The sensible policy is that relay should never be more restrictive than what is reliably getting mined in practice.
Anything more restrictive has the collateral harm of increasing centralization pressure by seriously hurting block propagation performance and by driving transactions to direct miner submission.
- gmaxwell
I disagree.
You can, but say why.
It would be better.
Although miners construct the blocks and receive the payout, the nodes are (/should be) more powerful. The nodes are what verify transactions so if a large percentage of nodes point to a different OS (Bitcoin Knots) or even OS version, the nodes ultimately make the decision of what chain to follow.
The nodes can give choices, but the marketplace makes the choice which chain to follow- therefore how big a change is this to not warrant a fork?
I’ve been expecting a fork for years. Bring it. I want to dump coins for more of the legitimate MONETARY Bitcoin.
Nodes follow the consensus valid PoW chain with the most work. Policy be damned
Feels like Bitcoin Core and Bitcoin Knots is playing a game of chicken to see who will fork. After that, I guess we’ll have to see which chain keeps up and/or plummets to $0.
No fork for policy changes.
Will need to look up what constitutes a policy vs a transactional rule.
Grep for these in Knots or score codebase. Iirc policy.cpp has the standard one.
This is what Maxwell had to say in full context. The point he is making is that you don't want to incentive alternative relay networks. Nodes, core or knots or otherwise, will still validate and relay valid blocks with valid transactions in them. But it hurts decentralization if you're validating mined blocks that have lots of txns that were never in your mempool. 
Big miners can profit more from these alternative txn relay networks and if you push profitable use cases to these alternatives simply because the public default relay network is too restrictive, then that is far more harmful than the alternative.
Yeah, I read his post. If Bitcoin mining companies have to result to accepting bribes from spammers, then that’s a terrible business model because eventually scamming trends end and others begin. They’ll be driven out, reputation damned, and then mining hardware will be sold to cover bankruptcy costs. The way I see it, more home mining will eventually take over.
How would they be driven out if it is economically more profitable and viable?
Scams die out so it doesn’t STAY profitable. Look at NFTs on ETH; scams couldn’t even last on a scamchain.
For sure. I don't like NFTs as much as the next person, but this isn't really about NFTs. You don't design network protocols based on market sentiment. As we all know, markets will stay irrational for longer than you can stay solvent.
The nodes are the ultimate decision-makers of the protocol. The most important part. We are the knots
I don’t want to give up on Core. It’s what Satoshi gave the world. How terrible would it be to have his gift destroyed and now we need a life raft for the life raft?
Sounds retarded.
Yes, totally agree, except for spam nodes and underinformed nodes of course. They shouldn't have too much of a say. You need to be able to devote resources in order to have a large sway on things.
Although my method is individual choice of filters and of OP_RETURN pricing per node for their mempool policy through Knots, I suspect there is a better solution, one that aligns miners with noderunners and disregards spam. Not sure what form it would take. Perhaps we don't realize that that mechanism already exists, or that we could make use of something that's being underutilized.
Being your own miner is something. There's the transaction fees we pay to miners to mine legit stuff. There's the higher fees we make spammers pay with our filters. A large portion of miners making private payment rails might not be all bad too... they'd still be interested in finding new transactions to mine, which they would continue to source from the P2P network (as shitty and quasi-permissioned as this would be, it would never completely censor anyone unless I made a mistake in my logic). We could perhaps require arbitrary data to also burn some bitcoin into provably unspendable UTXOs, which subsidizes all bitcoin holders and thus pays economic noderunners directly.
Do you agree?
I have also heard a similar perspective during RBF. That applying personalised policies may only reduce the efficiency of your node.
If your node discards a transaction that the miners are very likely to confirm, then when a block comes in, your node needs to validate it from scratch.
Under ideal circumstances, your node has already validated the transaction during idle time and can accept a block much quicker because of it, helping it to keep up with 10 minute blocks.
Instead, if your policy is irregular, you may have wasted idle time validating a transaction that was obviously unlikely to be confirmed.
Same outcome: optimise for what will get mined, not what you want to get mined.
Greg always comes through with a calm, insightful post in the middle of everyone freaking out.
Economically rational miners, aka majority hashrate, don't care about policy. It's irrelevant to their bottom line.
That leads to a race to the bottom, i.e. zero restrictions. Bitcoin would not survive that. Restrictions define a protocol.
seems like thats something that requires some flexibility and left for nodes to decide
Yep!
Not necessarily. There is a certain theshhold where there is effective pressure without meaningfully incentivizing the robust development of direct miner submission. Where that threshold is and where it will be in the future, I don't know, nor do I know how to keep miners' decisions in line with economic noderunners.
This is very much key to the issue, keeping robust choice in storage and mempool policy to enable spam to be kept out, while also keeping miners interested in mining blocks reflective of the choices of nodes that care about the Bitcoin ethos.
I think individual choices as regards spam filters and the configurable pricing for OP_RETURN that Knots offers is likely, though not certainly, a very good approach that puts pressure against spam while allowing individual choice AND preventing miners from bypassing the nodes in mass. There might be better solutions.
Totally agree—overly restrictive relay policies just choke block propagation and push transactions to miners directly, fueling centralization. Keep it practical, like Bitcoin’s ethos demands.
It sounds like an argument one would make to get nodes to CHOOSE such an option.
So because some spammers might go to the trouble and expense of direct miner submission we should abandon all attempts at stopping spam? This is not a credible argument.
nevent1qvzqqqqqqypzqquxdpn0xlh4zqw9k3patfqml9nnndqkyd9e642sfxzlycj5279pqy2hwumn8ghj7un9d3shjtnyv9kh2uewd9hj7qghwaehxw309aex2mrp0yh8qunfd4skctnwv46z7qpq2h0sstzrc7acfjk47xgpvx3v2yuhqz0w55ruqv2fxs0jevksnr8q69urpp
No because miners will scam the scammers by charging a premium to mine their transactions the network disapproves of. The bigger the miner the larger the premium. Subject spammers to THAT fee market. Filters work.
The issue in my eyes isn’t the specific pull request, but the precedent it sets for future contributions particularly how changes are evaluated, discussed, and accepted by maintainers and the broader community. Could be good or bad thing.
If something is filtered by 100% of nodes, it doesn’t reliably get mined. There’s an additional step and cost. Which is a business opportunity for somebody.
Knots or else beside Core and ossify L1
There will always be miners who serve the financial incentive to accept direct submission of transactions. This will trend to be more the norm in the long run as block space becomes more reliably congested (that is, unless Bitcoin disappointingly remains a niche project).
Let these blocks deal with disadvantaged propagation should the network as a whole choose to impose it.
