What do you think of Peter’s fix that he published to bitcoin-dev?
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2024-January/022326.html
I don’t do deep protocol design so maybe I’m missing something. But the fee-rate/depth argument looks pretty solid to me and I guess it must be actually deployed in Libre Relay nodes.
In your writeup, I don’t understand how step #3 is following one-shot rbfr rules. Isn’t that just a normal replacement with a tx of a higher fee? 16,500 > 5000
Not sure I understand your argument there.