Let us assume a fork produces two blocks in eight hours. On the Bitcoin main chain, the expected production in that window is forty-eight blocks. The observed production was two. At block height 961,632, a cohort of node operators triggered what they framed as a user-activated soft fork — BIP-110, a proposal to restrict non-financial data in Bitcoin transactions — by rejecting every block that did not carry the activation signal. Eight hours later, the child chain stood at height 961,633. The parent chain had advanced to 961,681. The 48-block divergence reads less like a chain split and more like a starvation event.
The hash is not the art; it is merely the key. This particular key fitted no lock.
BIP-110's purpose was subtraction, not scaling. It aimed to constrain the script-data capacity that enables Ordinals inscriptions — the non-financial payloads committed through taproot witness paths. In the 2,016-block difficulty period before the split, only 51 blocks carried the BIP-110 signal. That is 2.53 percent. The proposal's own activation math demanded 55 percent, or 1,109 blocks out of every 2,016 produced. Between 2.53 and 55 lies the entire story: a technical mechanism, an economic veto, and a governance deadlock that will not end with this fork's quiet death.
This article reconstructs the event mechanically — the activation design, the block-level evidence, the miner incentive structure — and then argues an uncomfortable conclusion: the failure of the consensus-level fork makes the next assault on Ordinals more likely, not less, and it will not arrive as a fork at all.
Context: The Proposal and Its Threshold
BIP-110 belongs to a family of Bitcoin governance instruments designed to adjust consensus rules without forcing every participant to upgrade simultaneously. Soft forks tighten constraints; old nodes continue to see new blocks as valid, while upgraded nodes enforce additional rules. The BIP in question targeted the data payload carried in transaction witnesses — the same technical capacity that Ordinals engineers used to embed arbitrary content into Bitcoin's permanent history.
The activation mechanism was a hybrid, and the hybrid deserves close reading. It asked miners to signal readiness in the version field of their blocks: at least 1,109 blocks out of any 2,016-block period, roughly two weeks, carrying the signal. That 55-percent bar placed the proposal between two historical precedents. BIP 91, the SegWit2x coordination device of 2017, required 80 percent signaling before locking in its commitments. BIP 148, the user-activated soft fork that broke the SegWit stalemate, required no miner signaling at all — it instructed node operators to reject non-signaling blocks after a deadline, forcing miners to capitulate or split the chain. BIP-110's 55 percent was a middle path: high enough to claim legitimacy, low enough to be reachable. On paper, reasonable.
The substantive target was the inscription economy. Ordinals inscriptions write arbitrary content into witness data, often using key-path spending and integer-indexed payload chunking to distribute large blobs across multiple transactions. The result is a permanent historical record: content addressed by hash, retrievable through full-node history, economically attached to a Bitcoin UTXO. For one faction, this is the first digital-native medium with genuinely permanent custody. For another, it is block-space pollution monetized as a tax on inclusion. The conflict, as usual in this domain, is not aesthetic. It is financial. If inscriptions produce meaningful fee volume, then restricting them is a transfer from miners to nobody in particular. The miners lose revenue, the proposers gain a purified block space, and the market is expected to applaud the transaction.
The observed signal rate — 51 blocks, 2.53 percent — suggests the market declined the invitation.
I keep the published on-chain figures here because the event is small enough to verify directly. A two-block chain is not a moving target; it is a museum exhibit. Run your own node, inspect the chain at height 961,633, count the blocks produced in the first eight hours. The numbers hold. The measurements are the easy part. The interesting part is why the mechanism failed so completely, and what the failure tells us about the future of Bitcoin's data governance.
Core
I. Activation Design and the Arithmetic of Consent
Every soft-fork activation design encodes an assumption about miner behavior. BIP 148 encoded the assumption that miners would capitulate in the face of a node-enforced deadline. They did. By the activation cutoff, roughly 95 percent of blocks signaled SegWit readiness, and the chain did not split. The threat alone — that a portion of nodes would reject non-signaling blocks and start a divergent chain — was sufficient to move a supermajority of economic actors. BIP 91 encoded a differently structured assumption: that a high-threshold lock-in could force coordination among otherwise stranded parties. It worked briefly and messily. BIP-110 encoded the assumption that a comfortable 55-percent threshold, combined with a node-side rejection trigger, would create a bandwagon effect. Miners would watch the signal approach, anticipate rejection, and join in self-defense.
The bandwagon never left the station.
Fifty-one signaling blocks is not a campaign; it is a protest note. It computes to 4.6 percent of the required consensus — 51 of the 1,109 needed. In any monitoring dashboard, that number would trip a critical alert. In governance terms, it is a statement that the proposal never achieved a toehold among the participants who control block production. The threshold design became irrelevant because the signal floor was never approached. The proposal's authors calibrated the bar for a negotiation that no economically relevant party was willing to enter.
This exposes something important about activation thresholds: they are not neutral parameters. The choice between 55 percent and 80 percent is not a choice about strictness. It is a choice about which negotiation you are trying to win. BIP 91's 80 percent assumed miners wanted SegWit enough to coordinate. BIP-110's 55 percent assumed miners were undecided enough to be flipped by a credible threat. Both failed or succeeded not because of the number itself, but because of the underlying preference distribution among the economic base.
I have spent years inside this kind of parameter-design problem. During my audit work in 2017, I identified integer-overflow vulnerabilities in the Golem distribution contract's pledge logic and submitted a patch with a mathematical proof of exploitability. The response from the founders: too academic. The patch was correct; the persuasion failed. Technical correctness never guaranteed adoption. BIP-110 shares the pathology. It is formally coherent — the 55-percent bar is reachable, the rejection rule is executable, the fork is mechanically reproducible. None of that matters if the economic base declines to engage.
There is also an arbitrariness problem. Why 55? The proposal's published text offers no first-principles derivation. I have spent enough time with parameterized systems — the interest-rate slopes of Aave and Compound are the canonical example — to treat arbitrary parameters as political positions dressed in mathematical notation. The 55-percent threshold triangulated between BIP 91's 80 and BIP 148's zero. Triangulation is not design. It is polling.
II. The Eight Hours: What Two Blocks Actually Prove
The eight hours following height 961,632 are the empirical core of this event. Bitcoin's target block interval is ten minutes. Eight hours at target difficulty should yield 48 blocks. The main chain produced 49 during the same window — operating normally, well within variance. The fork chain produced two.
A two-block chain is not a slow chain. It is an almost-nonexistent chain. The worst-performing mining pool in Bitcoin's history, on its worst day, has produced more than two blocks in 480 minutes. Two blocks in eight hours implies that no economically meaningful miner pointed hashrate at the fork. The most plausible explanation is that the proposers themselves mined the two blocks, on whatever hardware was at hand, to establish that a chain existed at all.
We can call them alarm-clock blocks. The term comes from small-scale mining: a miner computes at negligible hashrate but claims blocks by being the only participant awake when one solves. On a fork with near-zero total hashrate, anyone with a laptop and a compiled node can claim blocks on schedule. Two blocks in eight hours is the signature of a hobbyist operation carrying a banner, not an industrial pool building a competitor.
The security implication is not so much weak as nonexistent. A chain whose hashrate produces one block every four hours can be reorged by any entity with a few rented GPUs. The 51-percent attack on this chain would be a 0.001-percent attack. But the sharper observation is that no one would bother. There are no assets on the fork. No exchange lists its token. No wallet registers its address format. No indexer tracks its state. The chain is not a parallel economy; it is a press release rendered in consensus code.
And yet — the two blocks are not a mistake. They are the point. The operators did not fork to create a persistent competitor; they forked to prove that the trigger mechanism works. Nodes can reject blocks. A minority can instantiate a chain. The message to the miner majority was: the threat of division is real. The response from the miner majority was: a chain without us is not a threat. Both statements are true, and the two-block chain is their collision.
This reminds me of my 2021 research into NFT metadata persistence, where I spent weeks analyzing IPFS pinning across major profile-picture projects and found that over 60 percent of supposedly permanent tokens depended on centralized gateways already failing under load. The conclusion then: infrastructure stability, not artistic value, is the bottleneck. The two-block chain is that finding in extremis. It is not infrastructure; it is a signal wrapped in a chain.
III. Miner Economics: The 2.53 Percent Signal as a Profit-and-Loss Statement
No amount of governance theory explains BIP-110's failure as cleanly as a fee table. Miners earn a fixed block subsidy, decaying by halving schedule, plus variable transaction fees. The fee stream is where inscriptions matter. Ordinals transactions are real transactions: they pay fees, fill blocks, and contribute to miner revenue. In high-activity periods, inscription traffic has meaningfully lifted the fee component of block rewards.
BIP-110 proposed to amputate that fee source. A cap on non-financial data writes would reduce the volume of inscription transactions, reduce fee competition, and reduce miner revenue. No compensating mechanism was proposed. No substitute fee stream was identified. The proposal was, from the miner's perspective, a pure cost with zero benefit.
The 2.53 percent signal is the mathematical residue of that accounting. Miners did not need to publish position papers. They simply mined without the signal, and the proposal expired. Version bits are a ledger of economic consent, and this ledger said: no deal.
The contrast with BIP 148 is instructive. In 2017, miners ultimately signaled SegWit at overwhelming rates because the alternative — the August 1 chain split — threatened a much larger cost: exchange chaos, user confusion, community fragmentation. Miners capitulated because capitulation was the least-bad equilibrium. The preference structure aligned with the mechanism. BIP-110 inverted that structure. Miners preferred the status quo; the node-side rejection was a nuisance, not an existential risk. The mechanism fired, and nothing moved.
My 2022 work on MakerDAO's liquidation engine taught me the same lesson from the opposite direction. Debt ceilings and liquidation parameters behaved one way in calm markets and catastrophically differently during liquidity crunches. The parameters were formally identical; the incentive distribution among participants was not. Bitcoin's soft-fork activation mechanisms behave the same way. A 55-percent threshold is only meaningful within a preference distribution that allows it to be reached. When the distribution is hostile, the threshold is decorative. BIP-110 decorated a wall that then fell on it.
There is a second, quieter economic observation. The mining pools themselves have said nothing publicly in defense of Ordinals. Their silence is not neutrality; it is a de facto veto. By declining to signal for BIP-110, they preserved inscription fee revenue without ever declaring allegiance to the inscribed content. This is the cleanest form of protection the Ordinals market could receive: economic protection with ideological deniability. The inscription market, which had reason to fear a rule-level crackdown, instead received an eight-hour demonstration that the crackdown cannot pass without the permission of the exact parties who profit from its failure.
IV. Governance: The Separation of Initiation and Execution
The BIP-110 episode is the cleanest case study I know of Bitcoin's structural separation between two governance functions. Rule initiation on one side: anyone can write a BIP; anyone can run a node with custom rules. This is the high-freedom half of Bitcoin governance, and BIP-110 demonstrated it fully. A small cohort enforced a rule at height 961,632, and a chain was instantiated. Under the user-activated soft fork model, the user is defined as the node operator, not the coin holder. That distinction has a long and contested history.
Rule execution on the other side: a chain persists only if economic actors allocate resources to it. The dominant resource is hashrate. The two-block chain had none. The rule was written, executed by a handful of nodes, and then abandoned by everyone else. BIP-110 separated the rights precisely. The initiators exercised their initiation right. The executors exercised their execution right by declining to participate.
This asymmetry is load-bearing. It makes consensus changes expensive to impose and expensive to sustain. A minority can always initiate; a minority cannot sustain. The high cost of maintaining a fork is what keeps Bitcoin's governance conservative. The BIP-110 team could create a chain for nearly nothing. They could not create security, liquidity, or ecosystem integration for anything less than an enormous economic commitment. The two-block chain is the price of entry, and the entry led nowhere.
But the asymmetry cuts both ways, and this is the part that should worry the Ordinals market. The 2.53 percent signal says: no active enforcement threat at the consensus level. It also says: the majority is indifferent to the survival of inscription data. The "pure money" faction lost the consensus battle but gained strategic intelligence. They now know a UASF does not work without miner buy-in, and a fork without economic weight is theater. Strategic actors who absorb that lesson do not repeat the same move. They change the vector.
V. The Data Semantics: Polluted Ledger or Permanent Library
Let me put the ideological conflict under a microscope. The hash is not the art; it is merely the key. I wrote that sentence years ago about NFT pointers — the tendency of the market to confuse the hash of an asset with the asset itself. It applies with painful precision to the BIP-110 argument.
Ordinals inscriptions commit to content by distributing it across witness data, splitting blobs into chunks referenced by script path. The commitment is permanent, but the realization depends on the full-node set retaining historical witness data and indexers reconstructing the payload from chain history. The BIP-110 faction considers this a tax on every future node operator: the cost of validating and storing irrelevant content forever. The Ordinals faction considers it the first digital-native medium with genuinely permanent custody. The miners, inhabiting neither ideology, consider it a fee schedule.
The governance deadlock is not a fight between good and bad actors. It is a trilemma. The purity faction wants the chain to remain a monetary settlement layer. The inscription faction wants the chain to be a permanent library. The miners want fees. The BIP-110 fork attempted to resolve the trilemma by consensus-level force and failed because one of the three parties held the weapon — the hash.
The weapon is the point. In a proof-of-work system, the ability to write history is inseparable from the willingness to spend energy. The purity faction's proposal attempted to break that coupling: nodes would write a rule, and the chain's history would obey. The coupling did not break. The rule became a chain with two blocks, and the history continued on the main chain as if nothing happened. The economic reality of proof of work is that history is written by hashers, not by theorists.
VI. Market and Narrative Aftermath
The market's response to BIP-110 is best measured in units of indifference. Bitcoin price moved, at most, in the noise band. The event is too small, too technical, and too lacking in exchange integration to command macro attention. The only segment likely to react meaningfully is the Ordinals complex itself — and here, the reaction is a relief rally of narrative rather than of price. The threat of consensus-level restriction has been repelled, for now. ORDI and its peers may print short-term bounces as the policy-risk reprieve is priced in. The magnitude is likely limited: single-digit percentage movements concentrated in the most liquid inscription-adjacent tokens, followed by resumption of the prior regime.
The larger distortion is the narrative one. Headlines that frame this as "Bitcoin splits" are technically true and operationally false. A minority fork without hashrate is a message, not a split. The word "split" should be reserved for events that fracture economic consensus — the Bitcoin Cash fork of 2017, for instance, or the theoretical August split that BIP 148 avoided. A two-block chain is a signal, and the signal is the event. The chain itself is the stamp on the envelope.
For the sector, the correct response is neither celebration nor panic. It is calibration. Fork attempts now cost nearly nothing; anyone can trigger a UASF with a modest budget and a few dozen synchronized nodes. The market should treat fork noise as an expected feature of the system, not an input to price. The events that matter are the ones where the signal rate approaches the threshold. This one, at 2.53 percent, never did.
Contrarian: The Failure That Was Not a Failure
Here is the uncomfortable argument: the comfortable conclusion — that BIP-110 failed, therefore Ordinals are safe, therefore Bitcoin's data governance is settled — is backwards on every count.
Failure is information. The anti-inscription faction now knows, with experimental confirmation, that the fork vector is a dead end without miner support. That knowledge is not discouragement. It is a design constraint for the next iteration. And the next iteration will not be a consensus change. It will be a policy change.
There is no rule that requires nodes to relay inscription-bearing transactions. A future Bitcoin Core release — or more plausibly, a widely deployed alternative node build — could implement a mempool policy that filters transactions with large non-financial witness payloads. Inscription-bearing transactions would still exist, still be minable by pools that opt in, but the default relay network would refuse to propagate them. The economic effect is a liquidity strangle: inscriptions become harder to discover, harder to transmit, harder to settle. No fork is required. No version-bit signal is required. No miner coordination is required beyond the miners' own choice to run the filter.
I have built exactly this kind of layer in a different context. In my 2026 work on AI-agent transaction interoperability, my team faced a problem where autonomous agents were signing irreversible transactions based on hallucinated instructions. We did not fork the ERC-20 standard; forking the standard would have been a coordination disaster. We built interface layers that filtered what the agents could see and sign, reducing failed transactions by 40 percent without touching consensus. The filtering approach worked. The fork approach would have failed, for the same reason BIP-110 failed: the economic base does not upgrade for your convenience.
The anti-inscription faction has this exact toolkit available. They do not need a BIP. They need a node configuration and a release schedule.
There is also the compromise vector, which I consider the most likely path of real change. A future proposal could restrict only "large" inscriptions — payloads beyond a defined threshold — while preserving small data writes. The threshold would satisfy the purity faction's demand for limits on bloat while preserving the fee stream that miners value. Miners might accept that, especially if the next halving tightens their revenue margin. The Ordinals ecosystem would adapt, fragmenting into "small inscription" culture and "large blob" exiles migrating to alternative layers. That is not a catastrophe for either side. It is a market clearing.
None of this was made less likely by the two-block chain. It was made more likely. The fork was the opening bid. The negotiation now shifts to a venue where the purity faction has structural advantages and the Ordinals market is not looking.
Takeaway: The Next 2,016 Blocks
Watch the next difficulty period. Signal rates below five percent mean the anti-inscription movement is still gestating. Rates climbing past twenty percent would indicate the compromise negotiation has begun — and the Ordinals market is not pricing that scenario at all.
The hash is not the art; it is merely the key. BIP-110's key fitted no lock. The next attempt may not need a lock at all. It may simply change the shape of the door.
The two-block chain posed no threat to Bitcoin. It posed a question to the Ordinals ecosystem: if the default node software decides one day that your transactions are not worth relaying, what will your permanence be worth then? Count the blocks before you answer.
