@cypherdoc I only have a general overview, I think some of the changes create efficiencies and some sound worthwhile. Re: "force all hashers to become full nodes" I'm not sure this will be the case, although it's sold this way. and yes it'll be less efficient with a homogenous Core node relay policy, already Core censors my transactions with their "dust limit" degrading BTC's fungibility.
How I understand it; all functions of mining would be separated. There are the SHA256 hashing boards operators they are just responsible for finding cheap electricity and connecting the boards to the internet. They will probably be directed wherever there is a marginally higher profit. (they could keep their plausible deniability while actually collecting block rewards while taking part in a 51%)
Then there are the "Miner" they will run a node it can be anyone, a home miner, Bitfury or the NSA they just compile blocks of transactions and send needed data to the hashers who look for the golden nonce.
When a "Hasher" finds a golden nonce (a block) the "Miner" sends the block hash to the pool. The pools job it to propagate the block on the network and split up the block reward among the other "Hashers".
The ability to censor transactions anonymously becomes easy, it's being sold as a benefit. Eg. the benefit is I could keep running a node on my rPi at home, (theoretically not censoring transactions) I keep running a single S9 in effect mining on my own block similar to P2Pool.
The truth is I'm compiling blocks according to the Core developers default rules and I will never find a block. The larger Miners 10MW and up will have a "Miner" optimize the blocks for them - "Miners" may even pay Hashers in addition to the block reward that goes to the hasher. Probably the LN hubs owners would become Miners, and the people who run hashing bords just become cheap electricity finders.
One aspect of Matts BetterHash proposal I don't like is this neglect of the economic incentives to optimize blocks for propagation and revenue, it socializes benefit resulting from competitive transaction optimizing strategies giving full control to the Core developers.
In a free market, those miners who better predict market rates or optimize for propagation gain market share over time as they earn more revenue (a market mechanism that rewards optimal performance). A miner who is good at optimizing fee-paying transactions or soliciting business from high paying smart contracts is not rewarded. His reward is split by the pool who broadcast the block. (socializing the function of mining) "Miners" are not rewarded directly for optimizing transactions. Matt calls this feature "decentralization" - further distorting the meaning of the most abused word in Bitcoin.
As a small miner, running my own node and compiling my own blocks I have less say in that I can't choose a pool based on its performance or its reputation, there may just as well be 1 global pool as there is nothing for pools to compete on. While I am part of an independent pool I may be providing 0.0000001% decentralized transaction selection to a pool that is supporting 99% censored transactions. The kicker is the pool is not responsible for the censoring of transactions the Miner is and the miner is anonymous, a mining LN hub cartel is sure to form should Core's roadmap actually come to fruition.
I'm glad this is not happening on the BCH chain.
