Running a Bitcoin Node: What It Does, Who Should Do It, and How to Start
“Don’t trust, verify” is the Bitcoin world’s favorite slogan. Running your own node is what verification actually looks like.
If you have read any Bitcoin news this year, you have seen the names: Bitcoin Core, Bitcoin Knots, OP_RETURN, spam filters, BIP-110. For a beginner they sound like an argument among programmers. They are, but the argument turned out to be about the most basic question in Bitcoin, which is who gets to decide what the network is for. This guide explains the two pieces of software, the disagreement between them, and why a migration of node operators from one to the other became the seed of the 2026 fork.
What a node is, in one paragraph
A Bitcoin node is a program that downloads every block, checks every transaction against the rules, and refuses anything invalid. It is how you verify Bitcoin for yourself instead of trusting an exchange or a website. Nodes also relay transactions to each other before miners include them in blocks. If this is new to you, our beginner’s node guide explains why ordinary people run them and how to start.
Bitcoin Core: the reference
Bitcoin Core is the descendant of the original software Satoshi Nakamoto released. It is maintained by a group of volunteer developers, it is where most protocol work happens, and it is what the large majority of nodes run. Its defaults matter enormously, because whatever Core ships becomes the network’s practical behavior simply by being the thing most people install. Core has no authority in a legal sense; its power is convention.
Bitcoin Knots: the stricter cousin
Bitcoin Knots is a long-maintained alternative built from the same code base by Luke Dashjr, one of Bitcoin’s earliest contributors. It follows the same consensus rules as Core, meaning it accepts and rejects exactly the same blocks, but it ships with stricter policy defaults: rules about which transactions the node will relay and which it will quietly ignore. In particular, Knots filters transactions that use Bitcoin as a storage medium for images, tokens, and other non-financial data.
That distinction between consensus and policy is the key to the whole story, so it is worth stating plainly. Consensus rules decide what a valid block is; every node must agree on them or the network splits. Policy decides what an individual node chooses to pass along; nodes can differ freely. A Knots node and a Core node agree about every block. They disagree about what to do with a transaction before it gets into one.
The disagreement: what is a block for?
Beginning in 2023, techniques for embedding arbitrary data in Bitcoin transactions took off; our Ordinals and Runes guide explains how. Blocks filled with images and tokens, fees spiked for ordinary payments, and the blockchain grew faster than ever. Every full node, remember, must store all of it forever.
Two views emerged. One, broadly Core’s position, holds that a node should not be in the business of judging transactions: if someone pays the fee, the data gets in, and filtering it at the relay level is ineffective anyway because miners will include it regardless. The other, Knots’s position, holds that Bitcoin is money, that node operators volunteer their disk and bandwidth to verify money, and that a node has every right to refuse to carry data its operator considers spam. Filtering may not stop determined miners, but it expresses what the network’s verifiers actually want.
In late 2025 the disagreement became concrete. Bitcoin Core version 30 removed its long-standing default limit on OP_RETURN outputs, the main channel for embedding data. Core’s maintainers argued the limit no longer accomplished anything. To the money-only camp, it read as surrender.
The migration nobody organized
What happened next had no precedent. Node operators, individually, began switching from Core to Knots. Through 2025 and into 2026, Knots went from a niche choice to a meaningful share of the reachable network. There was no company behind it and no campaign budget. People who ran nodes because they cared about verifying Bitcoin for themselves decided they also cared about what those nodes would carry, and they changed software to say so.
It is hard to overstate how unusual this is. For fifteen years, the assumption was that Core’s defaults were effectively law because switching was too much trouble for too little gain. Thousands of operators proved otherwise. Whatever you think of the filtering question, the migration established that Bitcoin’s software is a choice, not a mandate, and that node operators can act collectively without anyone leading them.
From policy to consensus: BIP-110
Filtering at the node level had a ceiling. A Knots node could refuse to relay a data-heavy transaction, but if a mining pool included it in a block, the block was still valid and every node, Knots included, had to accept it. To actually keep data out, the restriction had to become a consensus rule.
That was BIP-110, drafted by Dashjr: a temporary soft fork capping large OP_RETURN outputs, restricting the script formats used for embedding, and limiting contiguous arbitrary data to 256 bytes, with the limits set to expire on their own. It asked miners to signal support in 55 percent of blocks. About 2.5 percent did. The pools that profit from high-fee data transactions were not going to vote for their own pay cut.
So the Knots community did what the software-switching had foreshadowed: they acted without permission. Their nodes activated the BIP-110 rules at block 961,632 on August 8, 2026. When the mining industry kept producing non-compliant blocks and the chain stalled after eight, the developers replaced the mining algorithm itself with BLAKE2b, which Bitcoin’s ASICs cannot compute, and resumed at block 961,640 on August 30. The full story, including what it means for your coins, is in our BIP-110 explainer.
Why this is good news even if you never switch
You do not have to agree with Knots’s filtering philosophy to appreciate what its rise demonstrated. Bitcoin’s founding document, which we walk through in our white paper guide, describes a system where participants enforce the rules by choosing which chain to accept. That only works if the choice is real. For years it was mostly theoretical; Core was the only serious option and the pools produced whatever they liked. The Knots migration, and the fork that followed, turned the theory into practice.
It also delivered an answer to the centralization worry that hangs over modern Bitcoin. A few custodians hold the coins, a few pools mine the blocks, and one project ships the software. Yet when it mattered, a loose community of node operators changed software, activated a rule, and, when blocked, changed the mining algorithm and carried the entire history of Bitcoin with them. If the network were captured, none of that would have been possible.
Which should you run?
If you run a node today, here is the honest picture as of September 2026.
- Bitcoin Core follows the legacy SHA-256 chain, the one exchanges, ETFs, and nearly all wallets mean when they say “Bitcoin.” If your goal is to verify the chain your coins trade on, Core does that.
- Bitcoin Knots without BIP-110 activation also follows the legacy chain, with stricter relay filters. It is a way to express the money-only preference without leaving the main network; check the release notes so you know which rules a given version enforces.
- Bitcoin Knots (BLAKE2b release) follows the new chain. Run it if you want to verify or mine the BLAKE2b chain. It will not track the legacy chain.
Many people in the Knots community run more than one. There is nothing wrong with verifying both chains while you decide what you think. A node costs little, and every independent verifier makes both networks a little harder to capture.
The honest caveats
This article presents the Knots side sympathetically because its story is under-told and, we think, genuinely important. It is not the only side. Core’s developers made a reasoned case that relay filters do not stop miners and that arbitrary-data limits are unenforceable in the long run. The BLAKE2b chain, as of this writing, has a small hash rate, no exchange listing, and a name most of the industry does not accept. And your pre-fork coins exist on both chains, so do not move them on either without understanding replay protection.
What is not in dispute is the lesson. Node operators are not passengers. In 2025 and 2026 they proved, first by switching software and then by forking, that the people who verify Bitcoin hold the final vote over what it is. That was always the promise. It is encouraging to see it kept.