On my last Afghanistan trip I was re-reading Axelrod’s The Evolution of Cooperation and it got me thinking again about the preconditions for co-operation, and how we design for this outcome. While simplistic (in that it’s derived from analysing a particular, one dimensional, game theory model), the lessons he derives should inform our design patterns for cooperation and community… Basically we need to think how to design and enable the following preconditions for the development of cooperation… (with thanks to Bokardo for kicking this off)
1. We must know there’s a likelihood of meeting in the future: there are consequences for our actions
If people don’t think they’ll meet again in the future, there are no repercussions for not cooperating. Threats of not cooperating are of little use. People will act selfishly if there is no future to the relationship. Therefore, the knowledge of the existence of future meetings changes our behavior because we feel some level of impending accountability for our actions. They have consequence.
2. We can identify each other – knowing who did what in the past
Identity is critical important for cooperation because it allows us to know who we’re dealing with. If people can’t identify who they’re dealing with, then they can’t hold that person accountable. This doesn’t mean that we have to know everything about the person, like their social security number – or even real name – it means that they are identified as a person to the system they’re in and the people they’re dealing with.
Question: how psuedonomous can you be for this to still apply? Depends on the trust in the persistence of the identity behind the nickname.
Question: how much are online tools like twitter and facebook changing our (western) threshold for identity?
3. We have access to a record of past behavior – who did what in the past, aka reputation systems
At the simplest level we assume that the best way to judge future behavior is by looking at past behavior. Reputation systems extend this to include the assessements of likely future behaviour by people whose judgement we trust – a proxy for knowing what they really did and the context in which they did it. Having a positive record of behavior leads to cooperation. E.g. the eBay’s seller rating (forget about how it can be gamed). Sellers accumulate status over time as they do business on the site. “Sellers who have a rich transaction history with a high percentage of positive transactions are much more likely to be successful than those with no history.” Rich reputation systems let us account for context when assessing the other party’s behaviour.
Question: if we make a reputation system too complex or abstract do we include too much context and thus stop remembering what they did clearly enough and we no longer trust them.
Question: how does this link to Dennis Mileti’s conception of “milling” (fact-checking and consultation) as a critical part of crisis decision making?
