Wednesday, 26 June 2013

Games, Interactive Behaviour, and Responsibility

Lately it's occurred to me that there's something called "Blaming the victim" by which the victim of some conflict is considered to be absolved of responsibility for the outcome of the conflict by dint of getting the worst of it. As in "Don't blame the victim!" After all, it seems somewhat awful to consider that the victims of rape, murder, and so on be blamed for failing to protect themselves adequately by wearing suitable clothing, avoiding bad parts of town, and not otherwise encouraging rapists, homicidal assholes, and so on. So why would anyone blame the victims?

In part the problem stems from an asymmetry between the attackers and their victims, and in part from how the consequences of those conflicts play out in legal systems by which the victim is further subjected to unwanted intrusion, and then the attacker is made victim as well. Not that I'm expressing sympathy for rapists and so on, but there's a certain knowledge gap here between our responsibility to protect ourselves, and our responsibility to protect others. And, naturally, I would like to work around the emotionally charged language of 'victim-blaming' or 'slut-shaming' or whatever catch-phrases are being used to drawn the discussion into a morass of competing interests rather than an attempt to identify solutions.

Firstly, I'd like to use games as models of conflict. In a game there is no attacker or victim, but nonetheless the players can come into conflict, and co-ordinate, and do so at each other's expense, do so in irrational ways, and take responsibility for their actions even when those actions do not yield the more desired outcome. Take the Hawk-Dove game as an example of a situation of conflict where players are motivated to maximize their own profit at the expense of their opponent: the Hawk option means that the player takes $1 from a pot while the Dove option means that the player gives their opponent $2. Where the players choose different strategies, one player gets all the money, $3, and the other gets nothing. Where a player is convinced that playing the Dove strategy, playing nice, will encourage an opponent to reciprocate, they will be victimized by a player mercilessly applying the Hawk strategy.

On the other hand, if a player avoids being victimized by playing the Dove strategy, they will not do as well when they shift to the Hawk strategy, or at least not as well as they would have if their opponent also played Dove so that both players get $2. It's from this point that many fallacies of cooperation stem, the notion that somehow or other it's in the interest of the players to co-operate and play Dove when players who play Dove lay themselves open to get nothing; at least players that play Hawk will get something. If both players applying Dove is co-operation, and at the cost of not maximizing their own putative interest in the money measuring the outcome of the game, and where only one player applying Dove is conflict, then what is it when both players apply Hawk?

It's interesting that a sense of responsibility lies outside of this model's power to explicate. To get a sense of responsibility, and dessert, we need to extend the Hawk-Dove game to be the Indefinitely Iterated Hawk-Dove game, an indefinite series of Hawk-Dove games where players attempt to influence the actions of their opponent in future game, and without perfect knowledge of those games (available in both Hawk-Dove and the family of Iterated-x-times Hawk-Dove games). Players are 'responsible' in the sense that they can be punished for actions taken in previous games; a player that plays Hawk against Dove may face a Hawk in return in the next round, and in the case of some mixed strategies such as GRIM face an opponent unwilling to play anything but Hawk thereon in. This strategy is counter-productive where the goal is to have the great amount of money possible when the indefinitely iterated rounds cease iterating .

Where we further expand the game laterally, among the sets or populations of players, we find populations of absolutely co-operative players will outperform populations of players in absolute conflict, and players with higher mixes of co-operation to conflict will outperform those with lower mixes where an Indefinitely Iterated Hawk-Dove game ends. Part of this is a result of players not knowing when the game ends, and so being motivated not to play Hawk on the last game, and instead to continue to encourage co-operation by playing Dove where an opponent isn't playing Hawk. However...

In the case of the Hawk-Dove game the players are motivated to subvene the matrix indicating their interests in the outcomes of individual games, or opportunities for conflict, to the matrix indicating their interest in the outcome of the over-arching game. Which isn't to say that players can't play sub-optimally, and do stupid stuff that adversely affects the outcome of their game, or that they can't get lucky facing a stupid opponent, or even that they cannot act randomly and thereby be immune to the threat of retaliation in the next game. It is to say that the players share responsibility in proportion to their ability to influence the outcome of the game. A player that can influence, but not be influenced (either because they are acting randomly, or according to some program of strategy that takes no account of interacting player choices), takes no responsibility for their actions in the game despite perhaps being culpable for choosing to act randomly or according to a program: Just following orders does excuse one from acts of murder carried out under orders, but abrogation of such responsibility is conceivably worse.

So what are the parameters defining a victim's responsibilities? The parameter suggested by the Hawk-Dove game is the one that people (bigots really) tend towards: Don't do things that will get you victimized: Don't dress sexy, don't go to the wrong part of town, don't play Dove. However the parameters further suggested by the expanded model, the Indefinitely Iterated Hawk-Dove game, suggest that responsibility is tied to a player's ability to influence an opponent. In other words, if a player's action can influence another player not to play Hawk, then they responsible when their opponent players Hawk on them. But if a player could not influence their opponent, encouraging or discouraging them from playing Hawk, then such players lack responsibility for the outcome.

This naturally leads to the sticky question of how such interactivity is measured in the complicated analog life beyond the discrete quanta of a game theoretic model. One could build more complicated into the model, but I think it's more useful to deduce principles and apply those to specific cases, or at least more realistic given the constraints. Could a soldier have prevented her death by shooting first, or shooting better? Yes, that's why soldiers killed in war aren't murdered: the whole point is that the conflict has broken down to the level of killing. Could a soldier have prevented her death by not showing up in the first place: Quite possibly, but soldiers accept the consequence, the responsibility, that they could die in the line of duty whether they properly consider that fact or not. Could a soldier's commander prevented her rape by not deploying her into a combat situation, or in close quarters with a collection of rapists? Could she have avoided it by not fraternizing with the rapists in a way that made raping her a possibility? Again, perhaps, but consider how the game changes.

In these cases the game is not the soldier in conflict with her killers and rapers, it is in conflict with her military and her commanders. In the case of preventing her death by not showing up, the soldier takes that risk by putting her life at the disposal of her commanders, and gives them the opportunity to send her to her death. In the case of preventing her rape by avoiding situations in which she is vulnerable outside of deployment, such as on base or on leave, we politely ignore all of the situations in which she would putatively not be vulnerable to rape. People, after all, get raped by family members, acquaintances, and even in the good part of town. They get raped behind locked doors, and often by those charged with preventing that assault. Husbands rape their wives, parents rape their children, and commanders rape their subordinates. The threat of rape persists in a fashion that absolves its victims of responsibility for not sufficiently protecting themselves; that's the thing about rape, that one cannot negotiate with one's rapists.

One could securely tie oneself up in the public square, or perhaps a back alley in a bad part of town, to be easily raped by any passersby, and the blame for those assaults would lie wholly with the passersby. Doing so is not equivalent to choosing the Dove in a game of Hawk-Dove, because rape is not the victim's choice, and involuntary by definition. Such a crime creates a victim because they are denied any agency or choice in the matter. Hence, as a society, we fail by considering how to avoid being raped rather than how to avoid raping. The rapist bears the sole responsibility for the crime because it is not a game, an interaction, or a negotiation between rapist and victim, but between the rapist and themselves, when that rapist decides to engage their capacity to harm with their desire to inflict harm on another.

The conclusion that I would like to summarize is that while responsibility can be considered as a conflict between attacker and victim, it is actually a conflict within the attacker. Likewise the responsibility of killing in war is a conflict between a soldier and her own army, rather than between opposing armies or soldiers; this provides a basis for war-crimes being actions that are extraneous to achieving the objectives of a war. In each case, the victim is not the person coming off worse, it's the person who could not have come off better. Likewise, the responsible party, the one choosing to opt for punishment, is the person who chose that someone else would be denied a choice.

No comments:

Post a Comment

Looking forward to hearing from you!