Thread of articles...

Desert Rat and BetaMine are enemies of Bitcoin.

perhaps not, but a MRSF/URSF by way of simply ignoring a UASF with 20% "based" consensus is existential. you just don't know yet that sha256 is a dead man walking.

Bounded per-template nonce space - A note for @SteveSimple @cguida6 and anyone else interested... Bitcoin’s original operating assumption was that the machine which searched for a proof-of-work hash was the same machine that selected transactions and checked the rules. Getwork (2010), and then Stratum (2012–2013) split that job. The node kept the mempool and the block template. The hardware received a compact work package—coinbase halves, merkle branches, a work root—and searched a wide nonce space without the transaction list. Validity on the network still meant “this header hashes below target and this merkle root matches these transactions.” It did not mean “the searcher possessed those transactions.” That is the split this note tries to narrow. The Blake chain already changed the hash function and the header. Under Knots #359 a Goldshell is given an 80-byte work buffer and searches an 8-byte nonce on a work root that includes a long extranonce. For one fixed transaction list that is a very large set of legal preimages. Pool software can send merkle branches once. The hardware can search for a long time. GetPoWHash accepts the result. Stratum version 1 remains a sufficient mining protocol.The proposed change is a second flag-day height. From that height, GetPoWHash includes a commitment computed from the transaction list in the block. CheckBlock requires that commitment to match the transactions the node received, and it caps coinbase extra-nonce and the rollable header extranonce. After previous block hash, nBits, nTime, and that commitment are fixed, only a short nonce field may change. A node that has the header and the body can verify all of this. A node that has only merkle branches cannot produce a legal preimage. Remote hashing without the transaction list is invalid. Remote hashing with the transaction list is valid and expensive in proportion to distance. There is no timestamp in the rule. The phrase “ten to twenty milliseconds” is how one chooses nHeaderNonceBits from measured hashrate so that one fixed template is exhausted on about a city-scale path. Blocks still propagate in the ordinary way. nTime remains coarse. A Goldshell in service today does not implement those checks. It hashes the 80-byte buffer it is given. After the flag day, firmware that only increments nonce8 on one work root will not produce headers that satisfy GetPoWHash. The same unit can still function if mining software that holds the transaction list computes a new legal work buffer each time extra-nonce or the commitment changes, and delivers that buffer to the device. The hash array still runs BLAKE2b on 80 bytes. Usable hashrate then depends on whether that software can refresh the buffer before the short nonce is exhausted. On a local network, with a competent controller, the rated TH/s can remain largely available; the loss is job-switch overhead and any period spent on a buffer the node will reject. Across a high-latency path the same unit wastes a large fraction of its hashes on work that is already exhausted or not yet replaced. Across a path that never carries the transaction list, its valid hashrate is zero.Hardware designed for the new preimage would search the short nonce and the transaction-list commitment directly, without waiting on an external translation into the old 80-byte layout. That hardware will convert electricity into accepted blocks more efficiently than a current Goldshell sitting behind a job queue. The point of the rule is not to create that product cycle. The point is to make Stratum version 1’s compact job insufficient for valid work, and to make a local template source—DATUM, or any process that takes getblocktemplate from a node that has the mempool—the only inexpensive way to keep hardware busy. What this does not restore is the 2009 identity of miner and node as one person. Pool software on the same local network as the hardware can still build a legal template, use a short nonce, include the commitment, and broadcast a block every peer will accept. Consensus will not see a difference. What it does restore, in part, is the coupling Stratum removed: legal search requires the transaction list, and moving that list far from the hardware has a cost in accepted hashrate. That is a partial return of work to the node. Th CheckBlock restores the part of the white paper that Getwork and Stratum made optional.

maybe there is some way to apply a verified public address mining consensus in the protocol such that it references a mempool guide and the next block is tentatively accepted provided: (i) if it is a new mining address = 1 yr coinbase maturity (ii) if mining address earns more than [10%] in a rolling 100 block window = invalid block in this way hostile miners that simply use a new address will be locked out of spending for a year. and hostile miners that want to control the chain will be locked out indefinitely until they scale back operations. we should not fear keeping large miners smaller. instead, we should fear small miners opting out of mining, since it is these miners that are the backbone of the network. we should preference the network towards 1000 small miners over 1 large miner. socially/culturally this scales significantly better in the long term (i.e. beyond our lifetimes). Cc @LukeDashjr @cguida6 @mattkratter @SteveSimple