Every payment system in history has answered one question with a trusted third party: whose ledger is correct? A bank, a clearing house, a government — someone you have to trust to keep the books straight. Bitcoin's founding move, in Satoshi Nakamoto's 2008 whitepaper, was to answer that question with physics and math instead of trust. The mechanism is Nakamoto Consensus, and once you see how its two halves fit together — Proof-of-Work plus the heaviest-chain rule — the whole system stops looking like magic.
Signal over noise: the most common misconception about Bitcoin is that "miners run the network." They don't. Miners order transactions; full nodes enforce the rules. A hashrate majority can reorder history — it cannot forge a signature, steal your coins, or mint money outside the schedule. Get that split right and everything else follows.
Proof-of-Work — the lottery you pay for in electricity
Mining is not "solving complex math." It's a brute-force guessing game. A miner assembles a block, takes its header, and hashes it with SHA-256 — twice (SHA-256d). The output is a 256-bit number. The rules demand a hash that lands below a target value; the only knob the miner can turn is a field called the nonce, so they increment it and re-hash, billions of times a second, until one attempt drops under the target. There's no shortcut — the only strategy is to guess faster than everyone else.
That's what "work" means: a hash below target is proof that, on average, an enormous number of guesses were burned to find it. Anyone can verify it in a single hash. The asymmetry — crushingly hard to produce, trivial to check — is the entire foundation.
Difficulty — the self-correcting clock
If work got cheaper as more machines joined, blocks would arrive faster and faster. Bitcoin prevents that with an automatic difficulty retarget every 2,016 blocks (about two weeks). The network reads the timestamps in those blocks' headers, compares how long they actually took against the 10-minute-per-block target, and moves the difficulty up or down to pull the average back to ten minutes. More hashpower joins → difficulty rises → block time returns to ~10 minutes. It's a closed feedback loop with no committee and no operator.
Nakamoto Consensus — follow the heaviest chain
Each block commits the hash of the previous block's header, chaining them so that altering any past transaction would require re-doing the work of that block and every block after it. That's what makes the ledger tamper-evident (the popular word "immutable" is shorthand — strictly, it's probabilistically final).
When two valid blocks appear at once, the chain temporarily forks. Nodes resolve it with one rule: follow the chain with the greatest cumulative Proof-of-Work — the heaviest chain, not merely the one with the most blocks. (Bitcoin Core switched from counting blocks to summing chainwork precisely because block-count was manipulable.) The losing branch's blocks become stale and its transactions return to the pool. Legitimacy isn't a vote — it's whichever history has the most energy poured into it.
Nodes vs. miners — who actually holds the power
This is the part most explainers get wrong. A full node is software that independently validates every transaction and block against the consensus rules: it checks that coins being spent exist and are unspent (the UTXO set), that no coin is double-spent, that every signature is valid, and that the block obeys every rule. Miners do one extra job on top of being nodes: they perform the Proof-of-Work.
The consequence is profound. Even an attacker with 100% of the hashrate cannot forge your signature to spend your coins (they don't have your private key), cannot create bitcoin outside the issuance schedule, and cannot change the rules — because every full node rejects invalid blocks, no matter how much work backs them. Miners propose; nodes dispose.
The 51% attack — what majority hashpower actually buys
A 51% attack is when one entity controls a majority of hashpower. With it, they can secretly mine a private chain, outpace the honest network, and then release it — the heaviest-chain rule forces everyone to adopt it, erasing the attacker's recent transactions and enabling a double-spend. That's the real, limited prize: they can undo their own recent payments, not loot the network.
Security rests on cost. To out-hash the honest network you'd need more machines and electricity than everyone else combined — an estimated ~$6 billion for a one-week attack at current scale, on top of watching the coin you're attacking crater in value. Two honest caveats: the "51%" line is a simplification — selfish mining can be profitable near ~33% — and mining-pool concentration is real (the top two pools have hovered around ~43% of hashrate). The threat model is economic, not theoretical.
Incentives — the subsidy, the halving, and the fee cliff
Why mine at all? Each block pays the winner a block subsidy of newly minted bitcoin plus the transaction fees it contains. The subsidy halves every 210,000 blocks — roughly every four years. It began at 50 BTC, and the April 2024 halving cut it to 3.125 BTC. This geometric decay is what caps supply near 21 million.
It's also the system's biggest open question: as the subsidy trends toward zero, security must be funded almost entirely by transaction fees. Whether a fee-only market can pay for enough hashpower to keep the chain secure is a genuine, unresolved debate among researchers — not a settled fact.
Forks — how the rules change without a boss
With no central authority, rule changes ship as forks, and the direction matters:
- A soft fork tightens the rules (some previously-valid blocks become invalid). Because the new valid set is a subset of the old, un-upgraded nodes still accept the new blocks — it's backward-compatible, with little split risk. SegWit (2017) and Taproot (2021) were soft forks.
- A hard fork relaxes the rules (previously-invalid blocks become valid). Un-upgraded nodes reject the new blocks, so without unanimous adoption the network permanently splits into two chains. Bitcoin Cash (2017), born of a block-size dispute, is the canonical example.
There's a governance twist worth knowing: a User-Activated Soft Fork (UASF) flips the usual script. Instead of waiting for miners to signal readiness, economic full nodes set a flag day and begin rejecting non-compliant blocks regardless of miner support. BIP148 (2017) did exactly this to break the SegWit stalemate — the clearest proof that hashpower doesn't govern Bitcoin's rules; the economic majority of nodes does.
Bottom line
Bitcoin replaces a trusted bookkeeper with a simple, brutal contest: whoever burns the most verifiable work extends the ledger, and everyone independently checks the result. Proof-of-Work makes history expensive to write; difficulty keeps the pace steady; the heaviest-chain rule picks a single truth from competing forks; and full nodes — not miners — draw the hard line on what's valid. The open edges are honest ones: mining concentration, the long-run shift from subsidy to fees, and the cost of majority attacks against deep-pocketed adversaries. But the core has held, unchanged, since 2009. Not by trust — by work.
Sources: Bitcoin developer docs — block chain · Bitcoin whitepaper (Nakamoto) · What is Nakamoto Consensus (Bitcoin Magazine) · What is a 51% attack (Chainlink) · What is a Bitcoin node (Unchained) · Bitcoin halving economics (Fidelity) · Soft fork activation (Bitcoin Optech)
Not financial advice. An automated technical explainer; consensus claims were adversarially fact-checked against primary sources where possible. Some figures (hashrate, pool shares, attack cost) are time-sensitive — verify before relying on them.