Who controls Bitcoin? Governance, the Knots–Core debate, and forks

The short answer to “who controls Bitcoin?” is: no one, and that’s the point. There’s no CEO, no board, no foundation that can push a change through. Bitcoin runs on rough consensus — improvement proposals get debated, reviewed, and written into software, and the people running full nodes decide whether to adopt it. The clearest evidence the model holds: in seventeen years, Bitcoin has reversed confirmed transactions only twice, both times to fix software bugs, and never once to undo a theft or enforce a policy. The current Knots-versus-Core argument and the contrast with Ethereum’s DAO fork both show what this design does — and doesn’t — allow.

Who’s actually in charge of Bitcoin?

No single party. Bitcoin borrows its governance from internet standards bodies: changes are written as Bitcoin Improvement Proposals (BIPs), discussed at length, reviewed by developers, and shipped in software that node operators choose whether to run.¹ Authority ultimately rests with users running full nodes, who decide which version of the rules to accept. Nobody can mandate anything.

The process is deliberately, almost frustratingly, slow — SegWit took around 20 months from proposal to activation; Taproot about three and a half years.¹ That conservatism is a feature: a security bug in Bitcoin could vaporise hundreds of billions in value, so changes get years of review, and a smaller, simpler codebase means a smaller attack surface. Slow is the price of safe.

Has Bitcoin ever rewound the chain?

Twice — and the reason why is the most important line in Bitcoin’s governance story. In August 2010, a bug let someone conjure 184 billion BTC in a single transaction; developers shipped a patch within hours and coordinated miners to orphan the bad blocks.² In March 2013, an unintended database change split the chain, and developers coordinated a rollback to the compatible version.³ Both reversed confirmed transactions — but both were bug fixes.

What has never happened: a transaction reversed to recover stolen funds, claw back a hack, or enforce any social or political outcome. That line — bugs yes, policy never — is what people mean when they call Bitcoin’s ledger immutable. It’s not that the chain can’t be altered; it’s that the network has refused to alter it for any reason other than fixing broken code.

Knots vs Core: is Bitcoin fracturing?

This is the live debate, and it looks like conflict but is really the system working. Bitcoin Knots is an alternative node client maintained by Luke Dashjr, one of Bitcoin Core’s earliest contributors.⁴ It takes a more conservative line than Core on what transactions nodes should relay — it filters Ordinals, Runes, and Stamps (protocols that embed arbitrary data in transactions) and enforces tight limits on data embedding. Bitcoin Core went the other way: v30, in October 2025, effectively removed those limits.⁵

This is a fight at the policy layer, not the consensus layer. Consensus rules define what’s valid — break them and you split the chain. Policy rules are just each node’s preferences about what to relay and store in its mempool;⁶ Knots’ filters slow certain transactions down but can’t stop one a miner accepts directly. Knots has grown fast — from roughly 394 nodes in January 2025 to over 5,200 a year later, about 22% of the network⁷ — which signals real support for the money-first view, but it’s a preference being expressed, not a rule being imposed.

The two camps hold genuinely different visions. Core treats Bitcoin as a neutral settlement network — block space is a market, whoever pays gets confirmed, data payload aside.⁸ Knots treats it as money-first infrastructure, where non-monetary uses just burn scarce block space. There’s a real concern on the Knots side: the broader token world’s manias (ICOs, NFTs, memecoins) did genuine harm, and keeping that off Bitcoin has appeal. But there’s a real risk on the other side too — once filtering “arbitrary data” is normal, the definition of “undesirable” can creep, from inscriptions today toward disfavoured entities or flagged addresses tomorrow. Reasonable people land in different places. The point that matters for governance is that the entire dispute plays out through which software people choose to run — not foundation edicts or token votes. Users decide which rules their nodes enforce, full stop.

How does this compare to Ethereum’s DAO fork?

The contrast is stark. In July 2016, hackers drained 3.6 million ETH (about $60 million then) from a smart contract called The DAO. Ethereum’s leadership proposed and executed a hard fork that reversed the transactions and returned the funds within weeks; a minority rejected it and kept the original chain alive as Ethereum Classic.⁹ That episode proved Ethereum’s history can be rewritten when social consensus among key stakeholders exists — the fork was framed as recovering stolen money, but the mechanism showed a coordinated group controlling the social layer can alter the ledger.

The market rendered a verdict. By January 2026, ETH traded around $3,000 and ETC around $12.80 — a ratio of roughly 234 to 1.⁹ Influence is concentrated, too: the Ethereum Foundation’s treasury (about $970 million in late 2024) funds much of the network’s research and direction.¹⁰

To be fair to the other side, there’s a real case for flexible governance. It let Ethereum respond fast to protect users from a hack, and the Merge — which cut Ethereum’s energy use by roughly 99.95%¹¹ — would have been impossible without governance capable of large protocol changes. A system that can never upgrade eventually ossifies. It’s a trade-off, not a clean win for either side.

Which model is better?

It depends on what you want the asset to be. Bitcoin’s base layer leans toward ossification on purpose — it keeps the monetary base boringly predictable and pushes innovation up into layers built on top. Ethereum’s keeps the base malleable so it can adapt and intervene. For a general-purpose smart-contract platform, adaptability is a reasonable priority. For money, the opposite is the feature: you want rules no coordinated group — however well-intentioned — can rewrite, because “we reversed the chain for a good reason this time” is precisely the discretion sound money is meant to remove. The same logic is why Bitcoiners reject proof-of-stake at the base layer — but that’s its own story.

Common questions

Who makes decisions about Bitcoin? No single entity. Changes are proposed as BIPs, reviewed by developers, and adopted only if the people running full nodes choose to run the new software. Ultimate authority rests with node operators, not any company or foundation.

Can Bitcoin transactions be reversed? Only twice in its history, both to fix software bugs (a 2010 supply bug and a 2013 chain split). No transaction has ever been reversed to recover stolen funds or enforce a policy — that distinction is what makes the ledger credibly immutable.

What is the Knots vs Core debate about? Whether nodes should filter non-monetary data (like Ordinals) out of the transactions they relay. It’s a disagreement about policy, not Bitcoin’s core rules, and it’s being settled by which client people choose to run — a live example of Bitcoin governance in action.

What was Ethereum’s DAO fork? A 2016 hard fork that reversed transactions to recover funds stolen in a hack. It worked, but it showed Ethereum’s history can be rewritten by coordinated stakeholders — the opposite of Bitcoin’s track record. The dissenting chain survives as Ethereum Classic.

Footnotes

1. Bitcoin Improvement Proposal (BIP) — the process by which changes are proposed, reviewed, and adopted only if node operators run the new software. SegWit took ~20 months from proposal to activation; Taproot ~3.5 years. Bitcoin Improvement Proposals (BIPs) repository, GitHub. https://github.com/bitcoin/bips ↩ ↩²

2. In August 2010, a value-overflow bug let someone create 184 billion BTC in one transaction; developers patched it within hours and orphaned the invalid blocks. Bitcoin Wiki, “Value overflow incident.” https://en.bitcoin.it/wiki/Value_overflow_incident ↩

3. In March 2013, an unintended database change caused a chain split, resolved by a coordinated rollback to the compatible chain. BIP 50. https://github.com/bitcoin/bips/blob/master/bip-0050.mediawiki ↩

4. Bitcoin Knots — an alternative full-node client maintained by Luke Dashjr, a Bitcoin Core contributor since 2010 and CTO of the Ocean mining pool. Bitcoin Knots project. https://bitcoinknots.org/ ↩

5. Knots enforces tight limits on data embedded in transactions; Bitcoin Core v30 (October 2025) effectively removed the prior cap, raising the default OP_RETURN size to ~100,000 bytes. ↩

6. Mempool — each node’s local store of unconfirmed transactions. Relay and policy rules differ from consensus rules, so a node’s filtering affects how transactions propagate but cannot stop one miners accept. ↩

7. Coin Dance — Bitcoin Knots grew from ~394 nodes (January 2025) to ~5,241 (~21.7% of ~24,197 reachable nodes) by January 2026. https://coin.dance/nodes ↩

8. Axiom Capital, “The Bitcoin Stack” — a framework analysing Bitcoin’s layers as distinct markets for work, settlement, and routing. https://www.axiombtc.capital/stack ↩

9. After the July 2016 DAO hack drained 3.6M ETH (~$60M), Ethereum hard-forked to reverse the transactions; a minority continued the original chain as Ethereum Classic. By January 2026 the ETH:ETC price ratio was roughly 234:1. Ethereum Foundation Blog, “Hard Fork Completed.” https://blog.ethereum.org/2016/07/20/hard-fork-completed ↩ ↩²

10. Ethereum Foundation — treasury of approximately $970M (October 2024), funding much of the network’s research and development. Ethereum Foundation Report 2024; figures via CryptoSlate. https://cryptoslate.com/ethereum-ecosystem-treasuries-top-22-billion-with-resources-poised-to-support-networks-future-growth/ ↩

11. The Merge cut Ethereum’s electricity use by ~99.95% (Crypto Carbon Ratings Institute analysis). Ethereum.org, “Energy Consumption.” https://ethereum.org/en/energy-consumption/ ↩