The Essential Question

"What should I read next?" Humans only need apply.

This seems intriguing, judging from the cover.

Few questions matter more than “What should I read next?” Yet we rarely think about our choice in any systematic way.

Most of our experience involves looking - at our parents, first loves, and fellow employees, of course, but just as much at the page and the screen, those engines of knowledge, introspection, and catharsis. Born into the world alone and leaving it that way, we spend the greater part of our time looking and reacting, hopefully to learn. It’s true even if you include the third we spend unconscious in bed.¹

A couple of weeks ago Benjamin Breen, author of the excellent Res Obscura Substack about History, Art, and Technology, took “What should I read next?” in a new direction, building a nonfiction book search engine based on the finalists of literary prizes over the past century. That may not at first sound promising, since we already have The National Book Award and The Pulitzers to guide us, but he draws from 72 different awards (Plutarch Award in Biography, anyone? Laura Shannon Prize in Literary Studies?)

My search topic was “libertarian gardening.” And sure enough, it gave me some interesting stuff I hadn’t heard about.

That’s a lot of judges, reading through an enormous number of texts each year and telling you what they liked. A good filter.

Like Breen, part of my rich (and sometimes paralyzing) interest in different subjects comes from days I spent shelving books in a library. Librarians are tolerant book lovers, and, I suspect, secretly like it when a young employee moves slowly through the stacks, perusing the contents of one or another volume before returning it to the shelf.

Or if you prefer, “AI and the Humanities.” The O’Gieblyn book is quite fun.

I benefited from two filters there. One was the Dewey Decimal System, which meant I could easily check out related volumes on either side of a book on an interesting subject. The other was the personal recommendations of strangers, since someone must have been interested enough in the book to take it out. I once spent five days in the annex, restoring DDS order to a pile of books as tall as me, the consequence of a cascading shelf collapse. That was unfiltered, and harder. When I spent three years in a warehouse in Singapore with 12,000 volumes (salesman, educational publisher) I had the benefit of categories, but no human filter.

That human factor is what appealed to me about Breen’s engine. Until recently, much of the way we consciously read the world in audio, books, even video, came to us through other souls. Today it may be through whatever is pushed at us in social media, those algorithmically sliced versions of stuff people are posting, with the scale and quality generally worse. Everyone knows something better has to come along, with a sustainable business model that doesn’t degrade the content.

Some of the most useful early search engines really aggregated people. Yahoo! circa 2000 had buildings full of specialists figuring out one or another of the ever-budding domains of knowledge on the Internet. The system collapsed before fully reaching its logical consequence, a Borgesian parody of infinite categorization of an even larger infinity of knowledge, to the point of overwhelming the Earth. Early Google search was based on mathematically determining how many human-crafted links a page received, and gave authority to a page because a lot of people gave it their time. The drive for quarterly earnings, along with the many interests out to “optimize” their (usually squalid) bit of content, have corroded that process. Certainly, it’s hard to be surprised by the results in a pleasant way.

When we have an issue with the quality and authority of what we’re looking at now, it’s usually because too many robots, often masquerading as people, are involved. To put it another way, a small group of human programmers have artificially distorted their authority, injecting junk into our choice. People remain the great filter, whether we ask a trusted friend what they’ve enjoyed reading, or a stranger of kindred spirit brings you their passionate obsessions.

Companies like Anthropic don’t just name a product “Claude” and treat it like a budding human intelligence to avoid culpability when something goes wrong (“Claude broke out of its sandbox!” is better PR than “we don’t know how to maintain safety standards on our own software!”) They want us to trust it, the way we’re quick to trust most people who give us a recommendation.

One more thing: People who recommend great books you have never heard of are, at their best, friends of the mind. Sometimes they are long dead, but that does not detract from the connection we feel with them. Great software? Useful tool, but there is no fellow feeling.

1. This recently occurred to me when I spent a few days in an isolated cabin, and found to my delight that I was reading less daily news, and scooping from our electromagnetic surroundings less video of our president, a creature who feeds on our continual attention, and dedicates much of his time and power to commanding it. Remember how, just a few years ago, you could go days without thinking about the president?. The technologic age is one of urgency over depth.