This week’s newsletter relays a security announcement for LN implementations, describes a non-interactive coinjoin proposal, and notes a few changes in popular Bitcoin infrastructure projects.
Upgrade LN implementations: one or more issues that are scheduled to be disclosed at the end of this month affect older LN implementations. If you are using any of the following software versions, upgrading to a more recent version is strongly recommended:
- C-Lightning < 0.7.1
- LND < 0.7
- Eclair <= 0.3
SNICKER proposed: Adam Gibson posted to the Bitcoin-Dev mailing list with a proposal for Simple Non-Interactive Coinjoin with Keys for Encryption Reused (SNICKER), a method for allowing wallets to create coinjoins non-interactively in two steps, a proposer step and a receiver step. In the proposer step, Alice’s wallet uses the block chain and UTXO set to find UTXOs for which she knows (or can infer) the owner’s public key. She selects one of those UTXOs whose value is less than the amount controlled by her wallet and creates a proposed coinjoin between that UTXO and her wallet’s UTXOs, producing three outputs:
Coinjoin output to Alice
Coinjoin output to the owner of the selected UTXO (Bob). Both coinjoin outputs pay the same amount to make them indistinguishable to third parties looking at the block chain
Change output to Alice returning any of her money in excess of the coinjoin amount
Addresses that already have transaction history can’t be reused or the privacy benefits of coinjoin are lost, so Alice generates new unique addresses for her two outputs. However, Bob has no way in this non-interactive protocol to tell Alice what addresses to use for his output. Instead, Alice can use Bob’s public key and Elliptic Curve Diffie-Hellman (ECDH) to derive a shared secret that Bob will also be able to derive from Alice’s public key. With the shared secret and Bob’s public key, Alice is able to create a new public key that only Bob can sign for. That new public key is used to create the new address for Bob’s coinjoin output. No one besides Alice and Bob can tell the difference between the addresses, ensuring privacy for the coinjoin.
With information about the inputs and the outputs, Alice creates a BIP174 Partially Signed Bitcoin Transaction (PSBT) containing the signatures for her UTXO or UTXOs. She can then upload this PSBT to a public server (and she can encrypt it so that only Bob can decrypt it). This completes the proposer step in SNICKER.
If Bob participates in the scheme, his wallet can begin the receiver step by periodically checking the server to see if anyone like Alice has sent him a proposed PSBT. If so, Bob can evaluate the PSBT to ensure it’s correct, add his signature for his UTXO to finalize it, and broadcast the transaction to complete the coinjoin.
The key advantage of this proposal is that no interaction is required between Alice and Bob. They each perform their steps independently and aren’t limited by having each other as potential partners. Alice can create as many proposals as she wants at no cost (except server storage space) and Bob can receive multiple proposals from different people via a variety of servers, selecting whichever proposal he prefers (or none at all). Either party can also spend their UTXO normally at any time, automatically invalidating any pending proposals without any harm done. The PSBTs can be exchanged using any medium that doesn’t require users to identify themselves, such as a via a simple FTP server over Tor, making it easy for anyone to host a SNICKER exchange server.
The main downside of the proposal is that it requires the proposer (Alice) know the public key of the receiver (Bob). Almost all transaction outputs today pay addresses that don’t directly include public keys, although that may change if the proposed taproot soft fork is activated and becomes widely adopted. In the meantime, the SNICKER proposal suggests scanning for reused addresses where public keys have been revealed via the block chain, or by using a public key from the input of a transaction that creates a UTXO. For a more detailed overview of the proposal, see Gibson’s blog post.
Notable code and documentation changes
C-Lightning #3002 makes several small changes to comply with recent updates to the LN specification, including the BOLT4 update to remove the
final_expiry_too_soonerror message as described in last week’s newsletter.
Eclair #899 implements extended queries as proposed in BOLTs #557, allowing an LN node to request only gossip updates that arrived after a certain time or which have a different checksum than those the node has already seen.
Eclair #954 adds a sync whitelist. If empty, the node will sync its gossip store with any peer. If not empty, the node will only sync with the specified peers.