Mousetext

It's ok to be confused

How to Sort Your Books

Let us say you own books. Where do you put them?

At the moment I write this, I own 402 books. This is the fewest I've had in a long time; I recently finished a ruthless cull, and sold and donated more than 100 books. It was a productive process (and I'm rolling in store credit), but it threw my shelving system into a tailspin. The question now is, what method should I use to rebuild?

You are probably familiar with a handful of common sorting systems. In bookstores, books are generally grouped by genre and subject, and then organized by authors last name. This scales down well to a personal library, and whatever categories best fit the collection and the collector.

In public and academic libraries we find Dewey Decimal or Library of Congress Classification. And, naturally, in some personal libraries as well:

But a home library need not be bound to external comprehensibility. It only requires that the people who live with it and regularly use it are able to navigate it. Books can be sorted by the order in which they were purchased, grouped by who owns them, piled up in the order they were taken out of a moving box in seeming randomness.

Humans are very bad at randomness; even a shelving system that was not intentionally designed has emergent order, or at least a reflection of small choices, however unthinking.

These systems, intentional or not, are all interfaces, ways for humans to find and pick up objects. A cataloguing system can't be implemented without considering how a person interacts with books as objects, and how the books interact with the space they inhabit.

At the most basic, books must be in reach, and they must physically fit on the shelves on which they are meant to sit. The theoretical system must map to actual bookshelves that exist in a room.

Consider books sorted by color. This is, perhaps, the most purely physical and spatial shelving system, setting content and metadata totally aside.


Source: slate.com

These books are first and foremost objects in a room, and unless you have a very busy household, chances are that your books spend much more time as objects in a room than they do as reading material[1].

Where things get interesting for me, though, is when systems take interest in physicality and in content. When you also map the content and subject matter of a book to physical placement.

Consider the Prelinger Library, an independent public library in San Francisco, created and curated by Rick and Megan Prelinger.


Prelinger Library

The collection is designed for browsing and serendipitous discovery; there is no catalogue with which to locate an item. Physical movement through the library is also intellectual movement. To quote the Prelinger Library's website:

The main shelves are organized according to the library’s unique geospatial taxonomy. This arrangement system classifies subjects spatially and conceptually beginning with the physical world, moving into representation and culture, and ending with abstractions of society and theory. It can be summarized as a walk through a landscape of ideas, from feet-on-the-ground to outer space.

I find that a much more mundane version of this connection between content and space happens organically in my own apartment. My cookbooks are on my kitchen counter, whatever I'm currently reading is piled on my nightstand. My grimoires are lined up in easy reach of the armchair where I like to sit and do research for grimoire.org.

But the rest of my books are sorted in their own particular way (or at least they were before I sold 20% of them), based on graph traversal.


My books, mid-reorganization

To make this possible (and to keep track of what I already own when I'm at bookstores), I catalogue and tag all my books on LibraryThing. This provides metadata about each book — the author, date published, characters, where it is set, subject matter, who recommended it to me, et cetera. I have a whole workflow of exporting, scraping, and MapReducing that gets me to a json file representing each book and what I know about it.

All of this data goes into a graph database, in which each book is a node, and the relationship between two books is weighed based on how much they have in common. For example:

  • The Fellowship of the Ring and The Two Towers, two books in the same series with numerous shared characters and other characteristics, have a very strong relationship, with the numerical weight of 61.
  • The Two Towers and The Hobbit, which are set in the same universe but are not direct sequels, have a weaker relationship, weighted at 39.
  • The Hobbit and Wintersmith are both young adult fantasy novels, and have a much more tenuous relationship, weighted at 6.

The order in which books are shelved is the order in which the graph is traversed using an algorithm that prefers to select books that are most similar to the books it has just selected.

The resulting system is meandering, exploratory, and sometimes jarring and confusing. Because I incorporate personal information into the graph, it will place together all the books that remind me of a certain friend, or group a novel with the nonfiction it inspired me to pursue.

There are, however, major problems with this system. It is difficult to know where to put a new book. Running the algorithm again with the updated data can mean that hundreds of books belong in new spots. Removing a book can create incongruities between the new neighbors.

It also treats book shelves as one long shelf, rather than a 2 dimensional unit where a book at the end of a shelf is much closer to the ones above and below it than the one at the start of the next shelf.

But I have the physical dimensions in my data. New systems may yet emerge.


Shoutout to everyone on twitter who told my about their own ways of sorting books. 💙

[1] Sorting books by color can also be a totally functional organization system at a small scale, which maps your visual memory of the book to its location. This may prove an easier connection to make than recalling information about the content of the book.

Mouse Reeve

Mouse is a software engineer at the Internet Archive, a historical occultism, and a true believer in nonsense. Nonsense is important.

San Francisco @tripofmice