Architects sort buildings into types — basilica, courtyard house, tower, shed — not because every basilica is identical, but because grouping by shared organisational logic lets a designer borrow solved problems instead of solving spatial arrangement from zero each time. A librarian classifying a new book into the 500s instead of the 900s is doing the identical intellectual move: this object resembles that group more usefully than it resembles this other group.
Types are arguments, not facts
Neither system is neutral. Deciding a building is "a library type" rather than "a civic hall type" shapes what precedents get consulted and what failures get avoided; deciding a book is sociology rather than history shapes who ever finds it on a shelf browse. Every classification is a small argument about what a thing fundamentally is, dressed up as administrative housekeeping.
When the type doesn't fit
Both disciplines have a name for the failure mode: a building or a book that genuinely belongs to no existing category, forced into the nearest available one because the system has no room for "none of the above." Architectural history is full of hybrid buildings that later got their own type once enough of them existed to justify the category. Library science handles the same problem with cross-references and multiple subject headings — an admission that a single shelf location was never going to be enough.
A good taxonomy doesn't just describe the world as it is. It quietly decides what counts as similar enough to matter.
Tagging systems on modern websites inherit this exact tension, usually without anyone on the team having read a word of library science.