What a blockchain is
A list, in order, where each entry contains a fingerprint of the one before it. That is the entire idea, and it is simpler than the word suggests.
Take a page of transactions. Compute a fingerprint of it — a short string that changes completely if any character of the page changes. Write that fingerprint at the top of the next page. Repeat.
That is a blockchain. Everything else is detail about who writes the pages and how disputes are settled.
Why chaining matters
Alter one transaction in block 500 and block 500's fingerprint changes. Block 501 contains the old fingerprint, so it no longer matches, and nor does 502, and so on to the present. Tampering is not prevented — it is made immediately visible to anyone holding a copy. That is the actual security property.
The vocabulary
- Block — a batch of transactions, plus a reference to the previous block and some metadata.
- Hash — the fingerprint. See hashing.
- Height — how many blocks deep a block sits. Block height 800,000 means 800,000 blocks preceded it.
- Confirmation — a block built on top of the one holding your transaction. More confirmations, more work to undo.
- Mempool — the waiting room of broadcast transactions not yet in any block.
Why blocks at all
Batching is a network design compromise. Agreeing on one transaction at a time across thousands of machines would be impossibly chatty; agreeing on a batch every few minutes is tractable. Bitcoin targets roughly ten minutes per block. Ethereum settled on about twelve seconds after the Merge.
The interval is a trade-off, not a technical limit. Shorter blocks mean faster confirmation and more frequent disagreement about which block came first.
What a blockchain is bad at
Every participant stores everything and verifies everything. Compared with a conventional database, throughput is lower by orders of magnitude and cost per record is higher by more. You accept that in exchange for removing the administrator. If your application has a trusted administrator — and most do — a blockchain makes it worse in every measurable way while adding complexity.
This is why a great many corporate blockchain projects launched in the late 2010s quietly ended. They applied a solution for the absence of a trusted party to situations where a trusted party was present, willing, and cheaper.
The part people get wrong
A blockchain guarantees that data has not been altered since it was recorded. It guarantees nothing about whether the data was true when recorded. Put a false statement on-chain and you have an immutable false statement. Nothing about the technology verifies the world — which is exactly the problem oracles exist to address, and only partly solve.