BIP-110: The 185-Block Countdown to a Non-Consensus Split

On August 7, 2025, Bitcoin was 185 blocks from a rule change that almost no one asked for. At height 961,632, Bitcoin Knots will treat any block without version bit 4 as invalid. The current signal rate is 2.62%. That is not a rounding error. That is the absence of consensus. BIP-110 is being deployed by a minority implementation with a mandatory activation clock. This is not a soft fork in the historical sense. It is a non-consensus rule set, embedded in the code of one client, about to be applied to a network that has not agreed to it. Read the code, not the pitch deck. The pitch deck says: smaller proofs, faster SPV verification, compressed block data. The code says: two definitions of a valid block. I have audited enough consensus state machines to know which one you should trust. The Context: What BIP-110 Claims BIP-110 is not a new idea. It emerged from the same scaling debates that produced the SPV proof-compression proposals of 2015-2017. The goal is straightforward: reduce the amount of data an SPV client needs to verify that a transaction is included in a block. In principle, that makes lightweight wallets faster and cheaper. That is a legitimate engineering objective. There is no serious argument that Merkle path data cannot be improved. But there is a serious argument about how to deploy it. BIP-110 chose a soft-fork path. A soft fork is supposed to tighten validation without breaking old clients. Old nodes continue to see new blocks as valid because the new rules are stricter. BIP-110 fails this test in a subtle way. The enforcing nodes do not simply add a stricter validity rule for Merkle data. They reject blocks that do not signal version bit 4 during a forced period. This is not a validation rule about the block's content. It is a rule about a bit in the block header. Old nodes never cared about that bit. They still do not care. So the same block is valid under Bitcoin Core and invalid under Bitcoin Knots. That is the difference between a soft fork and a rule schism. The closest historical analog is BIP-34, which also used version bits and forced signaling. But BIP-34 arrived with overwhelming miner support. It was a coordination mechanism for a network that had already agreed. BIP-110 is the opposite. It is a coordination mechanism for a network that has not agreed and, as far as Bitcoin Core is concerned, will never agree. The proposal is not controversial because of its Merkle-path math. It is controversial because of the legitimacy of its activation. The Activation Mechanics Now the numbers. Standard BIP-9 activation requires 95% hashpower over 2,016 blocks. BIP-110's stated threshold is 55%, or 1,109 blocks. On August 7, 2025, the observed signal rate was 2.62%. The threshold has not been met. It will not be met. Yet the mandatory signaling period is already scheduled. At 961,632, enforcing nodes start rejecting non-signaling blocks. At 965,664, if no lock-in occurred, the data-reduction rules are still scheduled to activate. The threshold is not a gate. It is a formality. The activation is a function of block height, not consent. The 55% threshold itself should concern anyone who has studied BIP-9's history. Even 95% is not always enough to prevent controversy. The 2015-2017 scaling war showed that a smooth soft fork requires more than a technical majority; it requires social alignment. BIP-110 has no such alignment. Bitcoin Core closed the implementation PR in March 2025. Antoine Poinsot, a Core contributor, said in June that Core will not enforce the proposal. The two major node implementations are on opposite sides of a rule. In every previous consensus change, the major implementation eventually agreed. This time, the major implementation is the non-enforcing side. The public signal rate comes from version-bit monitors such as mempool.space. That data is useful, but it has a hidden dependency: the monitor is only as neutral as the node it queries. If the monitoring infrastructure runs a Knots endpoint, the displayed signal rate may reflect a rule-biased view. I do not have access to that configuration, and the source article does not state it. But the possibility is structurally important. In a consensus dispute, the data pipe is part of the battlefield. The Divergence Now let's walk the worst case mechanically. A miner produces a valid block at height 961,633 without setting bit 4. Bitcoin Core accepts it. Bitcoin Knots rejects it. The next block is produced by another non-signaling miner. Core extends. Knots stalls. If OCEAN mines a block setting bit 4, Knots builds on that. If OCEAN mines a continuous string of bit-4 blocks, Knots follows its own chain. The network now has two candidates for the same height. The overwhelming majority of hashrate and Core nodes follow the non-signaling chain. The Knots/OCEAN chain is isolated. But isolation does not mean absence. Every exchange that uses a Knots-based API will see the isolated chain as valid. Every block explorer that queries a Knots node will display a different confirmation count. That is where user funds get stuck. Version bits were designed as a communication channel, not a law. The fact that a node can reject a block based on a bit that was never intended to carry consensus weight is a design flaw. BIP-110 weaponizes that flaw. It takes a signaling mechanism and turns it into a validity rule. The block is not being rejected because of bad data. It is being rejected because of a binary flag. That is not a soft fork. It is a protocol-level veto. OCEAN is not an accidental actor. OCEAN switched its default mining endpoint to signal BIP-110 on July 15, 2025, almost a month before the countdown began. That is not a miner voting with its feet after careful consideration. That is an implementation plan. OCEAN controls a small slice of global hashrate, estimated at 1-2%, but in a consensus dispute, hashrate is not the only relevant quantity. A pool that is willing to produce blocks on an alternate chain can force all enforcing nodes to follow. The resulting chain may be tiny, but it is a chain. The existence of a second chain alone is enough to disrupt exchanges, custody providers, and indexers that rely on Knots. The Hidden State Machine Now the part I have been waiting to discuss: the upgrade state machine. Complexity hides the body. The risk is not merely that two chains exist. It is that one node can switch between them without knowing it. Bitcoin Knots warned on August 7, 2025, that old non-enforcing software, explicitly including Bitcoin Core, may leave an unsafe chain state. That is a bombshell phrased as a footnote. An unsafe chain state means the node's view of history no longer matches its validation rules. BlockSlop, a technical reviewer, reproduced the exact failure in regtest. When moving from an enforcing Knots node to a non-enforcing node, the data directory still contained blocks accepted under the old, stricter rule set. On startup, the node did not immediately reattach inherited history. For a window, it was running with a rule set that contradicted the data it had already accepted. Knots has since merged a fix that scans inherited headers for forced-signal violations and reorgs them out. But the fix is incomplete by design. Header-level violations can be caught by scanning headers. Transaction-level and script-level violations cannot. Those require full reconnection of the UTXO set and, in many cases, a reindex. In my audit career, I have never met a reindex that did not produce at least one support ticket. The question is whether this time the support ticket is a withdrawal request. Let me be explicit about what I have seen in the field. I have audited more than a dozen Bitcoin-related node implementations, and the pattern is consistent. The most dangerous moment in any upgrade is not the new rule. It is the transition path between the old rule and the new rule. BIP-110's defenders focus on the Merkle verification savings. They ignore the state-machine residue. That residue is where the next exploit lives. A chain split is visible. A node that silently inherits blocks from the wrong rule state is invisible. The Market and Token-Value Question BIP-110 does not change Bitcoin's token supply. No new coins. No reward changes. No halving shift. The economic question is entirely about chain identity. In the event of a split, every BTC holder owns one coin on each chain. The minority chain will initially trade at near zero. It will have no exchange listings, no wallet integrations, no stablecoin pairs, no derivatives market. The majority chain, with 97.38% of hashrate and almost all Core nodes, is the natural economic winner. That is the BCH arithmetic. But the BCH split was deliberate. This split, if it happens, is accidental. And an accidental split does more damage to confidence than a deliberate one. The market has not priced this. Signal support is 2.62%, and most market participants have never heard of BIP-110. Tail events are not priced until they are no longer tail events. If OCEAN produces a visible alternate chain, short-term volatility may rise 2-5%. If nothing visible occurs, the event disappears from the terminal. But disappearance is not resolution. The precedent will remain in the code. In a bear market, the question is not how much you gain. It is whether your custody provider still sees the same chain you see. A rule divergence does not need to trigger a full split to cause damage. It only needs to trigger one exchange-wide confirmation delay. That delay will be called a bug. It will not be called what it is: two consensus rules competing inside one network. The Ecosystem Blind Spot Now let's look at the ecosystem. The downstream users are more exposed than miners. Miners simply continue mining the existing rules. Core node operators are unaffected. The exposure belongs to any wallet, explorer, exchange, or custody provider that runs Bitcoin Knots or depends on a Knots endpoint. Historically, Knots has been used by a minority of privacy-focused and independently minded operators. The source article does not list the affected services, which is itself a problem. I can think of at least one class of services that should be asking this question right now: any custodian that advertises multi-signature or self-custody infrastructure. Those systems are not always Core-based. A custody stack that uses Knots for UTXO indexing will show inaccurate balances after 961,632 if the pool signal does not match. That is not a theoretical concern. That is a settlement failure waiting for a block height. The Contrarian Case I do not join the easy chorus that BIP-110 is an attack. The bulls who call it an attack are missing the deeper structural lesson. Bitcoin Core's conservatism is not pure virtue. It has repeatedly delayed legitimate improvements, and it has allowed a culture where the default implementation's opinion is treated as law. That is an implementation monoculture. BIP-110 is not the cause of that monoculture. It is a symptom. Knots is the rebel faction that says the default is not fast enough and not strict enough. Some of its proposals are reckless. But the question of who gets to change Bitcoin's rules has no process. BIP-9 is a mechanism, not a social contract. BIP-110 exposes the absence of a social contract. The bulls also got one fact right: OCEAN and Knots have been more transparent than most upgrade attempts. They published warnings. They documented upgrade risks. They merged a fix after BlockSlop's reproduction. That level of disclosure is rare in crypto. But transparency is not legitimacy. You cannot create a consensus rule without consensus and then call it an improvement. The process flaw is the story. The technical patch is just a patch on a political wound. The Regulatory Shadow There is a wider, quieter risk. If a node implementation divergence causes exchange delays or temporary balance errors, regulators will ask why a supposedly decentralized network has no authoritative answer. That question is more dangerous than any price spike. The entire institutional narrative around Bitcoin is settlement finality. BIP-110 turns finality into a function of which node software your counterparty runs. Institutional investors entered after spot ETF approval. They rely on custodians. Custodians rely on node implementations. BIP-110 is not a retail event. It is a custody event. In a bear market, regulators are already suspicious. This is the kind of event they use to justify custody rules. The Accountability Call At block 961,632, the countdown ends. But the accountability begins now. If you hold Bitcoin, ask your custodian which node implementation it runs. Ask your exchange whether its indexer uses Core or Knots. Ask your wallet provider whether it tracks version bits. If they say it does not matter, they have not understood the event. Read the code, not the pitch deck. Complexity hides the body. The 185 blocks were a warning. The real test is whether the next 185,000 blocks see one chain, or two definitions of one chain.