pull down to refresh
0 sats \ 8 replies \ @Murch 31 Jan \ parent \ on: Call for nodes reconfiguration: minrelaytxfee=0.00000001 bitcoin_Mining
Accepting replacements that do not pay a higher feerate than the original means that the sender paid nothing for the relay of additional data.
That's not true. It means that the sender paid for the extra bytes at the same fee-rate.
reply
Let’s say I send a 200 vbyte transaction with a fee of 200 sats. Then I get to send a replacement for that transaction that is also 200 vbytes and pays 200 sats.
How much did my second transaction increase the available fees in the mempool? Zilch.
Ergo, the relay of my second transaction was free, and I can repeat that until one of my transactions gets confirmed.
reply
That's ok. In the end there is one transaction in mempools. A transaction is replaced with another one of the same size. What do you think justifies charging additional fee in such a case?
reply
reply
Let's not exaggerate. This setting reconfiguration doesn't cause endless amounts of data to be forwarded.
What makes you considering a transaction with fee at 1 sat/vbyte as okay while the same transaction with smaller fee, e.g. at 0.1 sat/vbyte as not okay ("useless data")?
reply
As a side note, consider also setting incrementalrelayfee=0 instead of the default 3000 in order to relay replacement transactions which fee-rates don't change.
- The default value for the incremental relay feerate is 1000 ṩ/kvB, not 3000 ṩ/vB.
- A single node setting this to 0 ṩ/kvB will not cause endless amounts of data to be forwarded, simply because other network participants will not even be aware, and we currently would not see any attempts to make replacements without paying more.
- Many nodes on the network setting this policy would be problematic: any malicious actor becoming aware of this would be able to endlessly cycle the same two transactions replacing each other among the part of the network that has decreased their incremental relay feerate to 0. This would waste the affected nodes’ bandwidth, and once their ISPs throttle them, essentially remove them from the relay network. We could argue over whether 1 ṩ/vB is too much or too little, but dropping the requirement that the fees in the mempool increase for replacements is definitely problematic: it’s asking nodes to open themselves up to bandwidth wasting and DOS vectors.
- I showed that your initial recommendation was problematic and now you are shifting goal posts without engaging with the substance of my argument. You can do better.