Replying to Avatar Dr. Bitcoin, MD

As the reference implementation, core should have “the best” policies from a genuine comp sci and networking perspective, not a commercial one.

Any major policy decisions should be handled at the fork level. Want more data carrier size? Fork it! The thought that this should be debated on github is laughable.
Avatar
Sjors Provoost 8mo ago 💬 1

I would recommend Greg Maxwell and AJ Towns latest replies on the mailinglist. It's harmful if people use different settings, whether that's by configuration option (-datacarriersize) or by code forks (Knots). The chaos of a "free market" for settings is disruptive for the network, which only helps its real adversary: governments.

The custom Knots default of 42 bytes caused problems in the past for Samurai wallet (which they in turn used for their drama marketing). In theory it should also interfere with compact block relay (I haven't seen measurements though).

https://gnusha.org/pi/bitcoindev/aBSVn7nJnrheLy5Z@erisian.com.au/

Reply to this note

Please Login to reply.

Discussion

Avatar
Dr. Bitcoin, MD 8mo ago

It’s tricky to parse what disrupts the network from relay and miner perspectives. I can see a miner having to validate a block with transactions it’s never seen before to be a disruption, but for non-mining nodes I don’t see it as a problem.

For non-mining nodes, relaying/accepting transactions that won’t get mined certainly doesn’t make sense; nothing-at-stake mempool attacks are an obvious consequence.

All that said, perhaps the real issue here is analogous to the misaligned incentives leading to Napster: people wanted downloadable music and there was only one way to get it. It wasn’t until music companies stopped suing would be customers and started selling digital music did the issue go away.