The Blocksize War: Who Gets to Change Bitcoin’s Rules

As use grew, blocks approached the limit and fees rose. Advocates of larger blocks warned that expensive transactions would drive out ordinary payments and turn Bitcoin into a network for the wealthy. Advocates of a more cautious course warned that rapid growth in data and requirements would weaken the ability to run a full node at home and concentrate validation in data centers.

Both sides spoke about decentralization, but measured it differently. One counted the people who could afford a transaction; the other counted the people who could afford to check the rules independently. Bitcoin XT, Classic and later Unlimited promised a direct increase. Development around SegWit sought a change to transaction format and more efficient use of space.

The one-megabyte limit was neither a law of nature nor Bitcoin’s original goal. Satoshi introduced it to protect the young network from excessively large blocks. The dispute was therefore not about whether one sacred constant could be changed, but how quickly capacity could grow without moving validation toward people with the fastest connections, largest disks and greatest patience for an initial synchronization. Even a small technical change carried a long-term vision of who would still be able to say: I checked the rules myself.

Andresen’s BIP101 proposed a jump to eight megabytes followed by predictable growth in the limit. Hearn placed it in the Bitcoin XT client, and both publicly asked miners and users to support the change. To them, a full queue of transactions violated the original promise of inexpensive electronic cash. If a new user had to compete with a high fee for a small payment, they believed the network would lose to cards before reaching ordinary life.

The other side was not composed of people who wanted to keep Bitcoin small. Pieter Wuille proposed Segregated Witness, while Gregory Maxwell and other developers discussed second layers and more efficient use of every byte. Their concern looked several steps ahead: a larger block today means greater bandwidth, storage and validation demands forever. When a home node ceases to be practical, a user begins trusting someone else’s server. Both camps defended accessibility; each saw it on a different side of the transaction.

In May 2017, a group of large companies and miners announced the New York Agreement. The SegWit2x plan coupled rapid activation of SegWit with a subsequent hard fork to a two-megabyte base block. Its signatories cited a large share of hashpower and transaction volume and spoke as though coordination among major businesses could close the dispute.

The agreement did not bind users running other software. Critics rejected above all the idea that a closed meeting could promise an incompatible change on behalf of the network. Miners can propose blocks and companies can give them a brand; a node can still declare them invalid.

The original announcement cited 58 companies from 22 countries and 83.28 percent of hashpower. Those figures carried weight, but did not measure the agreement of independent nodes or people holding bitcoin outside the signatories’ services. Combining SegWit with a later hard fork also created a political trade: each side would receive part of what it wanted. Once the first part activated while the second remained disputed, it became clear that the package had no institution able to enforce its completion.

Digital Currency Group founder Barry Silbert organized the agreement during the Consensus conference in New York. Representatives of exchanges, wallets, payment companies and mining firms sat at the table; Jeff Garzik led the SegWit2x implementation. To businesspeople exhausted by a long dispute, it looked like a mature compromise: influential people had finally agreed on a date and signed their companies beneath it. But Bitcoin had no membership register in which those signatures constituted a qualified majority.

Critics objected not only to two megabytes but to the way the decision was made. Adam Back, Gregory Maxwell and others warned that an incompatible change on a short schedule presented users with a fait accompli. The signatories controlled services through which many people passed, and miners might supply most hashpower. They could not remotely change the validation rules of a node running in someone else’s apartment. The agreement thus created an odd situation: the largest companies could announce what they would call Bitcoin, but every node owner retained a personal reject button.

BIP148 proposed a user-activated soft fork: beginning August 1, participating nodes would reject blocks that did not signal SegWit. It was risky pressure. If an economically significant part of the network remained elsewhere, competing histories or chaotic reorganizations could emerge.

Before the decisive day, however, miners began signaling conditions through BIP91 that led to SegWit locking in. The outcome cannot honestly be credited to one actor: the UASF threat changed incentives, miners coordinated signaling and existing nodes enforced the rules of a compatible soft fork. SegWit activated on August 24, 2017.

Nodes running BIP148 could neither mine SegWit themselves nor compel others to use it. They could only commit in advance that after the deadline they would refuse a category of blocks. The force of that promise depended on whether exchanges, merchants, wallets and holders whose coins gave a chain economic meaning would follow it. UASF therefore acted as a public commitment in a coordination game: its effect came not from an order, but from the credible possibility that a noncompliant chain would have hashpower and fewer buyers at the same time.

A developer using the name Shaolinfry published BIP148. People began wearing caps marked UASF and called August 1 independence day. Behind the memes stood a serious wager. Had miners continued without signaling while a significant part of the economy followed BIP148, histories might have split temporarily and users would not have known which incoming payment would remain valid. An activist running a node was risking more than a slogan; they risked their own ability to receive money safely.

James Hilliard’s BIP91 offered a path for miners to coordinate before the deadline and lock in SegWit signaling at a lower threshold. When support suddenly rose, each side interpreted the result in its own way. UASF supporters saw proof that user pressure had changed incentives. Miners pointed to their own coordination, and New York Agreement signatories to their compromise. History provides no controlled experiment in which one group is removed. What is certain is that the looming conflict made everyone act sooner.

On August 1, Bitcoin Cash split away with larger blocks and its own rules. It was not a rewriting of the original nodes, but a new incompatible chain sharing history up to the split. The market and users could then value and run both assets separately.

The second part of the New York Agreement was meant to activate a two-megabyte hard fork in November. On November 8, its organizers suspended it, acknowledging that they had not obtained sufficient consensus for a clean upgrade. High hashpower signaling had not become certainty that the economy would accept the new blocks as Bitcoin.

At the moment of separation, both chains shared their earlier history and the same set of unspent coins. Anyone controlling keys before the split could, under certain conditions, transact with corresponding assets on both sides afterward. From that point, however, each chain accepted different future blocks and built its own network effect. The split did not force old nodes to change their definition of a valid transaction; it gave supporters of different rules a way to bear the costs and the result of their choice.

Bitcoin Cash gave supporters of larger blocks something an endless debate could not: a working chain under their rules. Exchanges had to assign it a ticker, wallets had to address replay protection, and holders had to decide whether to sell, keep or ignore the new coins. Part of the dispute moved from mailing lists to the market. The prices, hashpower and usage of both chains began expressing an ongoing decision rather than a one-time vote for a sole winner.

SegWit2x was meant to be different. Its supporters wanted the two-megabyte chain to inherit the BTC brand and become the continuation, rather than a side branch. As November approached, it lacked broad client-software support and a clear plan for handling two chains safely. On November 8, Mike Belshe, Wences Casares, Jihan Wu, Jeff Garzik, Peter Smith and Erik Voorhees announced the suspension. They acknowledged that proceeding without broader agreement would divide the community they had wanted to unite.

The war is often told as a pure victory for users. The more accurate conclusion is harder. Developers control attention and code quality, miners the ordering of blocks, companies liquidity and branding, and users the choice of software and counterparties. No group holds every lever by itself; some nevertheless hold substantially more than others.

Bitcoin did not vote with a single number. It coordinated through code, public arguments, business decisions, deadlines and willingness to risk a split. The outcome was neither quiet nor cheap. That is its significance: the rules stayed the same not because the dispute did not exist, but because changing them required persuading more than the most visible room.

Nakamoto consensus determines the ordering of transactions within jointly accepted rules; it is not a parliament in which hashpower votes to rewrite those rules. A miner can produce a block carrying enormous work, but a node discards it if it violates its software. The choice of rules therefore occurs outside the blockchain itself: in code, client versions, prices, contracts and the people with whom users are willing to trade. The Blocksize War did not discover a simple Bitcoin constitution. It exposed a system in which change is possible, but no one has a guaranteed right to call it consensus.

People on every side paid a personal price. Mike Hearn publicly left the project in January 2016 and declared the experiment a failure. Gavin Andresen lost his former position in Bitcoin Core. Developers faced campaigns, companies faced boycotts and individuals faced attacks that turned technical disagreement into a test of character. The victor’s story often flattens those people into traitors and heroes. It is more accurate to admit that many were protecting different visions of the same goal and sometimes used pressure they considered a necessary defense.

After the war, the word consensus was used more cautiously. It no longer meant a simple count of miner signals, a list of companies or the agreement of leading developers. It meant a condition in which enough actors consider a change safer than a split and the others lack the will or strength to sustain competing rules. That is not an elegant voting system. It is continuous negotiation in which the ability to leave gives weight even to people without a title — while never guaranteeing that their chain will have value, users or security.