Gift Exchange

Interestingly, I recognized two of the authors of the articles. I've read "The Righteous Mind: Why Good People are Divided by Politics and Religion", a book by Jonathan Haidt, and have written an essay on some other article by David Brooks for a different class. Perhaps sadly, I don't think I have much to say about gift exchange.

I wonder, however, if the following is in any way related to gift exchange. A couple of years ago, I read "The Evolution of Cooperation", a book by political scientist Robert Axelrod. In the book, he explores the Prisoner's Dilemma, but from a new (to me) perspective. The book is a search for the most robust strategy in the repeated game of Prisoner's Dilemma. There're many models described in the book: always cooperate; always defer; randomly defer, etc. Surprisingly, the most robust one (I'll explain later what that entails) is the tit-for-tat strategy. Its definition is simple: cooperate on the first move, then keep retaliating to whatever the opponent's last action was. In short, the strategy gives cooperation a try at the risk of being exploited on the first move, and then simply reproduces the opponent's preceding move.

The robustness, first of all, means that the tit-for-tat strategy is fairly successful when playing 1-1 against any type of player. But more importantly, perhaps, is how it behaves in a group setting. With N players following various groups of strategies and playing with each other iteratively, tit-for-tat doesn't go extinct unless the proportion of the population it constitutes is relatively small. This isn't true for all strategies: consider, for example, always-cooperate vs always-defer approaches. In a group setting, a single always-defer player will definitively exploit always-cooperate players, even if those are outnumber that player by far. An interactive visualization of the iterative Prisoner's Dilemma and the findings of the book can be found here.

Now, the way I would like to tie this to gift exchange is the following. There's a strategy very similar to tit-for-tat that, despite its similarity, is not nearly as robust. The strategy is as follows: defer on the first move, then continue retaliating for the opponent's preceding move. Essentially, instead of taking the risk of being exploited on the first move, it chooses to exploit the opponent itself. The strategy can be thought of as tit-for-tat with a shifted phase (as in a periodic function).

The latter form of tit-for-tat not being as successful as the former form suggests, I would assume, that an attempt to establish gift exchange is a more robust approach even in the game theoretic setting. No wonder we all give gift exchange a try in real life.




Comments

  1. It's good that you brought in your prior reading, but I wish you considered the perspective slightly differently - from the perspective of somebody designing the game (a manager) rather than somebody playing the game (an employee who reports to the manager). The issue here isn't how to play the repeated Prisoner's Dilemma well. The issue is how to design a game where cooperation is the likely outcome.

    Also, the Prisoner's dilemma may not be the best example in which to consider willingness to share or fairness. So I encourage you to either, try to turn the Haidt article I had you read into a game in two stages - first stage is pull the strings, second stage is divide up the marbles - and then see how that game is like or is different from the Prisoner's dilemma. Then the next question is whether people "imagine" they are in a repeated game even when the actual situation is one-shot and not repeated. Try to think that through.

    ReplyDelete
    Replies
    1. Considering the manager's perspective, I wouldn't immediately know how to design a game that is conducive to cooperation. Now that I think about it, it indeed is a hard problem. A tangential approach that comes to mind -- managers can take advantage of hiring decisions that funnel out people who are not cooperative. The readings suggest that people naturally default to cooperation; maybe it's most important for a manager to pre-select people who are likely to cooperate. What the Haidt's article seems to suggest is to encourage "collaborative effort", not just cooperation. Gift exchange is more likely to take place if it is evident that there would have been no reward if not for equal collaboration of all parties. Maybe that means that it's effective for a manager to break teams up in smaller units that are each assigned more work than can be done by any member individually.

      I assume you are implying that we, generally speaking, don't realize we're playing a repeated game. I completely agree with that.

      Delete

Post a Comment

Popular Posts