There was some discussion about lowering the minrelaytxfee which resulted in a PR. A growing number of miners are accepting fee rates below 1 sat/vByte.
However, the always vigilant @0xB10C has noticed that this trend may be changing how efficient compact block relay is:
Since #32582 we also log the number of bytes that we requested to reconstruct the block. I plot the average kB requested per block in the following new chart. In June, we were requesting less than 10 kB worth of transactions per block on average for about 40-50% of blocks. Now, we are requesting close to 0.8 MB worth of transactions per block on average for about 70% of blocks.
Also see Antoine Poinsot's post on X about it:
https://x.com/darosior/status/1952753869814976638
So the problem is the miners are using more power for processing? And the network gets all jammed up?
I think the issue is that not very many nodes are relaying sub 1sat /vByte transactions among their mempools and so when a new block is found that has such transactions in it, lots of nodes have to request full details about these transactions which means the new blocks propagate more slowly across the network.
@petertodd suggested that this was only an issue for miners who weren't running nodes that relayed sub 1sat/vByte transactions.
Gotcha! Good explanation. I was on the right track. ahahah, appreciate it.
That's also how I read Antoine's comment "we might want to backport this PR at least to 29." right underneath - though I'd personally be cautious with that (adding policy changes into a bugfix release tree)
Which in turn trends towards miner centralization, given the greater head start to the miner of the last block. This is one of the heaviest arguments against actually bigger blocks.
What I find most interesting is that this is quietly the final scaling frontier. Since throughput is already unlimited, and ownership is a result of fee-relative supply divided by distribution, lots of scale scammers are going to cope hard because lower chain fees drastically shrinks the market for their their centralized shadowchains.
https://xcancel.com/peterktodd/status/1952754386725175408
Isn't the solution for everyone to lower
minrelaytxfee? I mean, once a critical mass is reached the only affected nodes will be the ones that keep it high. It's the filters conversation all over again.I think so, although @bluematt was pointing out on X that 1sat/vB provided some level of DoS protection. If fee rates go too low, what stops an attacker from dossing you with tons of transactions? I'm not sure this fully makes sense, though.
https://xcancel.com/TheBlueMatt/status/1952787518220570749
this seems interesting
https://xcancel.com/mononautical/status/1949452586391855121
Fascinating observation. It really shows how small policy shifts,like lowering the minrelaytxfee,can ripple out and affect network-level mechanics like compact block efficiency.
If more miners are including low-fee, non-relayed transactions (e.g., via direct submission or txpool differences), it makes sense that peer nodes would miss them during block reconstruction. More requests, more bandwidth, more latency.
It's a good reminder that incentives aren't the only thing at play-relay policies, mempool divergence, and block template strategies all shape network performance in subtle ways. Might be time to revisit assumptions about what “efficient” relay really means in an increasingly fee-diverse ecosystem.
Funny how this sub-sat stuff came about after I started getting vocal about the fake scaling solutions didn't address the cost of unilateral transactions that define ownership (and that only a supply increase or sub-sat fees could scale it)
Threw a grenade in the PR: https://github.com/bitcoin/bitcoin/pull/33106#issuecomment-3156258686