Hi Mike, I haven't gone through all the references but it seems we have two very different ideologies, from what's at the beginning of 'Contracts':
"Contracts ... allow you to solve common problems in a way that minimizes trust. Minimal trust often makes things more convenient by allowing human judgements to be taken out of the loop, thus allowing complete automation"
It's a shame that I can't go through all the details in your proposal for now, but from parts of your London talk and the wiki pages, I feel that they try to secure people's wallet without releasing their identities, by encouraging a system with minimal trust. Assuming it's technically feasible and bulletproof, I wonder what's the overhead will be.
The credit network in my mind seem to have conflicts with privacy, as it relies on the transaction history with real identities (at least accessible by the whole network that the person participates in) kept in the credit network. It seems to me the cheapest way that I can think of.
BTW here's an example (not as a proof that it's cheapest) here are some technical details to handle theft with the system:
If a wallet is stolen from A, the thief have 2 ways to do with it:
1. keep it
2. transfer the bitcoins to a target address T (for simplicity say only one target address)
1. If the thief keeps it forever, so it's just like your physical wallet was stolen on the street, there's nothing you can do other than tracking the thief through police, which for bitcoins is network security experts who may track the stealing. Any techniques that try to make the recovery easier can be complementary.
2. this is most likely. There are two cases:
a) the target address T is in the credit network, just need to ask that person B, who may or may not the thief, to give the money back. If B gives it back, B's credit doesn't change; if not, then some punishment, like can't transfer from that address again to any address in the network unless payback is done.
b) the target address T is not in the credit network, try to contact that address or any following target addresses (can be a publicly accessible bulletin) to ask it back. Treat the unresponsive addresses like in a)