N·01 · 2 February 2025 · archives, memory, notation
On keeping an index
Why the back of the book was always the interesting part.
I have a bad habit of reading indexes first. Before the introduction, before the first chapter, I flip to the back and read the alphabetical list of everything the author thought was worth pointing at. It takes four minutes and tells you more about a book's real preoccupations than the blurb does. You can see where the density is. You can see which names recur, which ideas got three sub-entries and which got one, and what the author considered a topic at all.
An index is an unusual kind of document because it is entirely secondary and entirely opinionated at the same time. It contains nothing of its own. Every line points somewhere else. And yet two indexers given the same book will produce different indexes, because the work is not extraction — it's judgement about what a future reader will come looking for without knowing the words the author used.
The cross-reference
The best part is see also. A plain index tells you where something is. A good one tells you what else you should probably read while you're in the neighbourhood. It's the indexer admitting that the categories leak — that the chapter on maps is partly a chapter on memory, and there's no clean place to file that fact except sideways.
Software mostly gave up on this. We got search instead, which is better at questions you can already phrase and much worse at the thing an index does best: showing you the shape of a collection you haven't read yet. Search answers. An index describes. You can browse an index in a way you cannot browse a search box, because the index is already an argument about what's in here.
Keeping one
So I started keeping one. Not a feed, not a folder of bookmarks that I would never open again, but an actual alphabetical index of the things I've read and kept, with the topics chosen by hand and the cross-references falling out of the overlaps.
The discipline is small but real. Every time something goes in, I have to decide what it's about — which is a different and harder question than whether I liked it. Two or three topics, no more, chosen from the ones already in use if possible. That constraint does the work. New topics only get created when nothing existing will honestly hold the thing, and by then I've usually understood why.
A year in, the index has started telling me things. There's a cluster I didn't know I had. There are topics with one lonely entry that I should either feed or fold into something else. It has become, in the way these things do, a slightly more accurate picture of my attention than I would have given if you'd simply asked me.