Knots Says Bitcoin Is the Altcoin. Its Own Chain Is Minting Coins Nine Times Faster.

Bitcoin Knots 29.4.1 became a final release at 19:02 UTC on Tuesday, September 2. Its release notes describe the software as "mitigating the ongoing attack" and say that on August 8 "most former miners abandoned Bitcoin en masse" and have been "falsely promoting a new altcoin as 'Bitcoin'." The altcoin in that sentence is the chain that roughly 925 exahashes per second of SHA-256 mining power, every major exchange and every wallet in ordinary use call Bitcoin.

Luke Dashjr, the fork's lead developer, put it more briefly half an hour later: "Bitcoin is back online! Scammers in shock."

Satoshi Gazette looked at what the two chains were producing on the same evening. The chain that calls itself Bitcoin has been issuing new coins at roughly nine times Bitcoin's pace since its difficulty reset. The chain it calls an altcoin produced blocks at its normal ten-minute rhythm.

Three chains and one name

The dispute started as a softfork. BIP-110 asked nodes to reject blocks carrying arbitrary data after August 8. Most mining power kept producing those blocks, so the rule never activated on the main chain. The project's own site says so plainly: "most SHA256d mining power kept producing those blocks, so the money-only rules could not activate."

The response was a hard fork with a new proof-of-work algorithm. The first BLAKE2b block is 961,640, mined on August 30, and the project describes the algorithm change as permanent. Its data limits, by contrast, are temporary and expire on September 1, 2027 unless nodes vote to continue them. The pull requests carrying the change, knots#359 for the algorithm and knots#357 for the new opt-in signature format, were still open on GitHub when the binaries shipped; the release was cut from a release branch.

There is a third chain. Bitcoin Purity, an anonymously run Knots fork, kept SHA-256d, made the data rules permanent and lowered its difficulty. Its site says its mainnet launched at 10:00 UTC on August 19 at height 961,637. Its block explorer shows heights that look like the BLAKE2b chain advancing and are not. Anyone reading an explorer this week should know which of the three chains it indexes.

Much of the week's coverage did not. Reports that "the fork died after two blocks" describe the August 8 SHA-256 attempt, not the BLAKE2b chain that has been producing blocks since August 30.

Nine times the issuance, for now

The most detailed public log of the BLAKE2b chain comes from a mining-guide site that reads heights from its own node. As of 11:24 UTC on September 2 it recorded the chain at height 965,704. Its first retarget, at 963,648 on September 1, cut difficulty 41.3 percent, from 30,393,776 to 17,842,860. Its second, at 965,664 on September 2, measured a window that ran at 31 seconds a block and raised difficulty by the maximum factor of four, to 71,371,438.

SG did the arithmetic those numbers imply. From block 961,640 at 06:14 UTC on August 30 to block 965,704 at 11:24 UTC on September 2 is 4,064 blocks in 77.2 hours, about 1,264 blocks a day. Bitcoin's schedule is 144. At the current 3.125-coin subsidy, the fork issued about 12,700 coins in three days, roughly 28 days of Bitcoin's issuance, on a difficulty about 1.8 million times lower than Bitcoin's 125.8 trillion. By height alone the fork is now 478 blocks ahead of the chain it forked from.

That inflation is temporary by design. Difficulty can rise at most fourfold per adjustment, and the site expects the next retarget around September 5. It is also unverified by SG: we have not run a Knots 29.4.1 node, so these heights are the operator's, not ours. But the direction is not in doubt, and it sits awkwardly beside a project whose home page says it "keeps Bitcoin as money only."

The same log undercuts the "ordinary computers can mine again" framing. Its first hours were mined by about a dozen operators running Siacoin-class BLAKE2b ASICs, and it calls CPU mining "non-viable from the start." A different ASIC is still an ASIC.

The "crawl" does not appear in Bitcoin's blocks

The Knots notes say the August 8 events "brought the network to a crawl." On the evening the release shipped, mempool.space showed Bitcoin at height 965,227, fourteen recent blocks averaging 12.1 minutes apart, block sizes between 1.58 and 1.72 megabytes carrying roughly 3,800 to 5,100 transactions each, and an average block time of 9.93 minutes across the current difficulty period, which was 78 percent complete and heading for an adjustment of about +0.76 percent.

That is what a normal fortnight looks like. Whatever the fork calls the majority chain, the majority chain has not noticed.

What a holder actually risks

Nothing about this fork requires a Bitcoin holder to do anything. Coins are not moving. The risks are at the edges.

The first is replay. Because the BLAKE2b chain shares Bitcoin's history through block 961,639, a transaction signed for one chain can be valid on the other. Bitcoin developer Murch wrote in July that on the forked chains "any transaction made on either chaintip would likely result in funds on both chaintips being moved unless they take special precautions." Bitcoin Optech's August 28 newsletter records that a minority chain had already been "subject to replay attacks" and that developers are discussing opt-in protection for future forks.

Knots' own answer is a new signature format, SIGHASH_UNIFIED. Its release notes tell anyone who wants to sell or spend "fake 'bitcoins'" on what it calls the altcoin to re-send their coins to themselves with the new format first. That works only in wallets that implement it, which today means the fork's own software.

The second is Lightning. Start9's operator guide, written before the algorithm change, says to "assume that anything you broadcast on one chain also spends your coins on the other," and warns that a channel counterparty could publish a revoked state on the chain you are not watching. Its advice was to close channels before the split or leave them alone until the picture clears.

The third is the market for the new coin. One report found a single small exchange quoting the forked coin under the ticker BTCB2 with bids at $82 and asks at $190, no major listing and negligible volume. Any "claim your BLAKE2b coins" service that asks for a seed phrase is the scam the Bitcoin community has been warning about since August.

What the record cannot settle

SG cannot say how much BLAKE2b hashpower exists, who is running it, or whether any exchange will credit the coins; the project itself tells exchanges only to "follow the BLAKE2b chain they intend to credit and list." Nor can SG say whether the chain persists once its difficulty catches up with its hashpower and the reward per hour falls back to Bitcoin's rate.

What the record does settle is smaller and more useful. A fork with a different algorithm is a different network with the same old ledger. Renaming the larger one does not move a single satoshi. And a chain that describes itself as money-only spent its first three days issuing money nine times faster than the network it left.