Imagine a network of concepts, each concept naming a combination of words (symbols). Imagine searching for words belonging to that concept through a collection of digital documents containing text.
Then each query corresponds to a path through that network, path which reaches a collection of documents. Then each document found can, essentially, be represented (as an alternative to its current representation as a series of words) as a collection of paths through that network.
This means that, given a refined enough ontology, an arbitrary digital document is formally rewritten according to that ontology through the search process (by collecting and inverting the hits).
This ultimately means that searching means rewriting too.
A consequence of this is that we have, in principle, an alternative solution to semantic authoring (authors using semantic mark-up to write new documents). Another consequence of this is that old, semantically flat, documents can recover semantic depth over time while the public gets involved in this playfully.