Essay
Why I track the connections between books
A reading list tells you what someone finished. It tells you almost nothing about what they think. The interesting information is in the links.
A reading list tells you what someone finished. It tells you almost nothing about what they think. The interesting information is in the links.
I have kept a reading list for years, and for most of that time it did what reading lists do, which is very little. It recorded titles and dates. It was a receipt. When someone asked what I had been reading, I could tell them the last three books, and when someone asked what I had learned, I could usually manage one idea per book, badly.
The problem was not memory. The problem was that the list had no shape. A list is a sequence, and a sequence implies only one relationship between any two items: which came first. That is the least interesting thing about two books.
I read Dark Money and then, about a year later, The Power Worshippers. Separately, they are two competent books about two different subjects — the financing of the American right, and the rise of religious nationalism. Read within a year of each other, they are obviously the same book approached from opposite ends. One follows the money into the institutions; the other follows the belief into the same institutions and finds the money already there.
My reading list recorded them eleven months apart, with about ninety unrelated titles between them. It had captured the two least important facts about that pair: when I read them, and in what order.
That is when I started keeping the connections instead.
Every book in this catalog carries a small set of explicit links to other books on the shelf. There are 1,256 of them across 321 books. They are made by hand, one at a time, and they mean something specific — not "these are both about finance," which is what a tag is for, but something closer to an argument:
Tags cannot do this. A tag is a bucket, and buckets are symmetric and shallow — everything in `politics` is equally in `politics`. A connection is directional and specific. It says this book, because of this claim, points at that one.
Three things, and only one of them was expected.
The expected one: recall improves. Not because the summaries help — they help a little — but because retrieving a book by its neighbors is far easier than retrieving it by title. I cannot always name the book about the collapse of Washington Mutual. I can always get to it from the 2008 cluster.
The unexpected one: the connections make the gaps visible. When a book arrives and has nowhere to attach, that is information. It usually means one of two things — either I have wandered into a subject with no foothold on the shelf, which is worth knowing, or the book is thinner than it looked and does not actually engage with anything. Books that connect to nothing are, in my experience, disproportionately books that said nothing.
The third: it kills the completion instinct. A list rewards finishing, which quietly encourages you to finish books that are not worth finishing. A connection graph rewards engagement, and a book you abandoned at page sixty having taken one real idea from it can still be a productive node. The scoreboard changes, and the reading changes with it.
No ratings. There is no star rating anywhere in this system and there never will be. Rating a book asks "how good was it," which is a question about the book. The question worth asking is "what did it change," which is a question about me, and it does not compress to a number.
The connection has to be specific. If I cannot say in one sentence why two books are linked, they are not linked. Vague adjacency is what tags are for. The moment "these are sort of related" becomes an acceptable reason, the graph fills with noise and stops meaning anything.
Roughly six links per book. More than that and everything connects to everything, which is the same as nothing connecting to anything. Right now a good number of books on this shelf exceed that, and pruning them back is real outstanding work rather than a finished discipline.
Asymmetry is allowed. If A points at B, B does not have to point back. Sometimes a later book genuinely comments on an earlier one and the earlier one has nothing to say in return. About seventy-five of the pairs here are one-way and that is deliberate. Both pages show the link; only one of them claims it.
This is not a system for reading more. I read roughly the same number of books before and after. It is not a productivity method, and the overhead is real — fifteen or twenty minutes per book, which over a few hundred books is a meaningful number of evenings.
It is also not a recommendation engine. Nothing here is scored, ranked, or inferred. If two books are linked it is because a person sat down and decided they were.
What it is, is a way of making a private habit legible — to me first, and then to anyone else who finds it useful. Most of what I know about how power and money actually work, I know from these books, and I would rather show the wiring than the list.
Two places to start: the standout reads, which is the short version, and the connection map, which is the whole thing at once and is best approached as a picture rather than a document.
Read something I’d love? Send it my way — I’m always adding to the stack.