# Luke Dashjr activates the BLAKE2b hard fork with no support and no exchanges

> Bitcoin Knots activates the BLAKE2b hard fork at block 961,640: Dashjr's chain starts out isolated, with no price and no exchange support
> By José Maldonado · Bitcoin · 2026-08-31

On Sunday, August 30, Luke Dashjr [**posted on X**](https://x.com/LukeDashjr/status/2093930821279183294) the parameter that was to seal the activation of his hard fork: `blake2b_headline=8-30 NYPost Deride And Conquer`, the canonical headline that the first block of the new chain had to include in its coinbase transaction.

![](https://content.observatorioblockchain.com/wp-content/uploads/2026/08/luke-blake2b.png)

Hours later, at block 961,640, the minority chain driven by Bitcoin Knots abandoned SHA-256d and began producing blocks with BLAKE2b, the proof-of-work algorithm change its group has been preparing since the BIP-110 soft fork split in two in early August. The trial announced for Sunday became a real activation.

## From failed soft fork to algorithm change

The activation caps three weeks of escalation within the Bitcoiner community. On August 8, BIP-110, the temporary soft fork pushed by Luke Dashjr to restrict arbitrary data in transactions (Ordinals inscriptions, text and images), entered mandatory signaling at block 961,632 with barely 2% of the hashrate.

But the node chain enforcing it stalled after initially producing only two blocks; the Roughnecks pool extended it in dribs and drabs under Bitcoin's inherited difficulty until mid-month, by which point the main chain was already more than 1,500 blocks ahead.

Unable to bend the SHA-256d miners, on August 11 Bitcoin Knots drew a new PoW algorithm by lot through a commit-reveal process on Testnet4 and confirmed BLAKE2b the following day, chosen from six candidates (BLAKE2b-256, SHA3x, BLAKE2s, Scrypt, BLAKE3x2 and Eaglesong).

Later, on August 15, Dashjr opened the draft of the [**hard fork pull request**](https://github.com/bitcoinknots/bitcoin/pull/359) in the Knots repository and, from August 21, the new header was tested on Testnet4. On Saturday the 29th, [**he asked**](https://x.com/LukeDashjr/status/2093527828948045982) SHA-2 miners to stop production before the trial and set release candidate rc4 as the one in charge of marking the last SHA-2 block. That same day his definitive departure from OCEAN was announced.

**[Chinese fork in Bitcoin: A former PBOC official launches Bitcoin Purity](https://observatorioblockchain.com/en/bitcoin/fork-chino-en-bitcoin-exfuncionario-del-pboc-lanza-bitcoin-purity/)**

## What changed at block 961,640?

The flag-day fork is permanent: from height 961,640, SHA-256d blocks are invalid and only those mined with BLAKE2b are accepted, with the initial target shifted 22 bits, equivalent to about 218 TH/s for ten-minute blocks. The difficulty adjustment every 2,016 blocks is maintained.

The change is not limited to the hash either. The new header goes from the historical 80 bytes to 164, and incorporates block height, transaction count, expanded nonce spaces (128 bits per ASIC level and 128 bits per machine), ASIC profiles, time-shift data, an XOR key and a merge-mining field.

The direct commitment of the transaction count closes the Merkle tree ambiguity behind CVE-2017-12842, and the optional XOR key mitigates block withholding attacks. SIGHASH\_UNIFIED is also added, a signature mode that commits to all amounts and offers one-way replay protection, though only when the sender chooses it.

The Reduced Data Temporary Rules (RDTS) limit blocks to 800,000 weight units, about 300 kB, until September 2027, with additional restrictions on Taproot scripts. The header reserves space for merge-mined sidechains and 40-bit timestamps, which will require another hard fork to activate.

## Solving the mining problem

The algorithm change seeks to attack the cause of BIP-110's failure, according to Dashjr. The BIP-110 minority chain inherited Bitcoin's full difficulty, a level of security it could not sustain with its small fraction of hashrate. By switching to BLAKE2b, SHA-256d ASICs are left out and the door opens to machines designed for the Sia cryptocurrency, such as Bitmain's Antminer A3 or the Goldshell SC5, and in theory to CPU and GPU.

But hardware being compatible does not mean the hashrate is committed. A code reviewer estimated that a version of the initial target required close to 870 TH/s to maintain ten-minute blocks, while the capacity measured on Testnet4 was around 50-70 TH/s. The problem here is another one: no one has shown interest in sustaining the infrastructure to mine, let alone to create an ecosystem around this "new Bitcoin".

On the other hand, the network's first retarget will not arrive for another 2,016 blocks, so if only that capacity shows up at launch, the production rate would be hours per block. In fact, the [**release candidate**](https://github.com/bitcoinknots/bitcoin/releases/tag/v29.4.1.knots20260508rc4) notes pointed to a lower initial target, but the gap between the power needed and the confirmed power remains the design's first test.

**[Bitcoin nears $80,000 and Saylor sets off the party on X with sequins](https://observatorioblockchain.com/en/bitcoin/bitcoin-80000-dolares-saylor-lentejuelas-x/)**

## The chain works, but the ecosystem does not exist

But producing blocks under common rules is only the first link. First, the 164-byte header breaks compatibility with Electrum-style clients, which expect Bitcoin's 80-byte headers. This means light wallets, indexers and explorers need to adapt to the change which, according to Dashjr, remains outside the scope of Bitcoin Knots. This means no exchange, custodian, explorer or Lightning implementation has publicly committed to supporting the chain, and the project's FAQ recommends that exchanges pause deposits and withdrawals around the split.

In addition, the replay risk remains. Both chains share the history and balances from before the fork, and a transaction that spends pre-fork coins can be valid on both. SIGHASH\_UNIFIED offers directional protection, but only if explicitly enabled.

In fact, more than 400,000 unconfirmed transfers have piled up on the minority chain since mid-August, many of them accidental, and the currency has no price or verifiable market.

This situation recalls the 2017 forks, but with one difference: back then the majority secured wallets, exchanges and a price; here none has appeared. A third fork has also joined the picture, [**Bitcoin Purity**](https://observatorioblockchain.com/en/bitcoin/fork-chino-en-bitcoin-exfuncionario-del-pboc-lanza-bitcoin-purity/), started by a former Chinese central bank official at block 961,637.

## A purge that leaves Dashjr alone

The activation comes at a moment of institutional isolation. On August 29, OCEAN, the pool he co-founded, [**announced the purchase of his stake**](https://cryptoslate.com/veteran-bitcoin-developer-luke-dashjr-exits-ocean-pool-will-hash-power-follow-him-to-new-pool/) and his departure as chairman, CTO and director; Dashjr moves to a new mining project, [**CONVOY**](https://x.com/Convoy_Mining), with no infrastructure or public clients.

This adds to the fact that, a week earlier, on August 11, Dashjr was [**removed as editor of the BIPs repository**](https://decrypt.co/375310/luke-dashjr-removed-as-bip-editor-after-bip-110-bitcoin-fork-stalls) by the other editors, who considered his handling of BIP-110 "a complete departure from the Bitcoin development ecosystem". The consensus of developers and the market, in Adam Back's words, already ignored the fork.

Tomorrow, September 1, the publication of Bitcoin Knots 29.4.1 final is expected, which would preserve the chain started on August 30; if problems arise, the plan contemplates a reset to the last SHA-2 block with an rc5.

But the chain's survival does not depend on code alone. It needs miners to keep production going and an economy to give it value. As of today, the BLAKE2b hard fork is an altcoin derived from Bitcoin with no price, no exchange and no ecosystem, not a new Bitcoin. The experiment has proven that the algorithm can be changed; it remains to be proven that anything will be built on it.

**[Luke Dashjr speeds up his Bitcoin hard fork to try to reach September 1](https://observatorioblockchain.com/en/bitcoin/bitcoin-hard-fork-dashjr-avanza-blake2b/)**
