On 03/17/2014 05:16 PM, Robert Williamson wrote:
> Escrow does require trust, with eBay, you have to trust that they're not
> colluding with the seller, the silkroad and stuff like that aren't quite so
> easy, since the parties are anonymous, you have to trust that the middleman
> isn't colluding with the seller (or that they're the same person).
>

This is why I propose randomising the choice of escrow/arbitrator for
the trade in such a way that neither party knows in advance which
arbitrator will be chosen. It is not the only defence in the protocol I
described
(https://github.com/AdamISZ/stegabank/blob/master/protocol.md), but it
is the principal one.

> Also it was trivial for a seller to create numerous buyer accounts and
> flood themselves with positive feedback, paying only a small % fee for each
> rating.

>
> This is easier to do in a decentralised system when there are no fees for
> creating a user and leaving feedback. Create multiple buyers and sellers,
> all giving each other good feedback, spotting things like this in the data
> is difficult, unless graphing techniques for the ratings are used.
>

Note that this attack is principally relevant in the case where there is
no objective verification possible of the trade. Example: localbitcoins
attempts to resolve disputes via photocopied bank details, hardly
reliable. Escrows for a Silk Road type trade can do even less. With ssl
logging this problem is basically completely removed for fiat transfer,
leaving only the collusion problem, discussed above.

Adam