How does a blockchain work, step by step?

What problem does a blockchain solve?#

Digital files are easy to copy. If money were just a file, you could send the same "coin" to two people. Bitcoin's original white paper names this the double-spending problem: the person receiving a payment cannot verify that an earlier owner did not spend the same coin twice^[1]. Banks solve it by keeping one master ledger. A blockchain tries to solve it without a single keeper.

The U.S. National Institute of Standards and Technology (NIST) describes blockchains as tamper evident and tamper resistant digital ledgers implemented in a distributed fashion — without a central repository — and usually without a central authority such as a bank, company or government^[2]. Tamper evident means changes are easy to detect. Tamper resistant means they are hard to make. Distributed means many computers, called nodes, each keep a copy.

What is a hash, and why does it matter?#

A hash is a fixed-length fingerprint of any piece of data. NIST explains that hashing applies a cryptographic hash function to data and calculates a relatively unique output, called a digest, for an input of nearly any size^[2]. The key property: even the smallest change to the input, such as a single bit, results in a completely different digest^[2].

Bitcoin's white paper uses SHA-256, a widely used hash function, as its example^[1]. You can see the effect yourself. We ran two almost identical sentences through SHA-256 in code. Only one character differs, yet the fingerprints share nothing recognisable.

How are blocks chained together?#

Transactions are grouped into blocks. NIST describes a block as having a header with metadata about the block, and block data containing a set of transactions^[2]. Crucially, every block header except the very first contains a cryptographic link to the previous block's header^[2]. That link is the previous block's hash.

Now the tamper evidence becomes clear. NIST spells it out: if a previously published block were changed, it would have a different hash, which would cause all subsequent blocks to have different hashes as well, since each includes the hash of the block before it^[2]. We built a three-block toy chain in code to show it. Each block's fingerprint is the hash of the previous fingerprint plus its own data.

Who decides which block comes next?#

If anyone could append blocks freely, people would publish conflicting versions. A consensus model is the rule for who may add the next block and which version everyone accepts. NIST describes two common ones^[2]:

• Proof of work — participants publish a block by being the first to solve a computationally intensive puzzle. Bitcoin uses this; its nodes are called miners^[3].
• Proof of stake — the chance to add a block depends on how much of the network's own token a participant has committed, on the idea that the more stake a user has invested, the more they want the system to succeed.

Bitcoin's white paper describes the puzzle as scanning for a value that, when hashed, produces a hash that begins with a number of zero bits^[1]. Finding such a value takes many tries; checking it takes one. Our toy version below asks for hashes starting with zeros in hexadecimal. Each extra zero makes the search about 16 times longer on average.

Where versions conflict, Bitcoin's rule is that the majority decision is represented by the longest chain, the one with the greatest proof-of-work effort invested in it^[1]. Rewriting history would mean redoing that work faster than everyone else — which is also why NIST lists a "51% attack", by someone obtaining enough computing power, as a known limitation^[2].

How does a transaction get onto the chain?#

Here is the life of a payment on a proof-of-work blockchain, following the sequence in Bitcoin's white paper, which begins: new transactions are broadcast to all nodes, and each node collects new transactions into a block^[1].

1. 1

   You sign a transfer

   Your wallet uses your private key to sign a message moving tokens from your address. NIST notes that an address is derived from the user's public key with a hash function^[2].

2. 2

   The transfer is broadcast

   Your wallet sends it to nodes, which pass it along to others.

3. 3

   Nodes gather it into a block

   Waiting transactions are bundled into a candidate block.

4. 4

   Consensus picks the block

   Under proof of work, the first node to solve the puzzle broadcasts its block.

5. 5

   Others check and build on it

   Nodes verify the transactions are valid and unspent, then start the next block on top of it.

NIST describes a full node as one that stores the entire blockchain and ensures transactions are valid^[2]. The private key in step 1 is the part you must protect — see our private key definition.

What can a blockchain not do?#

A blockchain is good at one narrow job: keeping a shared record that is hard to rewrite. It does not check whether the information put on it is true, whether a token's issuer will keep a promise, or whether a price is fair. It also cannot undo mistakes. NIST notes that if a user loses a private key, any digital asset associated with it is lost, because it is computationally infeasible to regenerate the same key^[2].

NIST also distinguishes permissionless blockchains, where anyone can read and write without authorization, from permissioned ones that limit participation to specific people or organizations^[2]. Public crypto networks such as Bitcoin are permissionless, which is why anyone can join — scammers included. Read the risks of crypto investing and Bitcoin basics next, or keep the short blockchain definition handy.

What mistakes do beginners make?#

1. Thinking "on the blockchain" means "true"

   The chain proves a record has not been altered since it was added. It says nothing about whether the record was honest in the first place.

2. Confusing the network with the asset

   A working blockchain does not make every token on it valuable. Judge each token separately.

3. Expecting a reversal button

   There is no central operator to cancel a mistaken or fraudulent transfer on a permissionless chain. Double-check addresses and amounts before you sign.

4. Assuming all blockchains work the same way

   Consensus models, speeds, fees and who may participate differ from network to network.