Block propagation consistency restored

[gmaxwell]

FWIW, I observe now that >95% of blocks are being received with 0 round trips on a node again (97.2% in the last day, to be more specific), which is back in line with historical behavior and the design of compact blocks.  This is significant because any round trip immediate doubles or more the time it takes to propagate a block,  and all delays in block propagation advantage larger hashrate miners relative to smaller hashrate miners.

This figure previously was hitting below 40% some months ago, due to mining policies of major miners diverging from node defaults.  But node defaults were widely updated to match reality and block propagation is now working correctly again.

(Please note that any posts whining about wanting to censor transactions at the expense of fucking up the network are offtopic in this thread and will just be removed.)

[philipma1957]

FWIW, I observe now that >95% of blocks are being received with 0 round trips on a node again (97.2% in the last day, to be more specific), which is back in line with historical behavior and the design of compact blocks.  This is significant because any round trip immediate doubles or more the time it takes to propagate a block,  and all delays in block propagation advantage larger hashrate miners relative to smaller hashrate miners.

This figure previously was hitting below 40% some months ago, due to mining policies of major miners diverging from node defaults.  But node defaults were widely updated to match reality and block propagation is now working correctly again.

(Please note that any posts whining about wanting to censor transactions at the expense of fucking up the network are offtopic in this thread and will just be removed.)

Nice job I assume this issue happened due to the 0.1 to 0.9 sat fees?

So are all nodes now taking 0.1 sats as the bottom number

[gmaxwell]

Nice job I assume this issue happened due to the 0.1 to 0.9 sat fees?
So are all nodes now taking 0.1 sats as the bottom number

And presumably the other standardness limits that were changed (e.g. data carrier) but primarily fees.

Not all nodes by any means, but enough that blocks are propagating well among most of those that do.  It doesn't take all of them, but just enough to form a short connected graph among those that do, and blocks will selectively ride the fastest paths around the network and route around nodes that propagate slowly because they blocked transactions that got included.

[satscraper]

FWIW, I observe now that >95% of blocks are being received with 0 round trips on a node again (97.2% in the last day, to be more specific), which is back in line with historical behavior and the design of compact blocks.  This is significant because any round trip immediate doubles or more the time it takes to propagate a block,  and all delays in block propagation advantage larger hashrate miners relative to smaller hashrate miners.

This figure previously was hitting below 40% some months ago, due to mining policies of major miners diverging from node defaults.  But node defaults were widely updated to match reality and block propagation is now working correctly again.

(Please note that any posts whining about wanting to censor transactions at the expense of fucking up the network are offtopic in this thread and will just be removed.)

Interesting observathions. Are these observations based solely on internal metrics ^{and it would be helpful if you could specify which ones} from your own nodes, or is there general statistics available that apply to all active nodes?I’ve only been able to find relevant data ^{gathered from 10 nodes}for 2024 , and nothing for 2025/2026.

[stwenhao]

and nothing for 2025/2026

You have more recent stats in the same topic: https://delvingbitcoin.org/t/stats-on-compact-block-reconstructions/1052/44

[Donneski]

It's interesting to know that this recovery happened without all nodes needing to align, just enough of the well-connected graph behaving efficiently.
This really reinforces how fragile propagation can be when miners drift from sane defaults. The fact that it dipped below 40% not long ago I think is a reminder that policy changes can quietly hurt decentralization before anyone notices.

[gmaxwell]

My figures are just from watching a couple nodes, though I expect similar figures on most nodes running current software, it's possible for some to be lower if they're surrounded by peers with more restrictive policy.