How Bitcoin mining works: the job, the reward and the race

Miners do two jobs: bundle transactions and win the right to add them

Every few minutes, someone somewhere adds a new block to Bitcoin’s ledger. The people and companies who do this are called miners, and the white paper lays out their job in six steps. New transactions are broadcast to every node, meaning every computer running the Bitcoin software. Each node collects them into a block. Each node then works on “finding a difficult proof-of-work for its block,” and when one finds it, it broadcasts the block to everyone else.

The other nodes then check the block. They accept it “only if all transactions in it are valid and not already spent,” and they show their acceptance by starting work on the next block, using the new block’s hash as the previous hash. That last step is the whole coordination mechanism. Nobody votes; nobody signs off. Miners simply build on the block they accept, and the chain with the most work behind it wins.

So mining is really two jobs. The first is clerical: gather valid transactions and put them in a block. The second is the race: be the first to find a proof of work for that block. The developer guide describes the point of it in one line: mining “adds new blocks to the block chain, making transaction history hard to modify.”

The work is guessing, and the answer is easy to check

The proof of work is a search. A miner takes the block’s header, a compact summary of its contents, and runs it through SHA-256, a hash function, meaning a fixed recipe that turns any input into a short fingerprint. The white paper describes the goal: a hash that “begins with a number of zero bits.” The miner changes a field in the header called the nonce, hashes again, and repeats until the result is small enough. There is no shortcut. As the paper says, the average work “is exponential in the number of zero bits required and can be verified by executing a single hash.”

That asymmetry is the trick. Finding a valid header can take an enormous number of attempts; checking one takes a single hash. The developer guide describes the loop in practice: mining hardware runs through nonce values, and if none of them produces a hash below the threshold, it “gets an updated block header with a new merkle root,” meaning a slightly different set of transactions, and starts again.

The intuition is a lottery, not a puzzle. Each hash is a ticket. Buying more tickets per second raises your odds but never guarantees a win, and a small miner with a tiny fraction of the network’s hashing power can still find a block occasionally, just rarely. The proof of work explainer goes further into why a lottery, of all things, keeps a ledger honest.

Miners are paid in new coins and fees, and the new coins shrink on a schedule

Miners are paid twice over. The white paper explains that “by convention, the first transaction in a block is a special transaction that starts a new coin owned by the creator of the block.” That is where every bitcoin comes from. The second payment is fees: when a transaction’s outputs are worth less than its inputs, the difference “is a transaction fee that is added to the incentive value of the block.” The developer guide sums it up as proceeds “from the block reward and transaction fees.”

The new-coin part shrinks on a fixed schedule. The email announcing the first release in January 2009 promised total circulation of 21,000,000 coins, “cut in half every 4 years.” In the code that every node runs, the rule is written in blocks rather than years: the subsidy halving interval is 210,000 blocks, and the genesis block’s reward was 50 coins. The Bitcoin Wiki traces the steps from there, 50 to 25 to 12.5 and onward. For the exact block heights and the arithmetic, see the dossier on halvings and supply.

There is a waiting period on new coins. The developer guide notes that the output of a coinbase transaction “cannot be spent for at least 100 blocks.” The reason is that a block near the tip of the chain can still be replaced if the network settles on a different branch, and coins from a replaced block would vanish. A hundred blocks of burial makes that vanishingly unlikely.

A thermostat keeps blocks arriving every ten minutes, whatever the hardware

If faster hardware simply meant faster blocks, the 21 million cap would arrive early and the ledger would be chaotic. The white paper anticipated this: difficulty “is determined by a moving average targeting an average number of blocks per hour. If they’re generated too fast, the difficulty increases.”

The developer guide gives the numbers. Every 2,016 blocks, the network compares how long those blocks took with an ideal of 1,209,600 seconds, which is two weeks, and moves the target in proportion, with a limit on how far it can move in one step. In Bitcoin Core’s parameters the two constants sit side by side: a target spacing of ten minutes and a target timespan of fourteen days. The difficulty adjustment article covers the details and the edge cases.

This is why mining gets harder over time without anyone deciding it should. More machines join, blocks come faster for a while, the adjustment kicks in, and the pace returns to ten minutes with each miner’s share of the reward now smaller. Mining is a competition against everyone else in the room, and the room keeps growing.

Mining moved from laptops to warehouses, and solo miners joined pools

In 2009 mining meant running the Bitcoin program on an ordinary computer. The Bitcoin Wiki traces what happened next: graphics cards, which turned out to be “drastically faster and more efficient” at the repetitive hashing than a general-purpose processor; then FPGAs, reprogrammable chips that drew less power; then, from 2013, ASICs, chips built to do nothing but SHA-256, which the wiki describes as “vastly faster than all previous technologies” for the power they consume. Each step made the previous one unprofitable.

The other change was social. As the network grew, the wiki records, individual miners “found that they were working for months without finding a block.” The answer was pooling: miners combine their hashing power, the pool submits the blocks, and members receive smaller, more frequent payments in proportion to what they contributed. The developer guide describes both modes, solo mining with all the reward and all the variance, and pool mining with steadier income. How pools are organized, and why their size worries people, is the subject of the mining pools article.

Why all that effort protects the ledger

It is fair to ask why a currency needs warehouses of hardware. The answer is in what the work buys. Because every block carries a proof of work, and every later block builds on it, an attacker who wants to change an old block “would have to redo the proof-of-work of the block and all blocks after it and then catch up with and surpass the work of the honest nodes,” in the white paper’s words. The cost of mining is the cost of rewriting history.

The paper also argues that the reward keeps miners honest. Someone with more hashing power than everyone else combined “would have to choose between using it to defraud people by stealing back his payments, or using it to generate new coins,” and ought to find the rules more profitable than undermining them. Whether that argument holds up is a live question, and the site’s dossier on how Bitcoin’s blockchain works follows a transaction from a wallet, through a miner, to a confirmation, with the rules checked at every step.