lightning-dev

CPFP Carve-Out for Fee-Prediction Issues in Contracting Applications (eg Lightning)

CPFP Carve-Out for Fee-Prediction Issues in Contracting Applications (eg Lightning)

Original Postby Matt Corallo

Posted on: January 8, 2019 14:46 UTC

In this email conversation between Rusty Russell and Matt Corallo, they discuss an alternative proposal for Bitcoin's Replace-By-Fee (RBF) protocol.

Russell argues that the proposal does not guarantee the ability to RBF as fees change, which still leaves room for an attacker to manipulate the system. Corallo suggests that the current "near the top of the mempool" criteria is not sufficient for Lightning Network's needs and instead requires certainty that a transaction will confirm by a certain deadline. Russell disagrees with this and argues that his proposed criteria, "in the first 4 MSipa", i.e. next block, would be more incentive-compatible than the current scheme. However, Russell acknowledges that block time variance makes this criteria unreliable. They also discuss potential attacks against the proposed system and agree that simplicity should be a determining factor in implementation.