Paper #:[4.3.2 Trust] #3
Title: "Reputation Systems: Facilitating Trust in Internet Interactions" by Paul
Resnick, Richard Zeckhauser, Eric Friedman, Ko Kuwabara


PROBLEM
The internet is a great medium to purchase products from individual sellers.
However, with this opportunity comes risks because there is no guarantee that the
seller can trust that the buyer will purchase the item that they may have won on
an auction or that the buyer will receive the item that he paid for in the
expected condition.  In order to alleviate the problem, "reputations" have been
used so that there are some consequences to being a poor citizen or rewards for
being an outstanding citizen in this community of shopping.  The opinions given by
users are not only important for making decisions on who to buy from but also with
making decisions on what to buy, which is why websites such as cnet.com are
useful.  In the end the success of a reputation system needs the following
characteristics:  future interaction is expected, information is easily available,
and other people care about previous feedback.

NEW IDEAS AND STRENGTHS
*The paper did a good job of determining the three characteristics needed, which
are long-lived entities, captured and distributed feedback, and old feedback aids
buying decisions.
*The reference to the economist was helpful with conveying the message that good
sellers need to be rewarded in order to prevent a "Market for Lemons".  This
market would result if all of the quality sellers left due to not being rewarded
for putting in the extra effort of being a good citizen.
*The idea of eBayers negotiating in order to avoid negative feedback is subtle,
but can drastically change the outcome of feedback.
*Blackmail and collaboration of people can also skew the true feedback results.
*The paper does not attack the current system because it seems to be working
despite the possibilities of people taking advantage of it.

WEAKNESSES AND EXTENSIONS
*The paper lists that eBay displays the net feedback and expresses the need to
know the amount of the transaction in order to know how much to weigh the
feedback.  However, eBay allows users to see recent transaction items as well as
the price of those items.  Therefore, with eBay this is not a problem.
*It would have been nice if the paper had done some more research on the effects
of users blackmailing each other in order to receive undeserved positive feedback.
While mentioning that this is a problem, it did not attempt to give any solutions.
*A possible extension would be to understand why the reputation system works
overall despite all of the possibilities of manipulating it.