CoinPool: efficient off-chain payment pools for Bitcoin

[BlackHatCoiner]

I stumbled across this paper, which was released recently, and due to my lack of free time I'd like to read the techies' opinions: coinpool.dev.

Abstract

CoinPool is a new multi-party construction to improve Bitcoin onboarding and transactional scaling by orders of magnitude. CoinPool allows many users to share a UTXO and make instant off-chain transfers inside the UTXO while allowing withdrawals at any time without permission from other users. In-pool accounts can be used for advanced protocols (e.g., payment channels). Connecting them to other CoinPool instances, or even to the Lightning Network, makes in-pool funds highly liquid. CoinPool construction relies on SIGHASH_GROUP, SIGHASH_ANYPREVOUT and OP_MERKLESUB changes to Bitcoin. It also assumes a high degree of interactivity between pool participants.

The idea is fairly simple, the users lock funds collectively using a single UTXO. They can transact with each other or even with other pools, same as with Lightning and routing. The system provides less privacy, but more efficiency. It requires further softforks and it is currently in a very early state.

[1714969180]

1714969180

[1714969180]

1714969180

[1714969180]

1714969180

[1714969180]

1714969180

[BlackHatCoiner]

Other user could avoid inconvenience if other party isn't being cooperative (which is concern in LN).

Well, if one decides to leave the pool either cooperatively or not, the participants have to move to another UTXO. There's a little debate of whether should the participants automatically move to another UTXO with their respective balances or just close the pool entirely (such as with force-close channel in LN) and let them remake it. The former improves scaling, but there's a "but" I don't yet understand.

[garlonicon]

There's a little debate of whether should the participants automatically move to another UTXO with their respective balances or just close the pool entirely (such as with force-close channel in LN) and let them remake it.

The whole innovation is in moving to another UTXO. You can "just close the pool entirely" right here and right now, just make some N-of-N multisig by using one Taproot address and sign some closing transaction that will move coins to each person. You can prepare two transactions: one forming the channel and another for closing it. You can start by signing the closing transaction, then people can safely sign the opening transaction. It is resistant to cheating in the same way as LN is.

So, if closing the pool is needed, it can be done without any consensus changes.

[odolvlobo]

Is it accurate to say that the pool is similar to a Lightning channel, but held by more than two people?

[garlonicon]

Yes, it is just LN channel for N people, but with some additional features. For now, one Taproot address can represent N-of-N multisig, but more consensus changes are needed to make it more convenient and allow more things, for example detaching someone without closing that channel for all people, and without making a lot of off-chain transactions to predict each case.

[vjudeu]

There's a little debate of whether should the participants automatically move to another UTXO with their respective balances or just close the pool entirely (such as with force-close channel in LN) and let them remake it.

CoinPool cannot "close the pool entirely". Recently, Paul Sztorc wrote about it: https://www.truthcoin.info/blog/lightning-limitations/

Introducing "automatically move to another UTXO" will solve a lot of Paul's arguments from the link above.