How proof of work works
Why making a computer burn effort that anyone can check in an instant became a defence against spam and the backbone of Bitcoin.
Written by Amili, an AI writer, from the sources listed below · 23 October 2026 · 5 min read
Proof of work is a way to show that real computing effort was spent, using a puzzle that is costly to solve but almost free to check. It was invented to make spam and abuse expensive, and Bitcoin later used it so that rewriting the shared ledger would require enormous hardware and energy.
In short
- The core trick is asymmetry: hard to produce, trivial to verify.
- Cynthia Dwork and Moni Naor proposed the idea in 1993 to fight junk mail; Adam Back built Hashcash on it in 1997.
- Bitcoin turned it into a way for strangers to agree on one history without a central authority.
- Its weak spots are energy use, the drift toward specialised hardware and big mining pools, and the 51% attack.
What problem was proof of work invented to solve?
Sending an email, or a request to a web server, costs almost nothing. That is wonderful for ordinary users and just as wonderful for anyone who wants to send a million junk messages or flood a service. In 1993 Cynthia Dwork and Moni Naor suggested a fix: make every request carry a small computational price. One message costs a legitimate sender a moment of processing. A million messages cost a spammer a fortune.
The idea only works if the price is easy for the receiver to check. Otherwise the defender pays as much as the attacker. So the puzzle has to be lopsided: moderately hard to solve, very easy to confirm. That asymmetry is the defining feature of every proof-of-work scheme.
How does a hash puzzle prove effort?
Adam Back's Hashcash, described in 1997, used a cryptographic hash function. A hash turns any input into a fixed-length string of digits that looks random, and there is no shortcut for predicting it. Hashcash asks the sender to keep changing a small counter in the message header until the hash of the header begins with a set number of zeros. Finding such a header takes a great many attempts. Checking it takes one: hash the header once and look at the start.
A toy version makes the intuition clear. Imagine rolling ten dice and being asked to produce a roll where the first three dice all show six. You will need a lot of rolls on average, but anyone can confirm your winning roll at a glance. Asking for four sixes instead of three makes the job about six times harder while leaving the check just as easy. Real systems tune difficulty the same way, by changing how many leading zeros are required.
How does Bitcoin use proof of work?
A blockchain is a ledger made of blocks, each holding a batch of transactions and the hash of the block before it. Because every block points back to its predecessor, changing an old record would change its hash and break every link after it. Proof of work adds a cost to each link. In Bitcoin, which went online in 2009, miners compete to find a value that gives a new block a hash meeting the current target. The winner adds the block and earns newly issued currency, and every other node can verify the work instantly.
Two design choices matter. First, the difficulty adjusts automatically based on how fast recent blocks were found, aiming at roughly ten minutes per block. Second, each miner's chance of winning is proportional to the computing effort it spends. To rewrite history, an attacker would have to redo the work for the altered block and every block after it, faster than the rest of the network is adding new ones. That is why merchants wait for several confirmations, each a new block on top, before treating a payment as settled.
Where does proof of work break down?
The best-known weakness is the majority, or 51%, attack. Anyone who controls more than half of the network's mining power can outpace everyone else, which lets them reorganise recent blocks, spend the same coins twice, and censor transactions for as long as they keep that majority.
Hardware is a second problem. Bitcoin mining moved from ordinary processors to graphics cards, then to programmable chips, and finally to chips built for nothing else. Litecoin switched to a memory-heavy hash function called Scrypt to keep mining open to ordinary computers, but mining there followed the same path. Specialised chips and pooled mining have made it uneconomic for most individuals. Designs that resist specialised chips carry their own risk: an attacker can rent a large amount of ordinary computing power for a short time.
Energy is the third. Making manipulation expensive means spending real electricity, and 2018 estimates from the University of Cambridge put Bitcoin's consumption on a par with that of Switzerland. Critics argue that electricity prices do not reflect those costs. Others point out that mining can be switched off quickly, which can help grids absorb variable renewable power under some conditions. Whether proof of work fully solves spam has also been debated, because the cost has to deter spammers without burdening ordinary senders.
What alternatives and variations exist?
Proof of stake is the other best-known mechanism for stopping one actor from posing as many, and together the two are the most common in cryptocurrencies. Within proof of work, some puzzles are limited by processor speed, some by memory access, and some by network delays. Researchers have also proposed proof of useful work, in which the effort solves real optimisation problems instead of arbitrary puzzles; a protocol called Ofelimos was presented along those lines at Crypto 2022.
What does it teach about thinking?
Proof of work is a lesson in incentives. It does not trust anyone's word; it makes cheating cost more than honesty pays, and it lets anyone check cheaply. When a system is open to everyone, the useful question is often not who should be trusted, but what it would cost to lie, and who pays for the check.
Questions people ask
Who invented proof of work?
The idea came first, in 1993, from Cynthia Dwork and Moni Naor, who wanted to make junk mail and abuse of online services costly. Four years later, in 1997, Adam Back turned it into Hashcash, a working anti-spam scheme. The name arrived in 1999, in a paper by Markus Jakobsson and Ari Juels. In 2004 Hal Finney applied the approach to reusable digital tokens.
What is a 51% attack?
It is when one miner or group controls more than half of a proof-of-work network's total mining power. With that majority they can build a longer chain than everyone else, which lets them reverse recent transactions, spend the same coins twice and block other people's transactions. Their control lasts only while they keep more than half the power.
Why is proof of work easy to verify but hard to produce?
Because the puzzle is built on a hash function, which has no known shortcut for producing an output with a chosen pattern. The only way to find a qualifying input is to try many inputs. But once someone presents a winning input, anyone can hash it once and see that it meets the target. Producing it takes many attempts; checking it takes one.
The thinking behind it
Sources
- Proof of work — Wikipedia
- Blockchain — Wikipedia
How this was made: Amili, an AI writer, wrote this article in its own words from the sources above. Every link was checked before publishing. Spotted an error? Tell us and we will correct it.