Merkle roots, plainly

Status — in design. Nothing is shipped, nothing is sold, and there is no customer.

A Merkle root is one short value that stands for an entire collection of documents, in a way that changes if any of them changes.

How it is built

1. Every document gets a fingerprint. A hash of its contents, plus its identifier, its licence and its length. Change one byte of the document and the fingerprint changes completely.
2. Fingerprints are combined in pairs, each pair producing another fingerprint.
3. Repeat until one value is left. That is the corpus root.

The whole tree for a corpus of millions of documents is a few megabytes. The root itself is a single short string.

Why that is useful

Because the root depends on every document, it works as a seal. A receipt issued under root R is a statement about the corpus as it was when R was computed. If anyone alters a document, adds one, or removes one, the root changes — and the old receipt no longer verifies against the new corpus.

Nobody has to trust that the corpus was unchanged. They can check.

Versioning, not overwriting

Every time the corpus grows, a new root is computed and appended to a log alongside the previous one. Roots are never replaced, so a receipt from six months ago stays checkable against the root it was issued under. The history is the record.

What we are not building

Not a blockchain. A root can be anchored to an existing public timestamp service if third-party proof of when is needed, which is cheap and requires running no chain of our own.