How does a blockchain work, and why is it so hard to change?

What is a blockchain, in plain English?

Think of a blockchain as a notebook of transactions that thousands of people keep a copy of at once. When someone wants to add a line, everyone checks it against their own copy, and only lines that pass are written in. No one owns the notebook, and no one can quietly tear out a page.

The U.S. National Institute of Standards and Technology (NIST) puts it more formally: blockchains are digital ledgers that are tamper evident and tamper resistant, run in a distributed way without a central repository and usually without a central authority². A ledger is simply a record of transactions. What is new is who keeps it: not one bank, but a network of independent computers called nodes.

Chaining records is older than crypto: NIST notes that in 1991 a signed chain of information was used as a ledger proving a set of documents had not been altered². The ideas were applied to electronic cash in a 2008 paper by the pseudonymous Satoshi Nakamoto, and the Bitcoin network followed in 2009². Nakamoto announced the paper to a cryptography mailing list on 31 October 2008⁵, and Bitcoin Core's code dates the very first block, the genesis block, to 3 January 2009 (18:15 UTC)⁶.

What is inside a block?

A transaction is an interaction between parties; with a cryptocurrency it is usually a transfer of coins from one user to another². Rather than add transactions one at a time, the network gathers them into batches called blocks. Ethereum's documentation explains why: batching lets every participant stay in sync and agree on one precise history³.

Each block has two parts. The body holds the list of transactions. The header is a short summary. NIST lists the typical header fields as the block number, the hash of the previous block's header, a hash that represents all the data in the block, a timestamp, the block size and a nonce, a number used in mining².

How do hashes chain the blocks together?

A hash function turns any input into a short, fixed-length output called a digest. NIST describes three key properties: you cannot work backwards from the digest to the input, and it is computationally infeasible to find a second input with the same digest, or any two inputs that share one².

Now the clever part. Every block stores the hash of the previous block's header². If someone edits a transaction in an old block, that block's hash changes. The next block still points to the old hash, so the link breaks, and so does every link after it. NIST notes that this makes altered blocks easy to detect and reject².

Who decides which block comes next?

Because there is no central operator, the network needs a rule for agreeing on the next block. That rule is the consensus mechanism. NIST describes several models, including proof of work, proof of stake, round robin, proof of authority and proof of elapsed time². The two you will meet most often are the first two, compared in our explainer on proof of work vs proof of stake.

If two versions of the chain appear, Bitcoin nodes treat the longest chain as correct¹. Bitcoin also adjusts the puzzle's difficulty every 2,016 blocks so that blocks arrive roughly once every ten minutes². Ethereum now uses proof of stake, with time split into 12-second slots in which one validator is chosen to propose a block³.

Why is a blockchain so hard to change?

To rewrite history on a proof-of-work chain, an attacker would have to redo the work for the edited block and every block after it, and then overtake the honest network as it keeps adding blocks. The Bitcoin paper shows that the attacker's chance of catching up falls exponentially as the number of blocks to catch up grows¹, provided honest participants control most of the computing power¹.

That proviso matters. NIST lists a "51% attack" as a known risk: someone who gains enough computing power (proof of work) or enough staked funds (proof of stake) could in principle control which blocks are accepted². NIST also stresses that changing a recorded transaction is extremely difficult but not technically impossible².

Participants differ too. A full node stores the whole chain and checks transactions; a lightweight node keeps no full copy and relies on full nodes to relay its transactions². Our guide to crypto wallets and keys explains what a wallet app actually stores.

What do beginners get wrong about blockchains?