A blockchain is a shared ledger that thousands of independent computers keep in sync, with no single operator in charge. Users sign transactions with private keys, miners or validators group them into blocks, and each block is linked to the one before it by a cryptographic hash, so later edits are obvious. A consensus rule decides which block comes next. The result is a record strangers can verify themselves and that is very expensive to rewrite.
What problem does a blockchain solve?
A blockchain lets people who don't trust each other agree on who owns what, without a company keeping the official record. Digital files copy perfectly, so digital money has an obvious flaw: what stops someone from spending the same coin twice? Banks solve this by keeping one master ledger and asking everyone to trust it.
Bitcoin, described in a 2008 whitepaper and live since January 3, 2009, took a different route. Everyone keeps a copy of the ledger, and everyone enforces the same rules for changing it. The US National Institute of Standards and Technology (NIST) calls blockchains "tamper evident and tamper resistant" ledgers. Those words are more accurate than "unchangeable," as we'll see below.
The five parts that make it work
A blockchain combines five ideas, and the first three come from the cryptography behind these networks.
Hashes
A hash is a fixed-length fingerprint of any piece of data. Change one character of the input and the fingerprint changes completely. Bitcoin uses a function called SHA-256. These are real outputs:
"Alice pays Bob 0.01 BTC"
29b769e0888d3af9aa1bd7fa84bbfab9cb156a4970b3e3ad82d773173c4a3665
"Alice pays Bob 0.02 BTC"
445000dfabc87935780f8ef2d571916ab39a44d8d1fb93fbc3a7fe48a7559a1a
Checking a hash is instant. Working backward from a hash to its input is not practical.
Blocks and the chain
A block is a batch of transactions plus a short header. A Bitcoin header is 80 bytes. It includes the hash of the previous block's header and a Merkle root, a single hash that summarizes every transaction in the block.
This is where the chain comes from. Alter an old transaction and its block's Merkle root changes. That changes the block's hash, which breaks the link stored in the next block, and so on up to the newest block.
Signatures
A digital signature proves that the holder of a private key approved one exact transaction. Anyone can check it using the matching public key, and nobody can forge it without the private key. Because the signature covers the transaction's details, nobody can change the recipient or amount afterward.
Your crypto wallet manages these keys. Many wallets derive them from a seed phrase, which is why losing one can mean losing the funds for good.
Nodes
A node is a computer running the network's software. A full node validates every block itself and takes nobody's word for it. A block that breaks a rule, such as paying its creator too much new money, gets rejected no matter who produced it.
Consensus
Consensus is the rule for choosing which valid block comes next. On Bitcoin, miners compete using proof of work. On Ethereum, validators who deposit 32 ETH take turns under proof of stake. Both make cheating costly, and our guide explains how proof of work and proof of stake compare.
Following one Bitcoin payment from signature to finality
Alice wants to pay Bob 0.01 BTC, and her wallet holds one unspent chunk of bitcoin, called an output, worth 0.05 BTC. Transactions must spend whole outputs, so she can't simply break off a piece.
- The wallet builds the transaction. Input: Alice's 0.05 BTC. Outputs: 0.01 BTC to Bob and 0.03999 BTC back to Alice as change. The missing 0.00001 BTC is the fee.
- The wallet signs it. Alice's private key produces a signature over the transaction. The key itself never leaves the wallet.
- The transaction spreads. Nodes check the signature and confirm the 0.05 BTC hasn't been spent. Valid transactions are relayed onward and wait in a holding area called the mempool.
- A miner picks it up. The miner adds it to a candidate block and hashes the header again and again, changing a number each time. At the difficulty set on October 16, 2026, a winning hash takes about 600 sextillion tries on average, and the whole network makes about 1 sextillion tries a second.
- A block is found. The winner broadcasts it and collects the fees plus new bitcoin, on a schedule that halves about every four years. Other nodes verify the hash in one step, recheck every transaction, and start building on top. Bob's payment now has one confirmation.
- Confirmations pile up. Blocks arrive about every 10 minutes on average. Each one buries Alice's transaction deeper, because rewriting it would mean redoing all the work stacked on top.
Bitcoin never declares a payment final. The odds of reversal just keep shrinking. The whitepaper calculated an attacker's chance of ever catching up from behind:
| Blocks the attacker is behind | Attacker with 10% of hash power | Attacker with 30% of hash power |
|---|---|---|
| 5 | 0.09% | 17.7% |
| 10 | 0.00012% | 4.2% |
Bitcoin's developer guide says high-value payments should wait for at least six confirmations, roughly an hour, and admits that number is "somewhat arbitrary."
Ethereum handles this differently. Validators vote on checkpoints, and when votes representing two-thirds of staked ETH link two checkpoints, the earlier one is finalized. That takes about 15 minutes. Ethereum's documentation says changing a finalized block would take an attack costing many billions of dollars.
Why "immutable" overstates it
Blockchains are hard to change, but NIST says they cannot be considered completely immutable. Three incidents show why:
- A software disagreement (March 2013). Bitcoin split into two chains when version 0.8 nodes accepted a large block that older nodes rejected. Two mining pools downgraded their software to end the split, so the network abandoned the chain that about 60% of hash power had been extending. At least one large double spend happened during the fork.
- A social decision (July 2016). After an attacker drained more than 3.6 million ETH from a contract called The DAO, Ethereum ran a hard fork that moved the funds to a recovery contract. Those who rejected the change kept the original chain going as Ethereum Classic.
- A majority attack (January 2020). Attackers rewrote recent Bitcoin Gold blocks and double-spent over $72,000 worth of tokens. Small proof-of-work chains are far cheaper to attack than Bitcoin.
The security of a chain rests on the cost of attacking it and on people's willingness to coordinate. It is strong, and it is not absolute.
How is a blockchain different from a database?
A database is controlled by whoever administers it, while a public blockchain follows rules that participants enforce on each other.
| Traditional database | Public blockchain | |
|---|---|---|
| Who can write | Whoever the administrator allows | Anyone who follows the protocol rules |
| Who can change history | The administrator | Nobody, short of an expensive attack or a community fork |
| What you trust | The operator and its auditors | Open rules you can check, plus the software and cryptography |
| Speed | Fast, limited mostly by hardware | Slow by design: about 10 minutes per Bitcoin block, 12 seconds per Ethereum slot |
| Cost to write | Very low | A fee per transaction, such as Ethereum's gas fees |
| Privacy | Access controls hide data | Public by default, with pseudonymous addresses |
| Fixing mistakes | Edit the row | No undo: the recipient has to send funds back |
| Main failures | Operator fraud, outages, breaches | Code bugs, lost keys, majority attacks |
Permissioned blockchains sit in between. Known participants run them, and NIST notes that their owners can replace any number of blocks if they choose. In practice they resemble a shared database with built-in tamper evidence.
When do you actually need a blockchain?
You need a blockchain only when several parties must share a record and nobody can be trusted, or wanted, as its keeper. NIST's list of suitable uses includes many distributed participants, no trusted third party, and a shared transaction history. Researchers Karl Wüst and Arthur Gervais published a similar analysis comparing blockchains with central databases.
A quick test:
- Do several parties need to write to the same record?
- Do they distrust each other, or want to avoid a middleman?
- Is a public, near-permanent record acceptable?
If any answer is no, a normal database is probably cheaper, faster and easier to fix.
Four common misconceptions
- "If it's on a blockchain, it's true." A chain can show that a record was made and not altered. It can't show that the record matches the real world. NIST calls this the oracle problem.
- "Blockchains are trustless." You still trust the cryptography, the developers, bug-free smart contracts, and that no group secretly controls most block creation.
- "Blockchains are anonymous." Most are public ledgers with pseudonymous addresses. Once an address is tied to you, its history can be read by anyone.
- "One chain can handle everything." Block space is limited, which is why layer 2 networks move activity off the main chain while relying on it for security.
You can explore more of these ideas in our guides to consensus.
Key takeaways
- A blockchain is a shared ledger kept in sync by independent nodes, with no central operator.
- Hashes link blocks together, so changing old data breaks every later link.
- Signatures prove who approved a transaction, and nodes reject anything that breaks the rules.
- Consensus picks the next block: proof of work on Bitcoin, proof of stake on Ethereum.
- Finality differs by chain. Bitcoin's reversal odds shrink with each block, while Ethereum finalizes in about 15 minutes.
- Blockchains are tamper resistant, not magic. Use one only when no party can be trusted to keep the record.
Sources
- Bitcoin: A Peer-to-Peer Electronic Cash System, Satoshi Nakamoto
- NIST IR 8202: Blockchain Technology Overview
- Bitcoin developer guide: transactions
- Bitcoin developer reference: block chain
- Ethereum.org: proof of stake
- BIP 50: March 2013 chain fork post-mortem
- MIT Digital Currency Initiative: 51% attacks
Get the weekly commit
New blockchain deep dives every week.
