BIP-110: how a soft fork became a new cryptocurrency

What BIP-110 asked for

BIP-110, titled Reduced Data Temporary Softfork and authored under the name Dathon Ohm, set out to limit how much arbitrary data a Bitcoin transaction can carry. Its main public champion was Luke Dashjr, who maintains Bitcoin Knots, the alternative node software that shipped it (opt-in from Knots v29.3, May 2026). For about a year (52,416 blocks) it would have made these invalid:

• new output scripts longer than 34 bytes, except OP_RETURN outputs, which may run to 83 bytes;
• data pushes and witness items over 256 bytes;
• spending undefined witness or Tapleaf versions, the Taproot annex, and Taproot control blocks over 257 bytes (a tree of more than 128 leaves);
• any OP_SUCCESS opcode in a tapscript, and executing OP_IF or OP_NOTIF in one.

Coins created before activation were exempt. The rules target inscriptions and other data stored in Bitcoin transactions, which their supporters regard as spam and their opponents regard as transactions that pay the fee like any other.

Activation was the part that mattered. Miners were asked to signal with version bit 4, and 55% of a 2,016-block period (1,109 blocks) would lock the rules in. Unlike an ordinary BIP9 deployment, which simply times out when miners do not signal, BIP-110 made signalling mandatory from block 961,632: any block without the bit would be rejected by the nodes enforcing it. That clause is what turned a vote into a fork.

8 August: the split

In the last voluntary signalling period, 51 of 2,016 blocks carried the bit — 2.53%, all of them reportedly from the OCEAN mining pool, which had switched its miners to signal by default in July. The last block both sides share is 961,631, mined at 19:33 UTC on 8 August 2026.

The next block on Bitcoin came from AntPool and did not signal. Every node running Bitcoin Core, and every Knots node not enforcing BIP-110, accepted it. The enforcing nodes rejected it and waited for a block that signalled. One arrived at 20:12 UTC, mined through OCEAN by the Roughnecks miner group, and from that moment there were two chains with a common history and two different blocks at height 961,632. This is the case the journal's hard-fork note describes: a rule change without the coordination it needs, run by two groups of nodes that no longer agree on which blocks are valid.

The BIP was marked Closed on 9 August, "following a chain split with stalled mining". On 10 August Luke Dashjr was removed as a BIP editor; on 29 August he left OCEAN, where he had been chief technology officer, and started a new pool, CONVOY.

Twenty days, eight blocks

Most coverage reported that the minority chain mined two blocks and stopped. The chain's own explorer shows eight: two on 8 August, two on 12 August, then one each on the 17th, 21st, 23rd and 28th — the last SHA-256 block at height 961,639. A proof-of-work chain retargets its difficulty every 2,016 blocks, so a branch that inherits the full Bitcoin difficulty with a sliver of the hashrate does not get easier; it waits. At the pace of August, the first retarget was years away.

That is the arithmetic every minority split faces, and it is why Bitcoin Cash in 2017 shipped with an emergency difficulty adjustment written in before the fork. BIP-110 had none, because a soft fork is not supposed to need one: it assumes the majority of hashrate is on its side.

30 August: the restart on BLAKE2b

The way out was to stop competing for SHA-256 miners altogether. Knots pull request #359, "Hardfork: New BLAKE2b proof-of-work algorithm", replaced Bitcoin's double SHA-256 with BLAKE2b — the algorithm of the Sia network, whose purpose-built ASICs from several manufacturers had little else to mine. The selection was reportedly made by a public random draw among six candidate algorithms; critics, Peter Todd among them, questioned whether the draw could be steered. The change also introduced a new 164-byte block header and cut the difficulty by a factor of 2²², so the new machines could find blocks at once.

Block 961,640, the first BLAKE2b block, was mined at 06:14 UTC on 30 August. The production release, Bitcoin Knots v29.4.1, followed on 2 September. With it the BIP-110 data rules switched on — at the proof-of-work change rather than by signalling — together with a temporary block limit of 800,000 weight units, about a fifth of Bitcoin's, both until 1 September 2027.

Two details shape how the chain behaves today. Addresses, network ports and transaction formats are the same as Bitcoin's, and replay protection is opt-in only, so a plain transaction signed for one chain is valid on the other unless the sender takes care. And the release notes invert the usual vocabulary: the BLAKE2b chain is "Bitcoin", and the chain every exchange calls Bitcoin is the "altcoin".

The chain today

Chain figures read from the mempool.guide explorer and the Knots releases on 21 October 2026.

Choosing which miners get paid

The miners who arrived with the BLAKE2b switch were not the miners the fork was built for. Within days a single pool briefly held more than half of the hashrate, and the project's developers began describing pools that do not build their own block templates as attacks on the network. Knots v29.4.2, released on 21 September, answered with a consensus change: block rewards mined between heights 973,440 and 979,919 cannot be spent for 6,480 blocks — about 45 days — instead of the usual 100.

That window closes at block 979,920, around the start of November at the current pace. Four competing proposals are open for what follows: keep a rolling 45-day lock until the data rules expire, extend it to 180 days, extend it to a year, or make the BIP-110 rules permanent with a lock of about 280 days. None had been merged on 21 October. Whichever is chosen, it is a new lever in a proof-of-work chain: the rules deciding not which blocks are valid, but when the people who mined them may spend the reward.

What it trades at, and why the site shows no market cap

No large exchange lists the coin, and neither CoinGecko, CoinMarketCap nor CoinPaprika carries it. It trades under three tickers on small venues:

Volumes from the venues' own market APIs, where published, on 21 October 2026.

The price on 21 October was close to $500. The early figures that circulated — first trades near $80 on 2 September, an "all-time high" of $3,348 three days later — came from a single thin order book and from a price tracker's record, not from a market anyone could fill a real order in. The pro-fork site bitcoinbip110.org itself warned on 1 September that Neoxa, the first venue, was a high-risk, unverified custodial exchange, citing withdrawal problems and no proof of reserves. CoinSpectrum's asset page lists NonKYC and NestEx and leaves Neoxa out for that reason.

Multiply $500 by the 20.1 million coins on the chain and the result is a market cap near $10 billion — for a coin that trades under $2 million a day. The multiplication is the problem. Those 20.1 million are Bitcoin's whole ledger copied at the split, including coins lost for a decade and wallets whose owners have never heard of BIP-110. Nobody knows how many of them are held by anyone active on this chain, and no source publishes a circulating figure. The Bitcoin Blake2B page therefore shows its price, volume, hashrate and supply rules, and no market cap.

Why it counts as a new cryptocurrency

The ranking counts an asset as a cryptocurrency when it has its own blockchain and its own consensus. Bitcoin Blake2B passed the first test at 20:12 UTC on 8 August, when its first block was one no Bitcoin node would accept. It passed the second at the same moment: its nodes enforced rules Bitcoin's nodes do not, which is the same test that separates Bitcoin, Bitcoin Cash and Bitcoin SV — three chains on one algorithm. The switch to BLAKE2b at block 961,640 made the separation visible to anyone with a miner, but it was already complete.

The name is a separate question, and the counting rule does not answer it. The project calls its chain Bitcoin; the exchanges call it BTCB2, XBT or BIP110; LiveCoinWatch lists it as Bitcoin Blake2B. The site lists it under the name its markets use, as it listed Bitcoin Cash in 2017 without ruling on which branch inherited the name. That question is argued in Bitcoin took two upgrade paths in 2017, and the answer there — that a name is inherited by convention, not by protocol — applies here too.

What it says about activating soft forks

In 2017 Bitcoin went through a similar threat. BIP148, the user-activated soft fork for SegWit, also made signalling mandatory from a fixed date, 1 August 2017. It never split the chain: in July miners activated BIP91, which enforced SegWit signalling before the deadline arrived, in large part because exchanges and businesses had said which chain they would follow. BIP-110 had no such backing. No exchange announced support, and the mandatory clause took effect against 97% of the hashrate.

The comparison with the 2017 splits shows what a minority fork needs to survive. Bitcoin Cash, the same month, kept SHA-256 so it could borrow Bitcoin's miners, wrote replay protection into the fork, and had exchange support on day one. Bitcoin Gold, that October, changed its proof of work to Equihash so it could recruit different hardware. Bitcoin SV's 2018 hash war showed that a minority on a shared algorithm can be attacked by the majority's machines. Bitcoin Blake2B took the Bitcoin Gold route — after the fact. It changed algorithm three weeks after splitting rather than at the split, with replay protection still left to the user.

The lesson the episode leaves is a narrow one. A signalling threshold measures who will say yes. Mandatory enforcement tests who will mine, and the answer to the second question arrives at the first block that disagrees.

Not to be confused with eCash (ECX)

A second Bitcoin fork ran on a similar calendar. Paul Sztorc's eCash (ECX) is a planned hard fork that keeps SHA-256 and adds drivechains (BIP-300 and BIP-301), launched in test stages since August with mainnet planned for about block 973,728, near the end of October. It is a different project from BIP-110, and also different from the existing eCash (XEC), the renamed Bitcoin Cash ABC chain. Until its mainnet runs and a market trades it, there is nothing for the ranking to count.

Chain, market and pull-request figures are as of 21 October 2026. Prices and volumes move with every data build; the fork heights, dates and BIP-110 rules do not. The coinbase-maturity decision at block 979,920 was still open when this was written.

Sources

1. BIP-110 specification — rules, activation parameters, Closed status
2. bitcoin/bips #2248 — removal of Luke Dashjr as BIP editor
3. Bitcoin Knots #359 — the BLAKE2b proof-of-work hard fork
4. Bitcoin Knots v29.4.1 release and v29.4.2 release notes
5. Bitcoin Knots #419 — the 45-day coinbase maturity; open proposals #429, #467, #468, #469
6. mempool.guide — block heights, timestamps, pools and hashrate of the BLAKE2b chain
7. bitcoin-blake2b.org and bitcoinbip110.org — project sites, parameters and the Neoxa warning
8. Bitcoin Optech, 14 August 2026
9. The Block, 8 August 2026 — the split
10. TFTC — signalling against enforcement
11. Bitcoin Takeover — the algorithm draw and the pool disputes
12. ecash.com — eCash (ECX) launch stages