A TRUST SYSTEM FOR ONLINE AUCTIONS
----------------------------------

* online auction systems such as eBay (www.ebay.com) are failing to combat
fraud (see http://law.richmond.edu/jolt//v7i4/note1.html).

* eBay allows users to rate each other, after they complete a transaction.
Problems with their current system:
        - it is too easy to create fake identities.  This means conmen can:
                * "respawn" under fresh identities
                * create puppet identities whose sole purpose is to give
                positive feedback (recommendations) to inflate the conman's
                reputation.
        - negative feedback (complaints) aren't viewed in context.  eg:
        negative reports are often used to punish genuine complaints.

* therefore, eBay needs a better metric.

* my proposed solution:
        - verify identities with a transitive trust metric such as
        advogato (http://advogato.org/trust-metric.html) or PageRank
        (http://citeseer.nj.nec.com/page98pagerank.html)
                This is a good choice because it limits the amount of
        false identities that can be created.
        - find and rate complaints (after verifying the identities with
        the previous trust metric) with a trust metric like that in
        www.pgrid.org/Papers/CIKM2001.pdf.
                This is a good choice because it is good at dealing
        with complaints (unlike transitive trust metrics XXXX)


XXXX

It looks like transitive trust metrics are actually capable of dealing
with complaints (and, the first results look good, although more
research is needed IMHO).  Just as I figured this out myself (grrr),
I found this really cool paper:

	http://citeseer.nj.nec.com/551866.html