Coordinates cannot do that — a Word document is a flow of paragraphs and has no pages until it is laid out, and where they fall depends on the fonts the machine happens to have. So an address names a node of the model:

apertura:1:9f3ac21b:body/t2/w4/c1/b0/r0+12
└ scheme ┘ │ hash  └ flow ┘└──── steps ────┘└ offset
           version

The eight hex digits are a hash of the file’s bytes, and they are why this can be trusted: an address arriving from a different file is detected rather than quietly resolved to whatever node happens to sit at that path. Silent wrongness is how every naive offset scheme fails, and it fails in production rather than in tests.

An anchor is not one address but three, in decreasing precision, following the W3C Web Annotation model. They fail at different times — the first when somebody saves the file, the second when the extraction options change, the third only when the words themselves go — and the resolver reports which one matched, so an application can see its documents drifting before a user reports a highlight in the wrong place.

// Once, when the document is ingested
const doc = await extract(bytes);
for (const chunk of doc.chunks({ maxTokens: 512 })) {
  await index.add({ text: chunk.text, meta: chunk.selectors });
}

// Later, when a question is answered
const hit = await index.search(question);
search.show(hit.meta);   // scrolled to and highlighted