Proof of work

Proof of work turns electricity into a number nobody can fake. A miner must find a hash below the target, which at difficulty 1 takes about 4.3 billion attempts on average and vastly more today, while a node checks the answer with two SHA-256 passes over 80 bytes. That asymmetry is what makes rewriting old history expensive rather than merely rude.

How it works

Proof of work is a deliberately one sided puzzle: brutally slow to solve, instant to check.

The puzzle is to find a header whose SHA-256 hash, computed twice, is numerically smaller than a threshold called the target. Hash functions offer no structure to exploit, so the only available method is to try, change a field, and try again. Each attempt succeeds with a probability equal to the target divided by the whole 256 bit output space, which makes the expected attempts per block equal to the difficulty multiplied by 4,294,967,296. At the difficulty the network has carried recently, above 120 trillion, that comes to roughly 500 billion trillion attempts for one block, shared across every machine on earth that is trying.

The idea is older than bitcoin. Adam Back proposed Hashcash in 1997 as a postage stamp for email: force the sender to burn a little processor time so that sending spam stops being free. What Satoshi Nakamoto added, in section 4 of the whitepaper, was to chain the stamps together, so the work proves not only that a message cost something but the order in which messages arrived.

Ordering is why nodes do not simply count blocks. A node follows the chain carrying the most accumulated work, summed header by header, which is why a short chain of very hard blocks beats a long chain of easy ones. Replacing a week old block means redoing that block's work and the work of every block after it, faster than the rest of the world extends the honest chain. The whitepaper works the probabilities out in section 11 and shows an attacker's chance of catching up falling away exponentially with each block added on top.

Why this matters when you buy bitcoin

Proof of work is the reason an exchange makes you wait for confirmations, and also the reason that wait is measured in minutes rather than left open ended.

Settlement here is probabilistic, not binary. One confirmation means one block of work sits on top of your payment; six means an attacker would have to outrun the entire network for about an hour to undo it. Exchanges turn that into a policy number: Kraken credits a bitcoin deposit after 4 confirmations, roughly forty minutes when blocks run on schedule. Nothing an exchange does can speed the chain up, and a platform that credits you instantly is lending you the balance and carrying the risk itself, which is a service, not a faster blockchain.

The same rule sets the honest boundary of the energy argument. Security here is bought with electricity and the bill is published: the Cambridge Centre for Alternative Finance estimated network demand at 16.09 gigawatts and annualized consumption at 141 terawatt hours on August 1, 2026, and its 2025 industry report attributed 52.4 percent of mining energy to sustainable sources. You can argue about whether that price is worth paying. You cannot argue that the protection is free, because a proof of work system that consumed nothing would be one that anyone could rewrite.

For a holder the practical rules are short. Wait for several confirmations on anything large, ignore the difference between zero and one confirmation when you are buying a coffee, and treat any product that promises instant final settlement as a custodian making a promise rather than a chain doing work.

Checking the genesis block by hand

The genesis block is the cleanest demonstration of proof of work, because you can verify it yourself with a hash function and no trust at all.

Its header, dated January 3, 2009, carries the nonce 2083236893. Hash those 80 bytes twice with SHA-256 and the result is 000000000019d6689c085ae165831e934ff763ae46a2a6c172b3f1b60a8ce26f. Difficulty was 1 then, and the difficulty 1 target begins with eight zero digits; the hash that was found begins with ten, so it cleared the bar with room to spare. Any laptop confirms all of this in under a millisecond, while producing it in the first place required the search. Cheap to check, expensive to create: that is the whole invention, and it is why a phone can validate a chain that cost billions of dollars of electricity to build.

Proof of work vs consensus

Proof of work is one rule; consensus is the entire rulebook. The consensus rules cover the 21 million supply cap, script validity, block weight, signature checks, timestamps, and much more, and a node enforces all of them before accepting a block. Proof of work answers only the narrow question of who may propose the next block and at what cost. A block with flawless proof of work that pays its miner 100 coins is thrown out instantly by every node, which is the detail people skip when they say miners control bitcoin. Miners choose the order of transactions. Nodes decide what counts as valid.

Proof of work vs SHA-256

Proof of work is the rule; SHA-256 is the ingredient it happens to use. SHA-256 is a hash function designed by the United States National Security Agency and standardized by NIST in 2001, with no original connection to bitcoin at all: the same function protects TLS connections, Git commits, and software package signatures. Bitcoin applies it twice to a block header and calls the output the block hash. A proof of work system can be built on a different function, and several other coins are. Changing the function changes which hardware wins the race, not what proof of work is.