What is a blockchain? NIST says tamper evident, not immutable
A blockchain is a shared ledger of transactions grouped into blocks, each one cryptographically linked to the block before it. NIST's own report deliberately dropped the word immutable, and the reason it did is the most useful thing to understand about how these systems work.
By Zain
Published · 8 min read
A blockchain is a shared ledger. NIST's definition, from its Blockchain Technology Overview, is that blockchains are distributed digital ledgers of cryptographically signed transactions that are grouped into blocks.
Unpack that one clause at a time and you have the whole design. The transactions are signed, so you can tell who authorised them. They are grouped into blocks rather than recorded one at a time. The ledger is distributed, so no single party holds the only copy. And each block is linked to the one before it, which is where the word chain comes from.
What is actually inside a block
A block has two parts. The header holds metadata. The body holds the transactions.
The important field in the header is the cryptographic link to the previous block's header. Every block carries one, except the very first, which has nothing before it to point at.
That link is a hash. Hashing takes an input of almost any size and produces a fixed-size output, and the property that makes it useful here is sensitivity: the smallest change to the input produces a completely different digest. Change one character in an old transaction and that block's hash changes, which breaks the link held by the next block, whose own hash then changes, and so on to the end of the chain.
So tampering with history is not blocked. It is made loud.
Why NIST will not say immutable
“tamper evident and tamper resistant”
This is a detail worth more than most of what gets written about blockchains. NIST's draft used the word immutable. The published report does not. It says tamper evident and tamper resistant instead, and the change was deliberate.
The reason is that immutable means cannot be changed, and that is not what the technology provides. What it provides is that a change is detectable and expensive. On a chain with substantial participation, rewriting old history means redoing the work that secured every block since, faster than everyone else is adding new ones. That is not impossible. It is uneconomic.
Which also means the guarantee is proportional to how much is securing the chain. A large network makes rewriting history absurd. A small or new one makes it merely costly. Treating the word immutable as a property of the data structure, rather than of the economics around it, is how people end up surprised.
Distributed, and what that buys
A conventional database has an owner. Someone runs it, and everyone else trusts them to keep it honest. That arrangement works and most of the world runs on it.
A blockchain removes the owner and replaces them with a copy held by every participant, plus a rule for agreeing what gets added next. That agreement step is the hard part, and it is a separate subject from the data structure. We have covered how the two main approaches work in proof of work versus proof of stake.
Append-only is the other half. Records are added. They are not edited in place and they are not deleted. The ledger grows, and the history stays visible, which is a feature when you want auditability and a liability when someone writes something to it that should not be there.
Append-only is a constraint, not a feature list
Of NIST's four characteristics, append-only is the one that shapes day to day use the most, and it cuts both ways.
On the useful side, it is what makes an audit trail trustworthy. Nobody can quietly revise last year's entry, because last year's entry is secured by every block that came after it. For anything where the question is what happened and when, that is genuinely valuable.
On the other side, there is no delete. A mistake is corrected by appending a correction, and both the error and the fix stay visible forever. Anything written to a public chain that should not have been, whether a private key in a memo field or personal data in a contract, cannot be taken back. Designing around that constraint is most of the work in building on one.
What a blockchain does not give you
It does not make the data true. A cryptographically signed transaction recording a false fact is a securely recorded false fact. The chain guarantees that the record has not been altered since it was written, not that it was correct when it was written.
It does not make anything private by default. On a public chain, append-only and shared among participants means permanently visible to everybody.
And it does not remove trust so much as move it. You stop trusting an operator and start trusting the code, the cryptography, and the assumption that no single party controls enough of the network to rewrite it.
The short version
A blockchain is a shared, append-only ledger of signed transactions, grouped into blocks, with each block cryptographically linked to the one before it. The link means a change to old data is detectable. The distribution means no one party can make that change quietly.
NIST's careful phrasing is the thing to carry away. Tamper evident. Tamper resistant. Not immutable.
Frequently asked
What is a blockchain in simple terms?
It is a shared ledger of transactions, grouped into blocks, where each block is cryptographically linked to the one before it. NIST defines blockchains as distributed digital ledgers of cryptographically signed transactions that are grouped into blocks. The links mean that altering an old record changes everything after it, so tampering is detectable rather than hidden, and the distribution means no single party holds the only copy.
Is a blockchain really immutable?
Not according to NIST, which is unusually precise about this. Its draft report used the word immutable and its published report replaced it with tamper evident and tamper resistant. The difference matters: immutable means cannot be changed, while tamper resistant means changing it is hard and tamper evident means the change will be noticed. On a large chain, rewriting history is uneconomic rather than impossible. On a small one it can be merely expensive.
What is inside a block?
Two parts. A header carrying metadata, and a body carrying the transactions. The field that makes the structure work is in the header: a cryptographic link to the previous block's header. Every block carries one except the first, which has nothing before it to reference. That chain of links is what turns a sequence of blocks into a tamper-evident record.
What does hashing do here?
A hash function takes an input of almost any size and produces a fixed-size digest. The property that matters for blockchains is sensitivity: the smallest change to the input produces a completely different digest. So if someone edits an old transaction, that block's hash changes, which breaks the link stored in the next block, which changes that block's hash, and the break propagates all the way to the newest block. The tampering announces itself.
What does distributed actually mean?
That the ledger is held by many participants rather than by one operator, with no central repository or authority. NIST lists distributed and shared among participants among the defining characteristics. The practical consequence is that there is no single copy to edit. Changing the record means convincing or overpowering the network rather than persuading an administrator.
What is the difference between a blockchain and a database?
A conventional database has an owner who can edit and delete records, and users trust that owner to behave. A blockchain is append-only, shared, and secured cryptographically, so records are added rather than edited and the history stays visible. The trade-offs are real in both directions: databases are faster, cheaper and easier to correct, while blockchains make unilateral quiet changes impractical.
Does putting data on a blockchain make it true?
No, and this is the most common category error. The chain guarantees that a record has not been altered since it was written. It says nothing about whether the record was accurate when written. A cryptographically signed transaction recording a false fact is a securely recorded false fact. Anything a blockchain says about the physical world depends entirely on whoever put that information in.
Is data on a public blockchain private?
No. Append-only plus shared among participants means permanently visible to everyone with a copy, which on a public chain is anybody who wants one. Addresses are pseudonymous rather than anonymous, and because the full history is public and never deleted, analysis across transactions can link activity over time. Privacy on a public chain is something that has to be engineered deliberately, not a default.
What is the first block called?
The genesis block. It is the exception to the rule that every block header contains a cryptographic link to the header before it, because there is nothing before it to link to. Every other block in the chain traces back to it through an unbroken sequence of hashes, which is what allows any participant to verify the whole history from a single starting point.
Does every blockchain work the same way?
The data structure is broadly common: signed transactions, grouped into blocks, each linked by hash to the previous block. What varies is who may participate, who may write, and how the network agrees on the next block. Those choices produce very different systems with the same underlying shape, and they are also what determines how much security the tamper resistance actually delivers in practice.
Sources, and what is behind them
- NISTIR 8202: Blockchain Technology Overview, National Institute of Standards and TechnologyPaperIndependent of vendor
- IR 8202, Blockchain Technology Overview (publication record), NIST Computer Security Resource CenterDocumentationIndependent of vendor
- NIST Report on Blockchain Technology Aims to Go Beyond the Hype, NISTPress reportIndependent of vendor