Zero-knowledge proofs explained: prove you know the password without saying it
A zero-knowledge proof convinces someone that a statement is true without revealing anything else, such as the secret behind it. The idea dates from 1985; in crypto it powers shielded Zcash payments and ZK-rollups on Ethereum.
Behind the village pond there's a walled garden with one door, and the door opens only with a spoken password. Anna says she knows it. Grandma Rosa would like to believe her, but Anna doesn't want to say the password out loud, not even to Rosa.
So they play a game. The path around the pond splits left and right and meets again at the garden door. Rosa waits at the fork. Anna walks down one side, out of sight. Then Rosa shouts: "Come back by the left!" or "Come back by the right!" If Anna knows the password, she can always obey, going through the door if she has to.
After twenty rounds, Rosa is convinced. She still doesn't know the password. That's a zero-knowledge proof: you learn that the claim is true, and nothing else.
What it is
A zero-knowledge proof is a way for one side, the prover, to convince another, the verifier, that a statement is true without revealing why it's true. The idea was introduced in 1985 by Shafi Goldwasser, Silvio Micali and Charles Rackoff. Their paper defines zero-knowledge proofs "as those proofs that convey no additional knowledge" beyond the fact that the statement is correct (Goldwasser, Micali and Rackoff, 1989 journal version). Goldwasser and Micali received the 2012 ACM Turing Award, computing's top prize, with zero-knowledge proofs named in the citation (ACM, 2013).
How it works
Back to the pond, with numbers.
- One round. Someone who doesn't know the password can still guess which side Rosa will call. He walks down the left and hopes. He's right half the time.
- Two rounds. He has to guess right twice: one chance in 4.
- Ten rounds. One chance in 1,024.
- Twenty rounds. One chance in 1,048,576. A faker who passes twenty rounds is luckier than almost anyone you will ever meet.
Notice what Rosa never sees: the password, or even Anna at the door. All she sees is Anna appearing where she was asked to. That record would look the same if Anna and Rosa had agreed on the shouts in advance, which is exactly why it gives away nothing. It only convinces the person who did the shouting.
Practical systems such as the zk-SNARKs used by Zcash replace the pond with maths and produce one short proof that anyone can check, with no back-and-forth. The original paper's own examples were already number-theory problems, not ponds.
Why it matters
- Private payments on a public notebook. Zcash, launched on 28 October 2016, calls itself "the first practical application of zk-SNARKs, a specific type of zero-knowledge proof" (Zcash documentation). A shielded transaction "encrypts the sender and receiver addresses", and the amount is hidden too. The documentation calls SNARKs "the engine that can quickly and efficiently verify a transaction". The village can check the sums without reading the names.
- More payments per page. On Ethereum, ZK-rollups "bundle (or 'roll up') transactions into batches that are executed offchain" and post a validity proof that the batch was handled correctly (ethereum.org). The main chain checks one proof instead of re-running every transaction.
- Showing less of yourself. The same idea could one day prove you're over 18 without showing your birthday, or that you hold enough funds without showing your balance. Whether anyone builds it well is a separate question.
Where people get it wrong
- "Zero-knowledge means anonymous." Not on its own. Zcash's own documentation says transactions between transparent addresses "work just like Bitcoin" and are fully visible, and even shielded ones show that a transaction happened and the fee paid.
- "Every ZK-rollup is private." The name is a little generous. Ethereum's documentation notes these systems hide a prover's private inputs only "where a system implements the zero-knowledge property". Many use the proof mainly for speed.
- "Bitcoin uses zero-knowledge proofs." Bitcoin's transactions show addresses and amounts on every copy of the notebook. Privacy coins were built partly because of that.
- "A proof proves the project is honest." It proves one precise statement. If the rules being proved are wrong, the proof is a perfect certificate for the wrong thing.
Ask what is being proved
When a project puts "ZK" in its name, ask three questions. What exactly is being proved? What is hidden, and from whom? Who wrote and checked the rules behind the proof?
If the stranger at the gate can't answer, notice what he did manage: he proved he has nothing to tell you, without revealing a single detail. Zero knowledge, as advertised.
Frequently asked questions
What is a zero-knowledge proof in simple terms?
It's a way to convince someone that a statement is true without revealing anything else, such as the secret that makes it true. The classic picture is proving you know a password by passing a locked door on request, many times, without ever saying the password. The idea was introduced in 1985 by Goldwasser, Micali and Rackoff.
How can you prove something without revealing it?
By answering challenges that only someone with the secret could answer reliably. A faker might guess right once, with a 50% chance in the simple door game, but passing 20 rounds by luck has a chance of about one in a million. The verifier ends up convinced but learns nothing usable.
Which cryptocurrencies use zero-knowledge proofs?
Zcash, launched on 28 October 2016, uses zk-SNARKs for shielded transactions that encrypt sender, receiver and amount. On Ethereum, ZK-rollups use validity proofs to show that batches of transactions processed off the main chain were handled correctly.
Are zero-knowledge proofs the same as encryption?
No. Encryption hides data so that only a key holder can read it. A zero-knowledge proof shows that a statement about hidden data is true. Systems like Zcash use both: encryption hides the details and the proof shows the hidden payment follows the rules.
Does a ZK-rollup make my transactions private?
Not necessarily. Many ZK-rollups use the proof to show the batch is correct, so the main chain can check it quickly, while the transactions themselves stay visible. Privacy only exists where the system is designed to hide the inputs.

