At 3:47 on an unremarkable Tuesday in the autumn of 2017, a tiny fraction of Bitcoin's node network executed an upgrade that no one had voted for and almost no one was prepared to accommodate. The upgrade carried a simple, ruthless logic: from that moment forward, these nodes would reject any block that did not signal support for a specific proposal. And when the wider network refused to bend to their demand, they walked. Not with a manifesto. Not with a press tour. Just a quiet bifurcation — a small cluster of clients diverging from the main chain along a seam that most market participants did not even know existed.
The event carried an identifier that history has reduced to a footnote: BIP-110.
I remember that autumn with the clarity that hindsight sharpens into a blade. I was in Zurich, forty pages deep into a technical autopsy of a privacy coin whose ZK-Snarks claims I had been hired to verify. The paper — fifteen pages when I finally published it — argued that the project's anonymity guarantees were compromised by transaction graph analysis. It went modestly viral among the crypto-native set, and the founding team, in a gesture that still amuses me, offered me an advisory seat rather than a rebuttal. But while I was hunting for ghosts in the transaction graph, the market was hunting for something far larger: the answer to whether Bitcoin could disagree with itself without dying. The Block Size War was not background noise in 2017. It was the epistemological weather. Every chart, every token sale, every institutional thesis was drafted under its influence.
BIP-110 was a footnote in that war. But footnotes, in a decentralized system, have a tendency to become battlefields.
Over the eight years since, the industry has told itself a remarkably comfortable story about 2017: the forks happened, Bitcoin survived, network effects won. The story is true. It is also dangerously incomplete. The episode that mattered most was not the fork that happened. It was the fork that was threatened and then never came. And the real lesson of BIP-110 is not that decentralization survived its test; it is that the test itself was the load-bearing wall, and we have spent the years since quietly dismantling it.
The War the Market Forgot
To understand BIP-110, you must first understand the stalemate that produced it. By 2015, Bitcoin had a scaling problem. The one-megabyte block size cap, introduced originally as a spam measure, had become an ideological partition. One camp argued for larger blocks: a simple, intuitive fix that would let the network process more transactions per second. The other camp argued for Segregated Witness, a more elegant restructuring that increased capacity while preserving the precious cap and, more importantly, preserving the protocol's immutability guarantees.
For two years, the debate was conducted through every channel a decentralized organization possesses: mailing lists, conferences, Twitter flame wars, and, most consequentially, code. Bitcoin Improvement Proposals were the ammunition. BIP 101 proposed eight-megabyte blocks. BIP 141 proposed SegWit. BIP 148 proposed a User Activated Soft Fork, under which nodes would simply reject blocks that did not signal readiness for SegWit after a certain date — regardless of what the miners preferred.
And then, into this alphabet soup of proposals, came BIP-110. Its essential gesture was coercive signaling. Nodes running BIP-110-compliant clients would refuse to validate or propagate blocks that did not carry a signal bit for a targeted upgrade. In practice, this was a threat expressed as protocol logic. It said: signal, or we will treat you as if you do not exist. It transformed the node's passive acceptance of the network's rules into an active instrument of political compulsion.
This moment forced a distinction that most market commentary still collapses: the difference between a soft fork and a hard fork. A soft fork tightens the rules so that old nodes still accept new blocks, even if they do not fully understand them; a hard fork loosens or changes the rules so that old nodes and new nodes produce incompatible ledgers. Rejecting non-signaling blocks falls into an ambiguous middle: soft in form — the chain continues to operate if everyone simply signals — but hard in effect, because any miner that refuses to signal produces blocks the new nodes will not accept, and the ledger splits. That ambiguity was not a bug in BIP-110. It was the design. The proposal wanted to make compliance mandatory while preserving the fiction that no hard fork was being threatened.
The New York Agreement of May 2017 — SegWit2x, a compromise that would activate SegWit and then follow with a two-megabyte hard fork — had nominally unified miners and a cluster of influential companies. But the agreement was inked on paper, not in code, and code was where legitimacy actually lived. BIP-110 was the counter-move: a threat by the node class to render the agreement unenforceable by refusing to propagate blocks that signaled for it. It was, to use the precise technical term, a veto.
And the veto, in Bitcoin's design, is the deepest power there is.
The Signal and the Threat
Bitcoin does not have a government. It has a signal-detection mechanism. Miners register support for proposals by setting a version bit in the blocks they produce; nodes tally these signals over difficulty periods; when a threshold — typically ninety-five percent — is reached, the upgrade locks in. This is BIP 9, the activation mechanism that made SegWit possible. It is elegant, transparent, and entirely capable of being gamed.
The flaw is structural: it gives miners the microphone while giving nodes the mute button. In the early years, this distinction barely mattered, because miners and nodes shared an interest in the network's survival. By 2017, the interests had diverged. Miners carried sunk costs in hardware and electricity contracts that demanded specific fee regimes; nodes carried ideological commitments that no revenue stream could bribe. When the New York Agreement attempted to force a synthesis — SegWit now, larger blocks later — the node class responded by refusing the premise. BIP-148 and BIP-110 were the twin expressions of that refusal. BIP-148 said nodes would ignore miner signals entirely. BIP-110 said nodes would punish miners for sending the wrong ones.
The distinction is worth pausing on, because it maps onto a difference the industry routinely confuses. Signal is not consent. A miner who sets a version bit in a block is expressing preference, not binding anyone. The consent layer lives elsewhere — in the nodes that decide which chain to treat as authoritative, in the exchanges that decide which coins carry liquidity, in the developers who decide which patches to maintain. BIP-110's novelty was the attempt to weaponize that separation. By refusing to accept non-signaling blocks, the node makes its interpretive authority visible. It stops being an audience and becomes a participant with a veto.
In any other governance system, that veto would be described as constitutional review. In crypto, it was described as a fork threat — which is precisely the point. The system had no codified procedure for resolving a dispute between the miner class and the node class, so the procedure had to be invented on the fly, in code, under market pressure. Consensus is not a moment; it is a siege.
A Coordination Game With No Referee
The game-theoretic structure of this confrontation deserves precise naming, because it recurs in every protocol dispute — from Ethereum's proof-of-stake transition to the wars over ordinal inscriptions and beyond. It is a four-player coordination game with no referee.
The miners control instantaneous throughput. They can process transactions, refuse to process them, or point their hash power at an alternative chain. Their horizon is short: hardware depreciates, energy bills arrive monthly, and the opportunity cost of a slow week is measured in lost fees.
The nodes control ultimate sovereignty. Running a full node is cheap, portable, and deeply partisan. A node operator's exit cost is approximately zero, which means threat credibility is maximal. They are the only player who can leave without losing anything but attention.
The exchanges control liquidity and user access. They decide which chains get a ticker, which chains get a withdrawal pipeline, which chains get priced into the market's collective imagination. A fork without exchange support is an orphan. A fork with exchange support is a market event.
The developers control the narrative — the constant stream of patches, forks, and explainer posts that determines who is perceived as the legitimate heir of the protocol's history. In 2017, this was the most contested ground of all. Both the SegWit2x camp and the Core camp produced not just code but histories, origin stories, and genealogies of intent.
A fork is what happens when these four players cannot reach a cooperative equilibrium and the default is a split payoff that no one desires but no one can unilaterally prevent. The fascinating detail is how rarely that split payoff is actually collected. In most cases, a credible fork threat is enough to shift the negotiation. The BIP-110 cluster did not need to win; it needed to prove the exit door was open. From an economic standpoint, that small, isolated chain functioned as a cartel discipline mechanism: enforce the rule, or we walk. The mere credibility of the walkout reset the terms of every subsequent argument.
There is a counter-intuitive property of the four-player game that the 2017 experience revealed: the size of the threat is inversely related to its effectiveness. A fork threat backed by five percent of nodes is more credible than a threat backed by fifty percent, because the smaller group has less to lose and more to gain by demonstrating resolve. A fifty-fifty split is a war; a five percent walkout is a provocation. The provocations change behavior; the wars rarely do. This asymmetry, more than any block-size argument, explains the eventual outcome of 2017.
The Economics of Thin Chains
And so we arrive at the part of the story the narratives usually skip. The nodes that adopted BIP-110 did fork. They formed a chain that was — and this is the important word — isolated and economically thin. It carried a small fraction of the network's hash rate, a negligible share of its economic weight, and none of its institutional plumbing. No serious exchange listed its coins. No wallet integrated its address format. No merchant processed its payments. It existed primarily as a demonstration that the exit door could be opened.
I have spent my professional life studying what happens to economically thin chains, and the pattern is so consistent that I now treat it as a law rather than a tendency. A chain's value is not a function of its code; it is a function of the willingness of independent actors to coordinate around that code as if it were binding. A fork preserves the code but destroys the coordination. It is the same constitution, read in a country that no longer has a government. And a constitution without a government is, as Bitcoin's history quietly demonstrates, just an academic document.
The key metric is the ratio of narrative weight to security budget. A chain whose narrative is large relative to its hash power is a honeypot — first for arbitrageurs, then for attackers. The Bitcoin forks of 2017 and 2018 proved this repeatedly. Bitcoin Cash, for all its block-size bravado, rarely commanded more than a sliver of the original's hash rate, and was eventually subjected to its own fission in November 2018: a literal hash war between BCH and BSV that incinerated hundreds of millions of dollars in mining capital to settle a disagreement no token holder had ever been asked to vote on. The market priced the exit and found it cheap.
There is also an operational risk that the coverage of the period tended to bury beneath the meta-narrative: transaction replay. When two chains share a history, a transaction signed for one is, by default, valid on the other. Without replay protection, a user who sends coins on the main chain could find those coins duplicated or drained on the fork — or, worse, accidentally spend both chains at once. Exchanges dealt with this by freezing deposits and withdrawals around each fork date. Wallets added assertions and chain identifiers. But the protocols for safety were improvised under fire, and the improvised nature of that response tells you everything about how unprepared even sophisticated players were for the governance era they were entering.
The same dynamic, in a different key, runs through every corner of this industry. In 2020, when I spent three months deconstructing Yearn.finance's vault strategies for a series I called The Alchemy of Idle Capital, I documented how liquidity mining farmed the same ten thousand users from protocol to protocol — supporting the illusion of adoption with the repeated spending of a subsidy. Stop the incentives, and the real users vanish. A fork coin is the same phenomenon in its purest form: its initial value is minted from the marketing of a schism, not from the creation of a service. In 2022, after the TerraUSD collapse, my team's audit of the algorithmic stablecoin's peg mechanism found the same flaw in a different costume — the death spiral of seigniorage shares was, at its root, a thin chain of confidence. The fork and the algorithmic stablecoin are siblings: both are instruments whose value depends on a promise severed from a foundation.
The BIP-110 chain vanished into history because it was nothing but that promise. It had no users, no applications, no settlement volume, no reason to exist beyond the protest of its authors. And this is exactly why the story is so instructive. The fork that happened was irrelevant. The fork that was threatened was decisive. The threat acquired power precisely because the threat was cheap to execute; the execution's success was measured not in the survival of the forked chain but in the behavioral change of the main chain's players.
How the Market Priced the Uncertainty
The market's behavior around the 2017 forks is a case study in information assimilation under existential uncertainty. Chasing the ghost of value in a decentralized void, traders can feel the pressure of consensus but rarely measure it. Yet the price data tells a coherent story.
Through mid-2017, Bitcoin traded with a persistent uncertainty discount — a volatility premium reflecting the risk of a contentious split. When the Bitcoin Cash fork materialized on August 1, 2017, the immediate reaction was chaotic but contained: the price dipped, then resumed its ascent. The Bitcoin Cash episode added an important data point: the initial price of the fork was a fantasy. BCH spiked to levels that implied the market believed the new chain would inherit a meaningful share of Bitcoin's monetary premium — a belief that ignored the mechanical reality of hash migration, exchange custody, and user inertia. Within a year, the fantasy had been arbitraged away, and BCH had settled into its permanent position as a satellite rather than a sun.
The far larger overhang was SegWit2x. From August through November, the market wrestled with the possibility that Bitcoin would split into two roughly equal, mutually hostile networks — Bitcoin Core and Bitcoin SegWit2x — each claiming the legacy ledger, each defended by an army of partisans. That scenario would have been, in my estimation, the true catastrophe. Not because the technology would fail, but because the narrative would shatter. A Bitcoin that could be cleaved into two legitimate claimants is a Bitcoin whose digital-gold mythology collapses into a liquidity grab. The two-equal-chains scenario is the only fork scenario that genuinely threatens the network, and mercifully, it has never occurred.
When the SegWit2x organizers announced the cancellation in November 2017, the market exhaled. Bitcoin accelerated into the most violent parabolic phase of its history, closing the year near twenty thousand dollars. The pattern has repeated with every major fork scare since: uncertainty before, relief rally after. The market has, by now, been trained to treat fork threats as noise.
And that training, as I will argue below, is precisely the danger.
The Fork That Saved the Chain
The prevailing narrative in institutional circles is that forks are governance failures — double-IOUs, marketing stunts, cracks in the constitutional plaster. This is, forgive the bluntness, a fundamentally centralized mindset applied to a decentralized object. A fork is not a failure of governance. It is governance of last resort. It is the capacity for exit — Albert Hirschman's exit, in the classic triad of exit, voice, and loyalty — that gives meaning to all the softer voices in the system. The ability to leave disciplines the ability to speak. If the node class lacked the power to fork, its objections would be purely rhetorical. The New York Agreement would have been enforced by economic coercion, and Bitcoin would have become a negotiated settlement rather than a social contract.
BIP-110's rejection of non-signaling blocks was, from this angle, not censorship at all. It was the opposite: the exercise of independent judgment by economic agents. Censorship is the concentration of speech into fewer hands. Rejection is the dispersal of authority into more hands. The nodes that walked out were not suppressing a debate; they were refusing to be bound by a debate they had never consented to join. In that sense, BIP-110 was one of the most constitutionally significant acts in the brief history of the medium — regardless of whether its technical implementation ever reached the level of a serious client.
This is the blind spot that the industry's comfortable post-2017 narrative has papered over. Bitcoin did not survive its forks because it was durable. It survived because the act of forking was still cheap. The threat was credible because the exit door was still open. And what has happened since — the slow, grinding consolidation of hash power into a handful of pools; the rise of institutional custodians whose legal teams would never sanction a rule change; the regulatory apparatus that treats any deviation from the canonical chain as a taxable event or a securities violation — has been the quiet closing of that door.
Think carefully about what has changed in the legal environment. In 2017, a fork was a curious event that regulators observed from a distance; the SEC's DAO Report had been published only months earlier, and the tax treatment of split coins was a subject for forum speculation, not IRS guidance. By 2021, the IRS had rules for taxed fork coins. By 2024, any node operator who participated in a contentious upgrade was navigating securities law, tax law, and the compliance policies of institutional counterparties whose risk tolerance for unusual ledger events is approximately zero. The cost of exit is not just technical. It is legal, regulatory, and reputational — and it has been rising every year.
Let me be direct about the measurement. After the fourth halving, mining revenue collapsed relative to the network's market capitalization, and the unit economics of security have deteriorated accordingly. Hash power, driven by economies of scale in hardware procurement, energy contracting, and capital access, is concentrating. If that concentration proceeds toward the three-pool scenario I have flagged in my market briefs, the credibility of any future fork collapses to zero. A fork requires willing miners, liquid exchanges, and a legal envelope that permits the experiment. All three are scarcer today than they were in 2017. The nodes retain the right to fork in theory. They no longer possess the ability in practice.
That, rather than the block size limit, is the actual lesson of BIP-110. The veto that saved Bitcoin in 2017 is the veto we have spent seven years pricing out of existence. Decentralization is not the state of having many nodes; it is the state in which the exit threat remains credible. We have optimized the first while destroying the second.
The Next Signal War
What does this mean for the next cycle? It means the next BIP-110 will not be a block-size fight. It will be a fight over the meaning of signaling itself — over who gets to declare what counts as a valid transaction, a valid message, a valid actor.
I have been working on the intersection of blockchain and machine agency since 2025, when I outlined the verifiable-compute narrative for autonomous AI agents transacting on-chain. The unsettling discovery is that the classic coordination game has acquired a fifth player: the machine. AI agents can signal, too. They can run nodes. They can, if instructed, refuse to accept blocks that do not conform to their training objectives. The Bitcoin of 2030 will not be governed solely by miners, nodes, exchanges, and developers; it will be governed by algorithmic exegesis — layers of interpretation between the raw ledger and the economic actors who depend on it. That introduces an entirely new class of fork: not a split in the chain, but a split in the interpretive layer. Two agents reading the same block and reaching incompatible conclusions is a fork that no block explorer will display and every derivative market will feel.
In the nearer term, watch the Layer2 ecosystem with the same lens. Dozens of rollups claim to scale Ethereum, but most of them slice an already-scarce liquidity pool into thinner and thinner fragments. Each new rollup is a soft fork of the user base — a polite exit, conducted through a bridge rather than a chain split. The mechanism is the same as BIP-110: a group of actors declares that it will play only by its own rules, and hopes the economic weight of that declaration forces the broader system to accommodate it. Sometimes it works. Sometimes it produces a thin, isolated island of liquidity that never achieves escape velocity.
So here is the question I leave with readers who believe they are investing in infrastructure: are you pricing the cost of exit, or only the promise of entry? The ghost of value in a decentralized void does not materialize from total value locked, nor from the ambition of a whitepaper. It materializes from the credible threat of walking away. The walkout, not the walk-in, is the founding act of any new settlement.
In 2017, the nodes who walked out taught the market that Bitcoin could be defended by being abandoned. In the years since, the market has learned to price that defense as a risk rather than an asset. I will not be surprised if the next walkout comes from a machine that no one thought to ask whether it consented to the rules. The only question — and it is the question every analyst should be asking — is whether the exit door, when that moment arrives, is still open. Don't wait for the fork to answer. By then, the fork will have already decided who you are.
