I don't remember this being discussed at the Reputation Symposium (at MIT in April) or on this list before, but has anyone studied the effect of a transaction fee on reduction of fraud?

For example, Friendster is also called Fraudster because many of the members are fake e.g. Jesus Christ. If there were, and probably will be, a member fee, most likely Jesus will disappear*. Transaction-based reputation system can exploit the relationship between transaction fee and fraudulent behavior even more powerfully -- the transaction fee itself discourages fraud. For instance, on eBay sellers planning to commit fraudulent activities first develop a positive reputation. Once they have some credibility they commit the fraudulent act (e.g. no delivery of product). However, the the transaction fee for creating the positive reputation (and also selling the fraudulent item) is high, it can serve as a deterrent. Even better, the marketplace (in this case eBay) could make a argument for the transaction fee being higher (in either % or actual value) for certain categories of products by claiming that the higher fee reduces fraud. It's kind of like castor oil, even thought it tastes terrible, it's good for you.

Has anyone on this list explored this relationship between reputation and fees? My organization, the SPM eGroup and its sibling professional guilds, are going to be charging a member fee soon. It would be terrific if I could point to some academic or other studies that "prove" fees actually raise the quality of the interactions.

Thanks.

Cynthia

*Just for the record, I don't think Friendster has a successful business model if it is a subscription fee. The person who is most needed on the site, the matchmaker, has no motivation to join. And the matchmaking is only convincing if the matchmaker knows both of the people who are seeking a date. Is it really necessary to have a system that tells you who you already know? And the reputation system rewards not matchmaking, but the gathering of "friends" whether real or fake. For more on this go to: http://typaldos.com/word.documents/profguilds/nodes/ .