by ajtowns

Over the last few years we’ve started seeing significant divergence in mempool policies, particularly adoption of full replace by fee, various Knots spam filtering options, uncapped datacarrier/OP_RETURN and most recently 0.1sat/vb minfee.

Background

A technical aspect that has influenced those discussions is the impact on compact block relay.

As a quick reminder, it’s fairly important for new blocks to be relayed quickly to other miners: if there are significant delays, that can both (a) increase the orphan rate for smaller miners, leading to higher comparative returns for larger miners, leading to centralisation, and (b) make attacks such as selfish mining easier to perform. Compact block relay, which was released in Bitcoin Core 0.13.0 in August 2016 provides a way to significantly improve block relay speeds, but relies on each node’s mempool making an reasonably accurate prediction of what transactions will be in the next block. When it gets it right, the current tip can be updated without requiring a round-trip over the network, potentially reducing relay time to a third of what it might otherwise be. This effect can be observed in the DSN Block Propagation History charts, driving block propagation times down from around 5s to very close to 0s:

...read more at delvingbitcoin.org