> It is possible to have a reputation system where the impact level of
> a good/bad review (in case of escrow service) would be influenced also by
> both sides's reputation (buyer / seller) -> review/feedback of reputable buyer
> would have a bigger impact than a review of seller with no reputation and
> vice versa. So this will be some kind of accumulated review and can be probably
> better used for escrow services. Everyone in this system would have some
> "virual karma" of his/her honesty :-)

To the extent you're describing a decentralized reputation aggregation
system, sure, I think something like this is possible and really
desirable! I think a key point is that it's nonsensical to have an
'absolute' or 'public' reputation, as that is easy to be sybil
attacked. Instead, something like Bitcoin-OTC's trust-graphs, where
inferred trust is a measure between a *pair* of people rather, is
promising. Lets continue talking about this! This is what the future
of markets will consist of.


> BTW, are now any real problems with escrow services (like SilkRoad?) or
> we are trying to solve purely hypothetical problems? :)

Well like most centralized systems, everything runs great and
efficiently until it fails catastrophically. Possible outcomes for
Bitmit (and silkroad for that matter) are: administrator runs away (or
gets "hacked") with all the bitcoins in accounts at the time, or
spammers could start to rake in more money by inflating reputations
for a fee.

Shouldn't the goal be to build infrastructure to replace such brittle
centralized markets with more robust decentralized ones, much as
Bitcoin is infrastructure for replacing brittle centralized payments?

--
Andrew Miller