The unsuitability of presigned transactions for vaults

[jamesob]

@AntoineP

Liana is not an emulation of a vault;

Liana may not explicitly be a vault, but if OP_VAULT were available, Liana would almost certainly be a user of it. There would be no need to dig up keys and roll timelocks with OP_VAULT; the opcode is general enough to cover the Liana case. See this twitter thread for details. Reproduced here in case Twitter goes away:

@AntoineP

In your post you are mixing up the Revault and ephemeral-key approaches with emulating a covenant using a trusted oracle.

I’m not sure where you think I did that mixup, but I’m aware of the difference in approaches. The reason I didn’t dwell much on using an oracle to enforce covenants is because if you’re relying to rely on an oracle to enforce transaction validity, you basically don’t need any script upgrades ever. Of course I think this is not a great security model in general, and especially for people without the resources to run live HSMs.

@AntoineP

But it’s also a fair point to raise: we don’t know of anybody allegedly interested in using vaults who reached the point in development where they have an MVP which emulates the covenant they need using a simple cosigning server.

Maybe no one’s done it because it’s basically a bad trade-off? But again I’m confused, because isn’t this what you guys tried with Revault – presumably because the conceptual notion of vaults is very appealing?