Idle ramblings of a wandering mind...

My last post on the group reputation system at Prosper.com focussed on the effectiveness of the system. Since then I have continued my research on the same and will likely deploy a survey to group leaders in next couple of months to test my hypotheses.

There has been a lot of research recently on online reputation systems like the ones used at eBay and Amazon. There are some important differences between the "individual" reputation systems at eBay etc and the group reputation system at Prosper.

• Seller Only Reputation: First of all, Prosper's reputation system is essentially a seller reputation system. Buyers do not need to prove their reputation as they bear all the risk (since they pay money upfront). This is a key difference compared to eBay where the reputation system includes both buyer as well as seller transactions.
• Credit Scores: While the reputation system is the ONLY window to a seller's reliability in eBay; in Propser, buyers have access to the credit history of sellers - which gives them all sorts of information to judge seller's reliability. So the reputation system here is of only incremental importance over and above the information available in the credit history. The group reputation system is essentially as secondary reputation system with the credit score being the primary reputation system.
• No Repeat Transactions: Unlike eBay and Amazon, sellers at Prosper rarely conduct repeat transactions (second, third loan). Since the sellers are unlikely to come back to the market, their incentive for keeping a great reputation in the market is quite low. Hence, there needs to be factors beyond the marketplace to force them into complying with buyer's expectations.
• Group's Core Connection: Prosper expects that if sellers are part of a tightly knit group then there would be peer pressure from the group for the seller to behave nicely. However, this would only be true if the group has a strong core connection (whether offline or online) which has some value to the seller. Currently most groups in Prosper do not have a core connection - they are just an online group of people with similar needs. This is unlikely to lead to good behavior by sellers and we have seen the evidence of same in the repayment data.
• Filtering by Groups: Prosper also expects the group to be a filtering mechanism where group leaders check the listings and only let trustworthy borrowers come to the market. However, many groups do not even enforce the mandatory review of the listing by the group leaders - so even this expectation is futile.
• Conclusions: Based on the above - we come to the following conclusion:
• Groups that share a strong core connection (offline or online) would perform better
• Groups that have a strong process for filtering listings would perform better

We have refined the conclusions above to generate research hypotheses and are testing these hypotheses using forthcoming survey data and data collected from Prosper. I will post the results once the analysis is done in next couple of months.

Links

1. Thanks are due to Prosperous PF Strategies for leaving a thoughtful comment. Link to the comment.
2. Rate Ladder is an excellent resource for Prosper participants. It is a little cluttered with ads - but has great information.

Labels: Prosper, Research