Today I gave a talk at AOL in Virginia about reputation. I also had a chance to talk to a small working group on reputation and to the Architecture Council. The discussions were very good and gave me some food for thought.

I came away more convinced than ever that what we need to build are reputation systems that bring more cues about people and their actions to bear, in a way that allows the user to control the privacy issues, and with as much emergent behavior as possible to avoid overt configuration. Such a system should reward people for participation so that they see real value for the loss of privacy that any effective reputation system requires.

Society extracts a price for bad behavior that is usually based on property--relatively seldom do people lose freedom (or their life). The loss of property is an effective deterrent to bad behavior and conversely the promise of increased property value is an incentive to good behavior. In fact, so much so that property and the rights associated with it are a key focus of the law.

In most online communities there is little of value--even the identifiers are cheap and disposable. Virtual worlds like those in World of Warcraft and Second Life are the exception. Where there is no property, it's difficult to create reciprocity and there is an attendant loss of social good. If we want the online world to become less like the wild west, we need to make it more costly to misbehave. That principle is at the heart of reputation systems.

The sldes from my presentation (PDF) are available online as is the current version of the bibliography my class used in their study of reputation. If you use a Mac, BibDesk is a great tool for viewing and manipulating the bibtex file.