Thinking this through once again: #ordisrespector does not seem dangerous knowing that:
- bitcoind does not consider the mempool for fee estimation (only the past blocks).
- LN implementations and Electrum servers rely on the data from bitcoind.
Please correct me if I am wrong.
Yes, unless you use specific software that takes local mempool into account when estimating tx fees.
Thanks, I know about two such software:
- mempool (the FOSS project)
- Bitcoin Knots
are there any more that might be affected by changing the content of my local mempool?
Knots? AFAIK ir has the same estimator as Core, haven’t noticed big differences in estimations. Maybe @Luke Dashjr can comment.
Don’t also Sparrow Wallet have custom estimator?
Only difference between Core and Knots fee estimates is Knots by default targets confirmation within 24 hours.
Looking at your current mempool for fee estimation is broken by design
What Wasabi does is looking at Knots estimates and then increasing fee if tx will not fit in top 200 MvB of current mempool with that estimated feerate.
a very useful correction here: https://nitter.net/mononautical/status/1725813952377127375
bitcoind does rely on the mempool for fee estimation.
the estimatesmartfee algorithm works by tracking how long transactions in different "fee buckets" wait in the mempool before getting confirmed.
which means it needs to see a representative sample in the mempool.
Oddbean new post about login | logout