I finally found time to compile my 2021 reading list. This year my reading shifted increasingly toward psychology, economics, and traditional statistics — partly useful for my doctoral research, and partly genuinely illuminating for everyday life.

Humanities

Working in Public is a book written by a GitHub employee about the production and maintenance of open source software. I first heard of it from Vue’s author Evan You. Having been drawn to open source software and actively involved in the open source community for years, I immediately bought and read it. The most thought-provoking points for me included the author’s framework for thinking about open source project models based on user growth rate vs. contributor growth rate (federated, club, hotel, and toy models), and the symbiotic relationships among distribution platforms, maintainers, contributors, and participants. Reading this book alongside my own growing open source experience, I increasingly set aside the naive idealism about “open source spirit” from an earlier generation. In a sense, this book answered my long-standing question about what “participating in open source” and “doing open source” truly mean — they are not the same thing. Anyone who genuinely wants to engage with open source should ask themselves: can I continue maintaining a project after losing the initial passion that drove me to create it? If so, what material conditions are necessary?

• Working in Public: The Making and Maintenance of Open Source Software

Psychology

This year my research gradually moved toward human rational decision-making, which inevitably led me to read a fair amount of psychology. The books centered around two key figures — Herbert Simon and Daniel Kahneman — and progressively unpacked foundational theories in modern cognitive psychology: bounded rationality, heuristics, and decision errors. When I read and experimented with these seemingly simple concepts, I was genuinely struck by their elegance. Without resorting to any so-called absolute rational analysis, they reveal and demonstrate through social experiments these pervasive human flaws step by step.

• Thinking, Fast and Slow
• Models of a Man: Essays in Memory of Herbert A. Simon
• The Adaptive Decision Maker
• Expertise, Communication, and Organizing

Statistics

Although I considered myself well-grounded in statistics from undergraduate study, I had long been limited by a rather superficial understanding of significance testing, maximum likelihood estimation, and the frequentist vs. Bayesian debate. But when my own research demanded precision on every detail, I was humbled by the many disciplinary nuances and their ingenuity. This year I feel I systematically encountered Bayesian analysis and causal inference methods for the first time. Most of what I learned came from the books below, which further sharpened my critical and methodological understanding of social science research.

• The Book of Why: The New Science of Cause and Effect
• Causal Inference in Statistics: A Primer
• Elements of Causal Inference: Foundations and Learning Algorithms
• Human-In-The-Loop Machine Learning: Active Learning and Annotation for Human-Centered AI
• The Art of Statistics: Learning from Data
• A History of Vector Analysis

Engineering

From where I stand today, engineering books are perhaps the least interesting category of all. This year I bought four engineering books of varying depth. The first is a classic for understanding multi-threaded programming — I had skimmed parts of it in undergrad but without deep enough knowledge of concurrent programming to absorb much. This time I read the English original to get a more thorough and systematic grasp of the subject, and it did fill in gaps in my understanding of memory barriers and transactions in shared memory. The other books were not fully read, especially the last two. I particularly liked the chapter on testing in the third book; the fourth was a reference I bought while building a rendering engine — more a source of guiding principles than hands-on recipes, but equally inspiring. The last one reads almost like a man page, bought specifically when I needed to understand system calls for memory allocation.

• The Art of Multiprocessor Programming
• Building Secure and Reliable Systems
• Game Engine Architecture
• The Linux Programming Interface: A Linux and UNIX System Programming Handbook

Review and Further Reading

Looking back at my 2021 reading, I increasingly notice that the volume of books seems to decrease year by year as I get older. On one hand, daily life demands more time, squeezing out reading time. On the other, I find myself wanting to experience things rather than merely gain knowledge from books. Undeniably, reading distills knowledge into structured systems — we can rapidly acquire large quantities of processed information. But only knowledge that has been truly lived can be deeply understood; that lived process is also more interesting and leaves a deeper imprint.

If you’re interested in previous years' reading lists: Reading List Archive.