bitcoin

BIP-110: The Fork Bitcoin Left Behind

BIP-110 tried to force new Bitcoin rules with just 2.53 per cent miner support. Instead, its nodes followed a barely moving minority chain, exposing the real limits of user-activated governance.

By CipherBot 09 August 2026 9 min read

A proposal to restrict inscriptions entered mandatory signalling with just 2.53 per cent miner support. Instead of forcing the Bitcoin network to comply, its enforcing nodes followed a barely moving minority branch. The result exposes what a UASF can, and cannot, actually do.

On Saturday, 8 August, a small group of Bitcoin nodes reached the block height at which their software was programmed to stop negotiating.

At block 961,632, nodes enforcing Bitcoin Improvement Proposal 110 began rejecting every block that did not signal support using version bit 4. The rest of the network continued operating under Bitcoin’s existing consensus rules. AntPool mined the first non-signalling block at that height and the dominant chain accepted it. BIP-110 nodes did not.

A separate BIP-110-compatible block was eventually mined at the same height, followed by another at 961,633. While Bitcoin continued producing blocks at close to its normal pace, the restricted-data branch fell rapidly behind. By the time the main chain had reached 961,659, the minority branch remained at 961,633, a gap of 26 blocks.

That is a chain split, technically. It is not Bitcoin dividing into two comparable networks. It is a small set of nodes enforcing rules that almost none of the network’s mining power has chosen to support.

BIP-110 has demonstrated that anyone can fork themselves away from Bitcoin. It has not demonstrated that they can take Bitcoin with them.

What BIP-110 is trying to do

BIP-110, formally called the Reduced Data Temporary Softfork, was authored by the pseudonymous developer Dathon Ohm, with Luke Dashjr credited for the original draft and advice. Its stated aim is to push Bitcoin back towards its primary function as sound, permissionless money by making arbitrary data storage more difficult at the consensus level.

The proposal responds to inscriptions, Ordinals, Runes and other techniques that use Bitcoin transactions to publish images, tokens and other non-monetary data. Supporters argue that these uses consume scarce block space, raise costs for people trying to transact, increase the burden on node operators and gradually turn permanent Bitcoin storage into a service for which miners collect the fees while every node carries the long-term cost.

For roughly one year, BIP-110 would add seven consensus rules. Among them, most new output scripts would be capped at 34 bytes, OP_RETURN outputs at 83 bytes and individual data pushes at 256 bytes. It would also temporarily disable several currently unused or lightly used Taproot features, including annexes, undefined witness versions and OP_SUCCESS opcodes, while restricting Taproot control-block size and certain uses of OP_IF and OP_NOTIF.

Coins created before activation are grandfathered. The authors have tried to preserve known monetary use cases, but the proposal itself acknowledges that some experimental Taproot arrangements could theoretically have funds frozen or lost during the temporary deployment.

The case for BIP-110 is not ridiculous. Bitcoin’s fee market pays the miner who includes data once, while thousands of nodes may be expected to download and retain it indefinitely. There is a genuine argument about whether an open monetary settlement layer should become an immortal file server simply because somebody can afford the block-space fee.

But BIP-110 does more than reduce node costs. It takes a disputed judgement about legitimate use and writes it into consensus. Bitcoin cannot tell whether a sequence of bytes represents a JPEG, a financial proof, a contract, an authentication record or a future protocol nobody has built yet. The rules can identify structures and sizes. They cannot identify intent.

That is why Strategy’s published case against the proposal argues that BIP-110 elevates a contested preference into protocol law. The question is not whether inscriptions are useful or obnoxious. The question is whether dislike of a use case is sufficient reason to make blocks containing it invalid.

The 2.53 per cent problem

BIP-110 offered miners an early route to lock-in if 1,109 of the 2,016 blocks in a difficulty period signalled with bit 4, a threshold of 55 per cent. Standard BIP9 deployments have historically used 95 per cent. BIP-110 lowered that bar because its authors considered the intervention urgent and temporary.

It did not get close to either figure.

Only 51 of the 2,016 blocks preceding the mandatory period signalled support, equal to 2.53 per cent. The activation client was nevertheless programmed to begin mandatory signalling at block 961,632. From that point, its nodes treated every non-signalling block as invalid, irrespective of the amount of work behind it.

This is the User Activated Soft Fork strategy. Its underlying claim is important: miners do not possess unilateral authority to decide Bitcoin’s rules. Full nodes independently verify blocks, and no amount of hashpower can force a node to accept something its software considers invalid.

That principle is true, but it is often stretched into something that is not.

A non-mining node can refuse a chain. It cannot add proof of work to the alternative. It cannot make an exchange recognise the alternative asset, make merchants price goods in it or make holders assign it value. A UASF gains force when a sufficiently large economic constituency credibly commits to rejecting non-compliant blocks. Miners then face the prospect of spending real energy to produce an asset the market may not recognise as Bitcoin.

Without that economic weight, mandatory enforcement is not a command to miners. It is an instruction for participating nodes to leave.

That is what happened at block 961,632. The dominant chain did not break its existing rules. BIP-110 nodes adopted stricter rules and rejected it. With little compatible hashpower, they were left watching a much slower branch.

This is not 2017

BIP-110 supporters have invoked the 2017 SegWit struggle, when the threat of the BIP148 user-activated soft fork helped break a prolonged activation deadlock. The comparison flatters BIP-110 by stripping away the conditions that made the earlier threat credible.

SegWit had years of development, extensive industry integration and broad support across wallets, businesses, exchanges and users. Miner signalling had become entangled with a wider political fight over scaling. The UASF mattered because it concentrated an existing economic constituency into a credible deadline. Miners were not simply frightened by node software. They were confronted with the possibility that valuable parts of the Bitcoin economy would reject the blocks they produced.

BIP-110 arrived at its deadline with 2.53 per cent signalling, low node adoption, no evident commitment from major exchanges to designate its branch as BTC and no comparable coalition of economic actors prepared to price the restricted chain as Bitcoin.

The lesson from 2017 was never that any group of users could write a flag day into software and thereby govern the network. The lesson was that coordinated users can influence miners when those users represent enough economic demand to make rejection costly.

Code can formalise a confrontation. It cannot supply the constituency.

Why block 963,648 is being misunderstood

Much of the reporting now points to block 963,648 as the next decisive moment, when BIP-110 is scheduled to enter its locked-in state. That description needs care.

There is no universal block 963,648 floating above both chains. Once two branches diverge, each has its own block at every subsequent height. BIP-110-enforcing nodes have already rejected the ancestry of the dominant chain. The main Bitcoin chain reaching height 963,648 does not cause those nodes to accept it, nor does it lock BIP-110 into the chain followed by Bitcoin Core, exchanges and the overwhelming majority of hashpower.

Under the proposal’s state machine, the BIP-110 branch is programmed to enter LOCKED_IN when that branch reaches 963,648, then become ACTIVE at 965,664. If its hashpower remains tiny, reaching those heights could take far longer than the normal ten-minute schedule. Unless substantial mining power migrates to it, the event will describe the internal state of a minority network, not an upgrade to Bitcoin.

This is the weakness in saying BIP-110 has “activated”. Mandatory signalling has begun in the software that implements it. The new restricted-data consensus rules have not activated, and BIP-110 has not become the consensus of the dominant Bitcoin chain.

The chain with accumulated work, economic acceptance and market recognition has continued without it.

The proof-of-work threat changes the argument

Some supporters have prepared for this outcome. Developer Chris Guida has reportedly rebased Luke Dashjr’s 2017 proof-of-work change code onto a recent version of Bitcoin Knots as an emergency fallback. The idea is that if existing SHA-256 miners refuse to support BIP-110, its backers could change the mining algorithm and replace the incumbent mining network altogether.

They are entitled to release such software. Open-source systems allow anyone to fork the code, alter the rules and try to persuade others to follow. But a proof-of-work change would remove any remaining ambiguity about what was taking place. It would be a hard fork launching a new network, with new mining hardware economics, a new security base and a new claim to the Bitcoin name.

It would also expose a strange contradiction. BIP-110 is presented as a defence of Bitcoin’s intended monetary neutrality against unwanted users and uncooperative miners. Yet its fallback is to discard Bitcoin’s existing proof-of-work infrastructure when that infrastructure refuses to ratify the proposal.

There may be circumstances in which a proof-of-work change is justified, such as an existential failure of the mining algorithm or a durable hostile majority. Disagreement over inscriptions, after 97.47 per cent of recent blocks declined to signal for the proposed restriction, is a much harder case to make.

At that point, “users control Bitcoin” becomes circular. The users are defined as the people running the preferred rules, and everyone else, including almost the entire mining network and the economic chain it secures, is reclassified as an attacker.

That is not decentralised consensus. It is a faction declaring itself the whole.

What the split reveals about Bitcoin

Bitcoin is not governed by miners alone. It is not governed by Bitcoin Core developers, and it is not governed by a count of reachable nodes. Each group holds a different kind of power.

Developers write and review the software through which rules are expressed. Node operators choose which software and rules they will accept. Miners order transactions and commit energy to a chain. Exchanges, custodians, wallets, merchants and holders decide which chain has liquidity, infrastructure and the BTC ticker. None can safely dictate to all the others for long.

Bitcoin’s resistance to capture comes from the difficulty of coordinating those groups behind a hostile change. Its governance is deliberately awkward because a system that can be upgraded easily can also be captured easily.

That does not mean Bitcoin is frozen or free from human power. The dominant Bitcoin Core implementation remains an influential coordination point. Mining pools remain concentrated enough to filter transactions in practice. Exchanges can decide which asset markets recognise. Wealth and institutional access shape whose preferences carry weight. “Code is law” does not eliminate politics. It moves politics into software distribution, economic coordination and the choice of which chain people are prepared to value.

BIP-110 is a useful demonstration of that reality. Its nodes exercised their sovereign right to reject blocks. The rest of the network exercised the same right to ignore them.

What holders should do

For ordinary BTC holders, the dominant Bitcoin chain has continued normally. The more immediate danger lies in attempts to trade or move coins on the minority branch.

Because balances and transaction history were copied at the split, coins may exist on both branches. Without robust replay protection, a transaction intended for one chain may be valid on the other. A holder trying to sell a speculative BIP-110 fork coin could accidentally expose or move real BTC. Unless a wallet or exchange has implemented reliable chain separation, interacting with the minority asset is an unnecessary risk.

The episode should not be inflated into a crisis for Bitcoin. A branch with negligible hashpower, limited infrastructure and no meaningful market recognition is not an equal rival to BTC merely because it shares Bitcoin’s history up to block 961,631.

The verdict

BIP-110 begins with a legitimate concern and answers it with an illegitimate level of certainty.

Inscriptions impose costs that the fee-paying user does not fully bear. Permanent arbitrary data may be a poor use of a global monetary ledger. Node affordability is part of decentralisation, and Bitcoin should be cautious about allowing fashionable applications to shape its base layer.

None of that proves that a small faction should impose new consensus restrictions with 2.53 per cent miner signalling, then threaten to replace the proof-of-work system when the network declines to follow.

The mandatory-signalling window was meant to show that miners serve users rather than rule them. Instead, it has shown that “the users” are not a coherent sovereign and that running code is not the same as commanding a market.

BIP-110 may continue producing blocks. Its backers may change proof of work, issue a separate asset and build an economy around it. That would be a fork in the honest sense: a new network competing for people, security and value.

What they cannot honestly claim, on the evidence so far, is that Bitcoin activated BIP-110.

Bitcoin kept moving. They left it.