No sybil resistance?
No zerolink?
I'll sit this one out. I applaud the experimentation but hope everyone that tries it is aware of the implications.
Discussion
> I'll sit this one out
Please don't, please make it better!
What can be made better? Do you have anything in mind to address the concerns above? Or would it require a totally new design?
Joinmarket does sybil resistance via fidelity bonds, that is, you only coinjoin with people who first prove they locked up a bunch of bitcoin in a timelocked address. Sybils would have to spend a lot of money to do that, thus bankrupting themselves, so you're only left with honest people. I think emessbee can do the same thing, so yay.
As for zerolink, its purpose is to prevent linking the equal amount outputs to their inputs and to prevent linking people's ip addresses to their inputs. I think I've managed to do the first part in a different way, so I don't think following the zerolink protocol would improve anything on that front. The other part seems achievable by just writing a client that runs over tor and creates a new identity for all three rounds. Thoughts?
If you do fidelity bonds for sybil resistance, there's not much that distinguishes it from JM, except for a nostr relay orderbook instead of p2p gossiping. That does protect IPs (if you dont use tor) but I would still say it's just a marginal improvement.
The point of zerolink is not protecting IPs, it's to break all deterministic links using equal output coinjoins. I actually see a future in which JM or your proposal could implement it. It could be something like an "advanced coinjoin market pool" that would require a previous tx0 type transaction that guarantees the utxo to mix is compliant with the pool amount.
