Proof of Work Explained Simply
Proof of work explained simply: why Bitcoin needs computing work, how the hash lottery works and how it differs from proof of stake.
Proof of work is the method by which everyone in Bitcoin agrees on one shared order of transactions, without a boss. Miners use computing power to roll the dice for the right to add the next block. Every block costs real electricity, which makes rewriting the past extremely expensive. That is exactly what protects you from someone spending the same bitcoin twice.
This article explains the principle behind Bitcoin. How miners operate and what they earn is covered in Bitcoin mining explained. Here the question is: why work at all?
The problem: who decides which payment came first?
Digital data can be copied at will. Money must not work that way. If you send 1 bitcoin to Anna, you must not be able to send the same bitcoin to Ben at the same time. This problem is called the double spend.
At your bank the solution is simple: the bank keeps the account, executes the first transfer and rejects the second. That works as long as you trust it.
Bitcoin has no bank. Thousands of nodes around the world each keep their own copy of the ledger. Messages take different amounts of time to cross the internet. A node in Zurich may see your payment to Anna first, a node in Tokyo the one to Ben. Each payment is valid on its own, just not both together. Someone has to say which one counts, and everyone has to stick to it.
A majority vote sounds obvious. But on the internet anyone can create as many identities as they like. Whoever starts a million fake nodes would have a million votes. Satoshi Nakamoto solved this in 2008 in the Bitcoin white paper: the vote is not one per head but one per unit of computing power. Computing power cannot be faked. You have to buy it and run it on electricity.
How rolling the dice with hashes works
The hash: a fingerprint for data
A hash function turns any data into a number of fixed length. Bitcoin uses SHA-256; the result has 256 bits and is usually written as a 64-character string of digits and letters. Three properties make it useful:
- Same input, same hash. Anyone can recompute a hash in a fraction of a second.
- Smallest change, completely different hash. Change a single character and the result looks freshly rolled.
- No way back. You cannot work out the input from a hash. If you want a particular hash, all you can do is try.
The lottery
A miner packs new transactions into a block. Then they compute the hash of the block header. The block is only valid if this hash is smaller than a target. You can see this because a valid block hash starts with many zeros.
Almost always the hash is too large. Then the miner changes a number in the block header, the nonce, and computes again. And again. And again. On average the whole network needs around 640 sextillion attempts (as of November 2026) before one fits.
This is a lottery, not a puzzle. There is no clever way to compute the answer; every attempt has the same small chance. Whoever has twice the computing power draws twice as many tickets and wins twice as often on average. Nothing more.
The trick lies in the asymmetry: finding is enormously costly, checking is trivial. Every node, even a small computer at home, computes a single hash and sees at once whether the block is valid. The work is thereby proven without anyone having to take the miner's word for it.
How the target works
A block is valid if: SHA-256(SHA-256(block header)) ≤ target
A hash is a number between 0 and 2²⁵⁶ ≈ 1.16 × 10⁷⁷.
The largest allowed target (difficulty 1, January 2009) is 0xFFFF × 2²⁰⁸ ≈ 2.70 × 10⁶⁷.
Target = largest target ÷ difficulty
Chance per attempt = target ÷ 2²⁵⁶ ≈ 1 ÷ (difficulty × 2³²)
With a difficulty of about 148.4 trillion (as of 12 November 2026):
Target ≈ 2.70 × 10⁶⁷ ÷ 1.484 × 10¹⁴ ≈ 1.82 × 10⁵³
Chance per attempt ≈ 1 ÷ (1.484 × 10¹⁴ × 4.295 × 10⁹) ≈ 1 ÷ 6.4 × 10²³
Attempts per block on average ≈ 6.4 × 10²³, about 640 sextillion.
Leading zero bits: log₂(6.4 × 10²³) ≈ 79. A valid hash therefore starts with about 79 zero bits, in the usual notation with 19 or more zeros.
Hash rate ≈ 6.4 × 10²³ ÷ 600 seconds ≈ 1.06 × 10²¹ attempts per second, about 1’060 EH/s. Because blocks recently came more slowly than every 10 minutes, mempool.space estimates the hash rate lower, at 936 EH/s.
You can see the newest blocks with their hashes here. Look at the zeros at the start:
- Block 975’721Previous fingerprintPaymentsOwn fingerprint
- Block 975’722Previous fingerprintPaymentsOwn fingerprint
- Block 975’723Previous fingerprintPaymentsOwn fingerprint
Why it stays at 10 minutes
If more miners join, blocks would be found faster. That is why Bitcoin adjusts the target every 2’016 blocks, roughly every two weeks. How exactly is explained in The difficulty adjustment: the heart of Bitcoin. For proof of work only one thing matters: however much computing power is in the network, a block always costs the share of the total work that falls on 10 minutes.
Why the work protects history
Every block contains the hash of its predecessor. If someone changes an old transaction, the hash of that block changes. Then the next block no longer fits, nor the one after it. Whoever forges a block has to recompute every later block, with the full work for each one.
And that is not enough. While the forger computes, the honest miners keep building on the real chain. He has to not only catch up with them but overtake them.
The calculation from the white paper
In section 11 of the white paper, Satoshi works with:
p = share of honest computing power, q = share of the attacker, z = number of blocks the honest chain is ahead.
If the attacker has less than half (q < p), his chance of ever catching up falls exponentially with each block: it is (q ÷ p)^z when he is already z blocks behind.
Values from the white paper for a chance below 0.1 %:
q = 10 % → z = 5 · q = 20 % → z = 11 · q = 30 % → z = 24 · q = 40 % → z = 89 · q = 45 % → z = 340
For q = 10 % and z = 6 the exact formula gives 0.0002428, about 0.02 %.
Computing power in the example: 93.6 EH/s ÷ 200 TH/s = 93’600’000 TH/s ÷ 200 = 468’000 machines. Power: 93’600’000 TH/s × 15 to 25 J/TH ≈ 1.4 to 2.3 GW.
What an attacker with more than 50 % can and cannot do
With more than half of the computing power, often called a 51 % attack, the attacker wins the race in the long run. He can then:
- reverse his own payments, that is, double spend,
- keep particular transactions out of his blocks.
He cannot:
- spend other people's bitcoin, because he lacks their private keys,
- create more new bitcoin than the rules allow,
- change the rules. Every node rejects invalid blocks, no matter how much work is in them.
So the 21 million limit holds even during such an attack. Why, is explained in Why there are only 21 million bitcoin. Smaller proof-of-work coins show that the danger is real: Bitcoin Gold and Ethereum Classic were attacked this way several times between 2018 and 2020. Their total computing power was small enough to rent. For Bitcoin there are nowhere near enough idle machines to rent.
There is also an economic argument. Whoever invests that much in machines earns the most by playing honestly. A successful attack would destroy trust, and with it the value of the very bitcoin in which he receives his reward.
The chain with the most work wins
Sometimes two miners find a block at almost the same time. Then for a short while there are two versions of the chain. Nodes keep building on the version they saw first. As soon as the next block is found on one side, that side is longer, and everyone switches to it. The other block is discarded and its transactions go back to the waiting area.
The rule is often called "the longest chain wins". Strictly speaking, what counts is not the number of blocks but the total work: the sum of the attempts contained in all blocks. A chain with many easy blocks does not beat one with fewer but harder blocks. So nobody can score with a home-made, artificially easy chain.
The nice part: a new node does not need to ask anyone which chain is the right one. It downloads every block since 2009, checks each one itself and follows the valid chain with the most work. No further trust is needed.
Proof of work and proof of stake compared
The best-known alternative is proof of stake. Ethereum switched to it in September 2022. Instead of computing power, participants lock up coins as a deposit. Whoever deposits more is chosen more often to propose the next block. Whoever cheats loses part of their deposit.
Both methods solve the same task. They have different strengths:
| Proof of work | Proof of stake | |
|---|---|---|
| What protects the network? | Computing power and electricity, costs outside the network | Deposited coins, value inside the network |
| Energy use | high, by design | low. Ethereum cites more than 99.9 % less since the switch |
| Who can take part? | Anyone with a machine and electricity, even without coins | Whoever has enough coins (32 ETH per validator on Ethereum) or deposits them through a provider |
| New coins go to | miners, who carry ongoing costs | those who have deposited coins |
| A new node finds the right chain | on its own, by the most work since the first block | needs a recent, trusted checkpoint |
| Finality | probabilistic, grows with every block | explicitly final after a fixed time, about 13 minutes on Ethereum |
| Cheaters lose | electricity and machine time | part of their deposit |
To be fair: proof of stake saves a lot of energy, and explicit finality is practical. The fundamental difference lies elsewhere. In proof of stake, those who already own coins decide. Security depends on how the coins are distributed, and the large holders collect the reward for depositing. In proof of work, anyone who wants a say has to keep bringing in something from outside: electricity and machines. Owning many bitcoin does not, by itself, give anyone a vote on the order of blocks.
For money that should never belong to anyone in particular, we think that is the more important point.
The electricity is the point, not a bug
Bitcoin is often accused of wasting electricity. From the point of view of proof of work, that is a misunderstanding. If a block were cheap to produce, a forgery would be cheap too. The cost of the work is exactly what an attacker would have to overcome.
The level of consumption does not follow the number of transactions but the reward. Miners invest in machines and electricity for as long as it pays given the reward of new bitcoin plus fees. If the value of the reward rises, the work rises, and with it the security. With every halving the new bitcoin per block fall, and over the decades fees have to carry a larger share.
How much electricity that actually is and where it comes from is worked out soberly in Bitcoin's energy use: facts instead of headlines.
What proof of work means for you as a bitcoin holder
You do not need to mine to benefit. Reversing a payment with a few confirmations costs more than almost any attack could bring in. And nobody gains power over the order of blocks just because they own a lot of bitcoin. The rules are still set not by the miners but by the nodes. With your own Bitcoin node you check every block yourself.
Frequently asked questions
What is proof of work, explained simply?
Proof of work is proof that someone has done computing work. In Bitcoin, miners search by trial and error for a hash below a target. Whoever finds it may add the next block. Anyone can check with a single calculation that the work is real.
Why does Bitcoin need proof of work?
Without a central authority, the network has to decide which of two conflicting payments is valid. Proof of work sets the order and makes it very expensive to change it later. That is how Bitcoin prevents anyone from spending the same coins twice.
Is proof of work the same as mining?
Almost. Proof of work is the principle, mining is the activity. Miners do the work and receive new bitcoin and fees for it.
Is proof of stake better than proof of work?
Proof of stake needs far less energy and offers fast, explicit finality. In return, those who already own coins decide, and new nodes need a trusted checkpoint. Bitcoin stays with proof of work because its security rests on costs outside the network.
Can Bitcoin drop proof of work?
Technically it would be a change of the rules that all nodes would have to adopt. A switch has no visible majority among Bitcoin users and would more likely split the network. As of November 2026 there is no serious proposal for it.
Sources
- Satoshi Nakamoto: Bitcoin: A Peer-to-Peer Electronic Cash System, 31 October 2008, sections 2, 4, 5 and 11
- Adam Back: Hashcash, a Denial of Service Counter-Measure, 2002
- Bitcoin Developer Guide: Block Chain, Proof of Work
- Bitcoin Core: target and proof-of-work checks (pow.cpp)
- ethereum.org: Proof-of-stake and Ethereum energy consumption
- mempool.space: Hash rate and difficulty, as of 12 November 2026
Not financial, legal or tax advice. Models are no guarantee.
The Bitcoin Power Law newsletter
An email when we publish a new article or calculator. Every issue has a one-click unsubscribe link.