bitcoin-dev

CheckTemplateVerify Does Not Scale Due to UTXO's Required For Fee Payment

CheckTemplateVerify Does Not Scale Due to UTXO's Required For Fee Payment

Original Postby Michael Folkson

Posted on: January 25, 2024 12:57 UTC

Michael Folkson's email addresses the comparison between two different Bitcoin transaction mechanisms: CheckTemplateVerify (CTV) and Replace-By-Fee (RBF), specifically in high fee environments.

He points out that CTV firmly aligns with Child Pays for Parent (CPFP) rather than RBF, highlighting a potential issue where paying for an additional CPFP transaction might be less efficient than simply replacing an unconfirmed transaction with RBF.

Folkson refers to Antoine Poinsot's contribution on Bitcoin StackExchange, which argues that RBF could be considered superior to CPFP from a transaction originator's perspective because CPFP is less efficient and sometimes not possible. A transaction eligible for CPFP must have a change output and its parent must pass certain policy checks; for example, it cannot be below the minimum feerate of most network mempools.

The conversation also touches upon the implications of these transaction policies for Lightning Network (LN) symmetry, comparing CTV-based LN-Symmetry to an Anchor Outputs (APO)-based LN-Symmetry. In theory, APO-based LN-Symmetry could adjust fees with every channel update according to the current market rate, whereas in the current pre-LN-Symmetry world, "justice transactions" for revoked states are constructed without consideration of future changes in market fee rates. This can lead to discrepancies when the market fee rate at the time of justice transaction execution differs significantly from the rate at which it was initially set.

Michael closes his email by offering contact information and a resource for learning about Bitcoin, providing a YouTube link to his channel.