Paul Resnick (has a blog!) is speaking at the conference on reputation systems. He uses eBay as his laboratory, because he can do semi-controlled experiments on things like the effects of reputaiton on sale price.
I've seen variants of his results before. The big picture, takeway for me is that reputation is defined temporally. That is, reputation is a factor of accumulation over time. The other aspect is that there needs to be an entrance cost to create an identity--and that entrance cost defines a maximum negative extent to which reputaiton will drop.
That is, if reptuation is costing me 10% of the cost of the things I sell--and he was able to show that negative reptuation does have a cost--and it costs $100 to join a site, then I have an incentive to hang out in the site until I want to sell $1000 worth of stuff. If it's free to create a new identity on the site, then I'll do so as soon as I have an even vaguely negative reputation. (Obviously, since cost and reputation are interfachangable, you can also start users off with a negative reputation at zero cost.)
I'm considering how this relates to the work at (for example) kuro5hin where the head is actively wrestling with reputation issues...March 29, 2004 04:07 PM | TrackBack | in Microsoft Social Computing