Back to A Better Way to Think About Tech.
Why Did Lawrence Wright Keep Research in a Filing Cabinet?
Cal Newport points to Lawrence Wright's filing cabinet to show why slow research can beat an AI agent. The Looming Tower won major awards and was an important account of the cultural, political, and religious conditions that led to the 9-11 attacks. Wright's system for it was handwritten cards in folders in a filing cabinet, not a fast database. The current instinct, especially among AI people on X, is to build an agent that files research into a cloud database and answers RAG-style queries, then spend days and days making that automation work. Newport says none of that is the core of writing a book. The core is the long, hard effort of touching ideas, moving them, and pulling the right card when a section needs it. Slowness gives the brain time to shift and to make something original.
The publisher did not pay Lawrence Wright for how quickly the research was organized. The publisher paid for how good a book he could write. Efficiency for its own sake, and automation just because it is possible, are fine as a hobby and miss the work that matters. Techno-selectionism keeps what is valuable: producing something worth reading, and using your brain in a way that is sustainable, enjoyable, and meaningful to the people consuming it. A filing cabinet that slows you down enough to think can be enough.
A Better Way to Think About Tech ·
From the transcript
Um, health, uh, We got here Syria, Jihad, and Wright would actually just write quotations or notes or ideas as he came across them and filed them in this filing cabinet. Okay, why am I pointing this out? Looming Tower was a magisterial book. It won major, major awards. It was a really important effort in understanding the cultural, political, and religious conditions that led to the 9-11 attacks.
And this is a pretty old-fashioned organizational system. It's handwritten cards in folders in a filing cabinet. Here's why I bring that up because it comes back to our thoughts about we're just talking about the good thing about hard things. This isn't a super efficient way to do things. And our current sensibility would be no, no, no, especially if you're again on X with the AI people.