delvingbitcoin

Replace-By-Fee-Rate vs V3

Replace-By-Fee-Rate vs V3

Original Postby glozow

Posted on: January 30, 2024 10:35 UTC

The discussion in the BIP v3 alternatives section addresses numerous common suggestions for Bitcoin's transaction replacement policies, explaining why certain ideas are unfeasible or appropriate only for specific scenarios.

A key point is that any transaction replacement should ensure faster confirmation than the original, meaning higher fees and feerate should result in quicker confirmations. Additionally, it's emphasized that replacements should increase the "miner score," a metric reflecting miner incentive compatibility.

A critical issue with some proposals, like those advocating for two sets of Replace-by-Fee (RBF) rules based on feerate or miner score, is the persistence of free relay problems, which can lead to infinite replacements. This challenge is exemplified by discussions on emergency RBF and feerate prioritization, found in proposals such as "Emergency RBF (BIP 125)" and "Fees in Next Block and Feerate for the Rest of the Mempool." The implementation of alternative rules that are both safe and useful is challenged by the current inability of the mempool to efficiently calculate miner scores due to unbounded cluster sizes and the complexity of the existing proposals, including one that attempts to cache the top block's worth of transactions.

Despite these challenges, the suggested path of upgrades, encompassing v3 package RBF, sibling eviction, 1p1c package relay, and ultimately cluster mempool, appears to be the most comprehensive solution. The v3 proposal, sometimes mistaken for being underdeveloped due to its simplicity, offers valuable benefits, serving both as a foundational building block for further improvements and addressing specific problems within the current system. It strategically avoids implementing full cluster limits by codifying a cluster limit count of two, utilizing package limits to prevent infinite clusters, and facilitates progress towards the cluster mempool, which promises improved calculation of metrics like miner score and incentive compatibility across the entire mempool.