What Librarians Figured Out That Note Apps Forgot

Before Melvil Dewey published his 42-page pamphlet in 1876, most libraries organized their books by arrival order. New acquisition? Spine at the end of the shelf. The catalog was essentially a receipt log. If you wanted to find something, you described it to a librarian, who carried the real index inside their head. This made libraries navigable only for regulars who knew the staff and the collection. Everyone else was fishing in the dark.

Dewey changed this with a classification number, a string of digits that encodes a book's place in the universe of human knowledge, not its place in the acquisition queue. 510 is mathematics. 616 is disease. 822 is English drama. The number becomes an address, and the address is a function of content, not chronology.

Fifty-five years later, the Indian librarian S.R. Ranganathan formalized what Dewey had intuited into five laws. The fourth is the one that matters: Save the time of the reader. Not the cataloger. Not the librarian. The reader — the person arriving later, in a different mental state, trying to find something they half-remember. Every organizational decision, every cross-reference and subject heading, was evaluated against that single criterion: does this help the person who comes later?

Your notes app is a pre-Dewey library

Apple Notes organizes by last-modified date. Notion shows you a homepage of recently edited pages. Most Obsidian vaults, if you look at them honestly, are organized by when their owner started using Obsidian — a folder called "2023," a folder called "Projects," a folder called "Misc" containing everything that wasn't obviously one thing or another.

That's arrival order. It works well enough when you're writing notes, and terribly when you're trying to find them.

The problem is that you write in one mental state and retrieve in a completely different one. You wrote the note when the context was fresh. You need it weeks later, when all you can half-recall is what it was about. Research by Elsweiler and colleagues, tracking memory lapses in real users over a week, found that people consistently remember the contents of what they stored — a conversation, a decision, an idea — but not the formal descriptors: the file name, the folder, the tag. You remember the substance. You don't remember what you called it. This is exactly the wrong alignment when your search requires you to know those things.

McKinsey's 2012 research put a number on the consequence: knowledge workers waste roughly one day per week searching for information they already have somewhere. That figure has probably gotten worse, not better, because we've added more places to store things without improving how we retrieve them.

What the card catalog actually was

The insight at the heart of library science isn't classification. Classification is just the implementation. The real insight is that the document and its index are separate things.

A book lives at 616.85 regardless of when you acquired it or what you happen to call it in conversation. The card catalog — rows of small drawers, each card cross-referencing title, author, subject, related subjects — was a separate artifact entirely. It let you arrive from multiple directions: by author's name, by title keyword, by subject. You could find the same book whether you remembered who wrote it or only what it was about.

Note apps mostly don't separate these layers. The note's location is also its only index. Folder placement is classification. The note's title is its catalog entry. If you can't recall the folder or the title, you're doing full-text search and hoping the exact word you remember appears in the document.

OCLC printed the last physical library catalog cards on October 1, 2015. Exactly as the second brain movement was cresting. As Notion was gaining traction. As Obsidian's community was forming. We stopped making the things that worked at the precise moment we started building the replacements that don't.

What structure costs, honestly

I don't want to oversell this. Maintaining a real classification system is labor. Librarians are professionals. The Library of Congress Subject Headings runs to thousands of pages. Ranganathan's five laws are simpler than his colon classification system, which took decades to develop and remains genuinely difficult to apply.

This is why most note vaults collapse into chaos. The organizational maintenance costs more than the retrieval benefit feels worth. So the structure erodes, the folders multiply, the "Misc" drawer fills up, and eventually you stop filing things deliberately and just use search and hope.

The honest version of this problem isn't that people are lazy or undisciplined. It's that good cataloging requires time and attention precisely when you have the least of it — right after a meeting, mid-conversation, late in the afternoon. The professional librarian's job exists precisely because this work doesn't happen spontaneously.

Ranganathan's fifth law is the one nobody ever quotes: A library is a growing organism. Not a static archive. Something alive that requires tending. A cataloger's work is never done because knowledge keeps being created, and the index has to keep pace.

That's the part that really didn't transfer to personal knowledge tools. We got the storage. We got, occasionally, the search. We didn't get the idea that maintenance is part of the design — that a knowledge system without upkeep isn't a library. It's a filing cabinet no one's opened in years.

---

Asgeir Albretsen is the founder of Harbor.