who have been looking for Clarence (he stole the drugs from them earlier in the film) bust in and point their guns at the cops and bodyguards. The only attempt at reason—Lee pleads for everyone to calm down—falls on deaf ears. At this point there are six cops, four mobsters, and two armed guards with guns pointed at one another, all screaming “PUT IT DOWN, PUT YOUR FUCKING GUN DOWN.” It’s not difficult to see how this scenario plays out: everyone shoots at once and almost everyone gets killed.
This situation doesn’t really lend itself to a rational solution in the first place; the people involved aren’t really talking with one another. Given the conflicting interests and goals of the participants, however, a rational solution based on co-operation would be necessarily unattainable even if they were to engage in conversation.
Let’s look at ECDS’s in which persons don’t die. In fact, these ECDS’s are more pertinent than those just discussed. This is because in cases where people die, it follows, necessarily, that the optimal state of affairs—co-operation—doesn’t obtain. Hence, such cases don’t count as resolved ex-convict dilemmas. The category of ECDS’s in which persons don’t die can be broken down into two subcategories: ones in which fate or luck intervenes and ones in which reason plays a role in resolving the situation.
We’ve already discussed the two ECDS’s in which persons don’t die and fate or luck intervenes. The first is the one early in
Reservoir Dogs
involving Mr. White and Mr. Pink. The standoff ends when Mr. Blonde enters the room distracting each and causing them inadvertently to lower their guns. Here there is a co-operative solution, but it is not one that is chosen by either White or Pink—it just happens.
The second is the one in
Kill Bill Volume 1
. The standoff between Beatrix and Vernita ends—temporarily at least—when Vernita’s young daughter, Nikki, comes home from school. Beatrix and Vernita both, almost instinctively, hide their weapons behind their backs in an effort to make it appear to Nikki that nothing bad is going on (despite the fact that Vernita’s living room has been completely destroyed moments earlier during their fight). Again, the result is a co-operative solution (albeit a short-lived one, as Beatrix kills Vernita shortly thereafter). While this solution
is
chosen by the persons in the scenario,
it is not chosen for its own sake; rather, it is chosen because each recognizes that getting out of the situation is what is best for Nikki. It would be disastrous for Nikki to see her mother kill someone or be killed by someone.
Although having good luck or having external events come into an ECDS may get one out of the situation, they certainly are not reliable means of doing so. Let’s turn our attention to ECDS’s in which people don’t die and reason plays a role in resolving the situation.
Consider the ECDS in the final scene of
Pulp Fiction
. Here the situation is potentially ripe for a co-operative rational solution. The persons in the scenario, Pumpkin (Tim Roth), Honey Bunny (Amanda Plummer), Jules Winnfield (Samuel L. Jackson), and Vincent Vega (John Travolta) don’t desire to be in the situation and don’t have any particular animosity toward one another; rather they just end up in the situation when Pumpkin’s and Honey Bunny’s attempt to rob diners of a restaurant is thwarted by Jules’ pulling a gun on Pumpkin. The situation mirrors the second ECDS in
Reservoir Dogs
almost perfectly. Jules has a gun pointed at Pumpkin, Honey Bunny has a gun pointed at Jules, and Vincent has a gun pointed at Honey Bunny. While the result is, for all intents and purposes, the co-operative solution—that is, each eventually puts down his or her gun—it’s not in virtue of group rationality that the co-operative result obtains; rather, it’s in virtue of Jules’s willingness to buy off Pumpkin and Honey Bunny (he gives them $1,500). In an important sense, true co-operation doesn’t actually occur: Pumpkin and Honey Bunny profit more from the situation than do Jules and Vincent.
Perhaps the best example of a co-operative solution to an ECDS occurs in
Kill Bill Volume 2
. Here the persons in the situation actually reason their way out, and neither has to buy the other out. Here’s what happens. Karen Kim (Helen Kim) is an assassin sent to kill Beatrix. Karen shoots her way into Beatrix’s hotel room and they end up with guns pointed at one another. Beatrix convinces Karen to put her gun down and walk away by informing Karen that she (Beatrix) is pregnant. Beatrix convinces Karen that in her pregnant state she doesn’t desire to be in the ECDS nor does she desire to harm Karen.
While the co-operative solution is attained by means of reason and rationality, it, once again, isn’t the sort of rationality employed in the group rationality solution to the prisoner’s
dilemma. Rather, it is the recognition that each person in the ECDS has a greater interest in something else: the well-being of Beatrix’s unborn child.
117
Don’t Get Into One
The most promising solution to the prisoner’s dilemma—the group rationality solution—fails, as it doesn’t apply to a wide enough range of instances of the prisoner’s dilemma. Specifically, it doesn’t apply to a considerable number (if not all) of ECDS’s.