Two block Bitcoin fork shows the risk of enforcing rules too early

Bitcoin’s BIP-110 fork attempt produced two blocks over the weekend and then stalled. By Tuesday, the dominant Bitcoin chain had climbed to block 961,959, more than 300 blocks ahead of the minority branch, which has not moved since Saturday. The split began at block 961,632, when nodes enforcing BIP-110 rejected a block, mined by Antpool, that did not signal support for the proposal.

Miners had signalled support in only 51 of the previous 2,016 blocks. That’s 2.53% and far below the 55% needed even to lock the proposal in, let alone put it into effect. The fork did not change that: it only separated a minority of enforcing nodes from the rest of Bitcoin.

The dominant chain has continued normally. The enforcing branch has produced no block since Saturday. Its main source of mining power walked away over the weekend before signaling it might return. A separate dispute over the proposal’s editorial handling has already cost one of its authors his role as a Bitcoin Improvement Proposal editor.

What does it mean for a blockchain to fork?

Bitcoin’s ledger is a single chain of blocks. Each one records a batch of transactions and points back to the block before it. Every computer running Bitcoin software, called a node, checks new blocks against the same set of rules before accepting them. As long as every node agrees on the rules, there is only one chain.

A fork happens when two groups of nodes start applying different rules and end up disagreeing about which block belongs next at the same position in the chain. That is what happened at block 961,632. Antpool produced a block without a specific marker, called a signal, that BIP-110 requires. Nodes running ordinary Bitcoin software accepted that block, because it broke none of their rules. Nodes running BIP-110 software rejected it, because their rules required the marker, and instead built on a different block supplied by another miner, Roughnecks.

From that point on, two separate versions of Bitcoin’s history exist side by side. Both started from the same blocks up to 961,631 and diverge after. Miners can choose which version to keep extending, the same way they choose which valid block to build on any other day. That choice is not symmetric. Miners get paid in whichever chain’s coins they mine. But those coins are worth something only if someone accepts them in exchange for goods, services or other currencies. Almost every miner kept extending the chain that exchanges and merchants already recognize, which is why the BIP-110 branch attracted only one significant miner and lost it within a day.

>>> Read more: Bitcoin BIP-110 Proposal Divides Developers and Miners

What triggered the split at block 961,632?

BIP-110 seeks to temporarily restrict methods used to place images, text and other non-payment data inside Bitcoin transactions. Supporters argue that these limits would protect block space for monetary use. Critics say users should remain free to buy block space for any valid transaction.

The proposal uses a signalling system based on a field inside each mined block. Miners had a first chance to adopt BIP-110 voluntarily: if 55% of blocks in a single stretch carried the signal, the rule would lock in early. That never came close to happening.

The mandatory-signalling window that followed, running from block 961,632 through 963,647, was not a fallback improvised after voluntary signaling failed. It was the plan from the start. Nodes running BIP-110 software would simply reject any block without the signal, forcing the issue instead of continuing to just count how many miners agreed. The same approach forced SegWit through in 2017. But that only worked because most of the Bitcoin economy, exchanges, businesses, ordinary node operators, was already on board before enforcement began. Miners had little choice but to comply. BIP-110 reached its enforcement date without anything close to that backing. It split off the small group enforcing it; it did not bring miners into line.

Antpool’s block at 961,632 carried no signal, so BIP-110 nodes rejected it and followed a rival block instead, built by Roughnecks using Ocean’s DATUM system, which lets individual miners assemble their own block templates and choose whether to signal.

Why did the minority chain stall after two blocks?

The BIP-110 minority chain added blocks 961,632 and 961,633, both mined by Roughnecks through Ocean’s DATUM system. No third block had appeared by Tuesday. Bitcoin’s dominant chain, meanwhile, reached block 961,959 in the same stretch, a gap of more than 300 blocks.

Finding a valid block requires solving a computational puzzle. Bitcoin sets how hard that puzzle is, called mining difficulty, based on how much total computing power the network has. That difficulty only resets every 2,016 blocks. The BIP-110 branch split away carrying the same difficulty level Bitcoin had at full network power, then lost more than 99% of that power within a day. It is now trying to solve the same puzzle with a small fraction of the computers, so its blocks arrive far slower than the usual ten minutes.

That imbalance also determines when the branch could catch a break. A live tracker of the minority chain put its path to an easier difficulty target at roughly 6.3 years away as of Monday, up from an estimate of 350 days just one day earlier. The estimate is recalculated from recent block production, so every hour without a new block pushes it further out. Michael Saylor, Strategy’s chairman, offered his own estimate on Aug. 9, putting the branch’s hashpower at about 0.15% of Bitcoin’s total and its path to a difficulty adjustment at roughly 25 years. The two figures come from different sources and different moments, but they agree on the direction: the longer the chain sits idle, the more distant its own fix becomes.

Is anyone trying to revive it?

Roughnecks halted its mining operation on Sunday and told other miners to stand down, a decision that coincided with Ocean’s displayed mining output falling by roughly 96.5% over the weekend. The retreat was brief. Roughnecks has since signalled it may resume mining the branch. BIP-110 supporters are now discussing an even more drastic option: changing the branch’s underlying computational puzzle to separate it from Bitcoin’s existing specialized mining hardware entirely. However, that would require new or repurposed equipment and would restart the branch’s computing power from zero.

The episode also complicated the idea that Ocean backs BIP-110 as a pool. Simple Mining, which also mines through Ocean, produced a non-signalling block on the dominant chain days after the split and said the decision not to follow BIP-110 was deliberate. Ocean’s DATUM system leaves that choice to individual miners; the pool itself takes no position.

Why is a Bitcoin Improvement Proposal editor now out of a job?

BIP-110’s stall was not the only fallout. On Aug. 9, Bitcoin developer Mark Erhardt, known as Murch and one of the volunteer editors who maintain the official Bitcoin Improvement Proposal repository, recommended removing fellow editor Luke Dashjr from that role. Erhardt alleged Dashjr had used his editorial authority to favor BIP-110, a proposal Dashjr helped draft, including trying to assign it a BIP number before it had been discussed on the developer mailing list and merging a related change within minutes of it being opened.

Dashjr rejected the allegations and said Erhardt should be the one removed instead. Other editors and Bitcoin Core contributors, including Olaoluwa Osuntokun and Matt Corallo, backed Erhardt’s motion. On Aug. 10, editor Jon Atack confirmed Dashjr no longer held administrative access to the BIPs repository, and the change was merged into the project’s editor list. Dashjr called the move an abuse of power.

BIP editors hold an administrative role: assigning numbers to proposals and checking that they followed the required process. It is not a technical role. The removal is a governance fight over whether BIP-110 received fair procedural treatment on its way to the fork that split Bitcoin’s chain.

Is this really over?

Not everyone treats the outcome as settled. Himanshu Sahay, co-founder of the Bitcoin infrastructure firm Arch, said it remains too early to call the branch a failure from two blocks of data alone, since changes to Bitcoin’s rules depend on coordination across miners, developers and the wider ecosystem that can shift over weeks, not days.

The proposal itself still has room to run. The specification allows BIP-110 to lock in as late as block 963,648, with its restrictions taking effect one difficulty period after, at block 965,664. What has not happened, so far, is any sign of the miner support that would get it there. Bitcoin Core has not endorsed the proposal. No major exchange has listed the minority branch’s coins, which is what actually keeps them from functioning as money regardless of how many blocks the branch produces.

What does this mean for ordinary holders?

Most Bitcoin holders have nothing to actively track here. Their coins and transaction history stay on the chain that exchanges, wallets and payment processors already treat as real. The people who do need to pay attention are node operators enforcing BIP-110, whose software is now stuck on the slower branch, and anyone who tries to move Bitcoin across both chains at once.

That second group faces a “replay” risk. A transaction signed to spend coins that existed before block 961,632 is valid on both chains, since neither has changed the ownership since the split. If that signed transaction is deliberately resubmitted to BIP-110 nodes as well as the main network, it moves the same coins on both, which matters to anyone trying to sell or trade the two histories separately. It does not affect people who simply spend from a normal wallet through the normal network, and the branch is not currently mining any blocks that could include a replayed transaction regardless.

Holding coins from before the split does give a technical claim on both chains: the same private keys control matching balances on each. That resembles the way past Bitcoin forks such as Bitcoin Cash in 2017 also handed existing holders a claim on a new coin. But that doesn’t mean that claim is worth anything at this point. With no exchange listing and no active mining on the branch, there is currently nowhere to spend or trade the BIP-110 side of it.

Whether Roughnecks resumes mining the branch, whether a change to its computational puzzle attracts enough hardware to matter, and whether the editorial dispute settles the question of how BIP-110 reached this point at all, are the threads still open as the gap between the two chains keeps widening.