What the base promises

A number like “all documents indexed” is a claim, and claims are cheap to state and expensive to verify. The knowledge base does not simply remember having indexed something once and repeat that memory forever. It re-checks its own coverage on a schedule, and reports what it actually finds — which is a slower, less flattering habit than trusting a cached count, and a more honest one.

Fig. 1.2 — A trustworthy base is not one that claims to be complete. It is one that keeps checking, and says so.

Put plainly, for a reader with no technical background: this is the difference between a filing cabinet with a sign on it that says “everything is filed” and a filing cabinet that is actually opened and counted every so often, with the count posted where you can see it. The first is a promise. The second is a fact, refreshed on a schedule, and the base is built to prefer facts.

This matters more than it sounds like it should, because everything else in this book — the graph in Part II, the answers in Part III, the workflows in Part IV — is only as good as the material underneath it. A system that lies to itself about what it has read will eventually lie to you about what it knows, however confidently it phrases the answer.