Generating my own books

I have not been lucky this year with books. The last few I read were not especially revealing. I read mostly to understand the world better, and lately the books I find are either too dense or too shallow for what I am looking for.

For example, I recently read a short book called 1848, the Revolution of the Intellectuals: I was expecting an introduction to this fascinating period of history where many of today’s nation-states were created. I found a very dense essay instead that offered no introduction whatsoever and was difficult to follow. I also read Why Politics Fail and here the problem was the opposite: I found it devoid of interest and full of platitudes, a rather simplistic pop-science book.

This is a frustration I keep running into. When I want to learn about a niche topic, say, the exact economic structure of the Tokyo metro system, how it evolved and how it is organised today, I struggle to find the right resource. I know I can read Wikipedia or various websites, but this is rarely satisfying. The articles I find often lack the perspective or the angle I would find useful, and very frequently they lack a proper introduction, which forces me to navigate what I call the “forest of distractions”: what starts as reading about the Tokyo metro can end up, in a surprisingly small number of clicks, in a Wikipedia article about the spread of Buddhism in China. After the reading, one is left with the impression of having read about many things but having learnt nothing solid. And I do not particularly enjoy reading long texts on a screen, which further reduces my motivation.

To solve these issues, I have been playing with an idea over the last few days. Presenting the Yongle Dadian Collection.

The Yongle Dadian is the name of a Chinese encyclopedia, most of its content sadly lost, written in the early 15th century, whose size was only surpassed by Wikipedia in 2007.

With AI, I am generating very short ebooks about the topics I want and with the angle I want, that I can read comfortably on my Kindle. These booklets are not perfect, but that does not matter: they are still a better alternative in many cases than reading random posts on the internet or books that are either too simplistic or too complex. In the Yongle Dadian Collection I will curate a collection of ebooks on the most diverse topics, really anything that catches my attention, publicly available and downloadable for anyone who shares these interests. The ebooks will vary in length but will be mostly self-contained and adhere to a few rules I consider valuable:

• Zero bullshit
• Mechanistic explanations
• Enough context
• High signal, low noise

Think of the Yongle Dadian Collection as something similar to the A Very Short Introduction series, but even shorter and more to the point. Think of them as very specific Wikipedia articles formatted in a pleasant reading format. Topics will be highly specific and niche, and certainly biased towards what I find enjoyable. I make no pretence of offering a neutral sample of topics. The ebooks are produced with a combination of prompts and AIs: they are based on deep research outputs from Gemini and refined and rewritten with the help of Claude.

I will maintain this collection until I get bored, but it would be satisfying to see it grow over the years into a decent library.

You can find the collection here.