Every time someone calls a website's structure its "architecture," they are cashing a cheque written centuries earlier by librarians. The library was the first large-scale attempt to organise more information than any one person could hold in their head, and to make it retrievable by strangers. That is also, more or less, the job description of the internet.
What the shelf solved
A library solves retrieval through three overlapping systems: a fixed physical address for every object (the shelf), a classification scheme that groups related objects (the call number), and a separate index that lets you find the address without walking the stacks (the catalogue). Split retrieval from storage, and you can reorganise the index — add new subjects, cross-reference, correct errors — without moving a single book. Search engines inherited exactly this separation: the page does not move when you change how it is found.
Where the metaphor breaks
A book can only sit on one shelf at a time. A web page can belong to a hundred categories simultaneously, linked from a thousand directions, with no single "true" location. The internet did not just digitise the library; it removed the constraint that gave library classification its discipline. That discipline is worth remembering — not as a rule to enforce online, but as a reminder that a structure without constraints tends toward no structure at all.
The catalogue was always the first search engine — it just had a longer index time.
Studios working on digital products still borrow from library science more than they admit: primary navigation is a classification scheme, breadcrumbs are call numbers, and a sitemap is nothing more than a catalogue rendered as a document.