From that pastebin:
"9. Alice's Escrow awakens in 1-5 minutes, and starts periodically checking (via SEPA calls) to see if BOB_SEPA has received the Euros. If so, she automatically releases the Escrow to Jorg's silver account on OT."
This escrow must be able to check the banking transaction via SEPA calls, and I suppose without a formal contract with banks, he won't be able to listen to those calls
Just like a gateway in ripple, this is the point of weakness for all the P2P exchange design: That escrow/gateway must be able to communicate with banks through an authorized channel and that channel is controlled by banks
The core problem is, all of the exchanges today still operate inside a banking framework (the exchange must have a bank account to operate).
This.
And further,point 11 in the pastebin on silver/euro SEPA:
11. If Alice disputes (claiming the SEPA transfer was never received by Bob) then the escrow splits the silver grams between Alice and Jorg.
This is not a fair resolution method, is it? If the escrow has no way of knowing, then Alice is always incentivised to dispute.
Unless the bank wire is publically verifiable as well as irreversible, the fiat part of these exchange models always seems to be broken.I believe dansmith has the essential elements of the solution in the discussion we've been having here:
https://bitcointalk.org/index.php?topic=173220.0I'll give you the precis as that's a lot of reading. In case of dispute, the BUYER of fiat (not the seller) can be requested to run an open source plugin that reroutes their banking SSL session through a proxy (after they've logged in), the proxy can be the escrow agent or another user of the P2P software, the encrypted data is dumped and using the master key for the session, the data can be read such that it's verified that the fiat buyer did/did not receive the fiat funds (the fiat buyer only needs to show their bank statement for the period). Thus it can be made public knowledge whether the fiat was transferred or not.
Agree fully with oakpacific in that digital signatures, which would be trivial for banks to implement, on bank statements would immediately solve the problem. However banks are only using cryptography for the privacy functionality - they are not interested in non-repudiation of messages, even though it would be helpful to customers for a lot more than Bitcoin, it's not useful to them so they don't care. But I see that discussion as a tangent since we are not going to change bankers minds, we just have to get rid of their archaic system.