At the end of 2025, I put together the list of books I had read. The order is random, not a ranking.
Nonfiction
- Clean Architecture
- Create: Doing Valuable Work in Unconventional Ways
- Understanding Linux Processes and Memory
- The Effective Executive
- System Design Interview, Vol. 2
- How Bezos Runs Meetings
- How to Read a Book
- Mini Habits
- Nonviolent Communication
- From Zero: Principles and Practice of Large Language Models
- StatQuest Illustrated Guide to Machine Learning
- The Fifth Discipline
- Jensen Huang: Nvidia's Core
- The Trend
- The Wealth Equation
Fiction
- Day and Night
- The Murder Case Code
- Night Walker Records
- Genius on the Left, Madman on the Right
- Bleach
- Flaw
- Corpse Case Investigation Unit
The one that mattered most
The best book of the year was, without question, the one about how Jeff Bezos runs meetings. It stood out because I actually put it into practice.
The biggest influence was the way it made me think about which meetings should exist at all, and which should not. The other useful idea was the one-page meeting document: condense the meeting content onto a single page and have people read it before the meeting starts. In practice, this improved meeting efficiency a great deal. Because I could apply the ideas and feel the change, the book became genuinely meaningful to me.
New thoughts
This year brought so many AI tools. I spent a lot of time trying out the changes they made possible, and I watched more videos and other material than I read books. That led to some new thinking.
People ask me whether reading is still useful now. My answer is yes. Reading is just one way of taking things in. With so many tools available, we have more and better input methods. But I still believe in the subtle, slow influence of reading. A persuasive short video or a beautifully made AI explainer may disappear from your memory quickly. Many ideas I encountered in books are still vivid to me.
Of course, what I read next year may change a lot. I have realized that many books on fundamentals or underlying principles need to go. I have already thrown out several Rust syntax books. Put simply, the era of rote 'standard answers' is coming to an end. The Spring source-code material I read before can go too, because I really will not use it. What remains is the design thinking behind it.
That will also change my blog. I will probably write fewer tutorials and source-code walkthroughs, and focus more on design, approach, and solutions to common problems. In this era, design feels like a promising path: how to design for a particular scenario, and how to quickly judge whether one of several AI-generated design proposals is feasible for the situation at hand.
I hope the new year brings a new rhythm and a clearer way of thinking, and that I can catch the wind this remarkable era is offering.