Feels Like Progress

I played that game in a middle school in Yantai, a port city in northern China, and in a high school in Shanghai. I played it at a liberal arts college in America and in the economics PhD program at Cornell. Then I played it at Amazon, at Facebook, and at Tencent. By 2022, I was a vice director at Tencent, leading a data and AI team of about thirty people, collecting top performance ratings, and being prepared for a general manager role.

I don’t usually lead with my résumé. If an argument needs the author’s credentials to hold it up, the argument isn’t finished. But in this book, the résumé is part of the evidence. It’s a record of how good I was at the game, and so a measure of how long the game could hide from me what I was actually doing.

This book is about what I found when I finally took a good look at the prize.

Three moments

The thirty-six pages were the first moment. They felt like progress. I had worked hard, the methods were sound, and the document looked like serious work. It simply didn’t help anyone decide anything.

The second was a conversation. In 2024, I sat down for my channel with Mike Develin, a data scientist who had led data teams at Instagram and Shopify. Mike had watched many smart technical people stall in their careers, and he traced it to beliefs they had carried out of school: that every real problem has a standard answer, that the textbook is enough, that harder things are more valuable. I recognized almost every one of them in myself. Much of what I believed about what was right at work, and what was valuable, had been formed in school, and it didn’t match how the world works. Paul Graham’s definition of wealth is the simplest I know: what people want. Not what a grader rewards. Only once I could see what was real could I see what was fake.

The third was a sign by the road. In the fall of 2022, outside Seattle, I passed a McDonald’s with a sign out front: NOW HIRING $17.75. I asked myself, half as a joke, whether I would rather have that job or my boss’s. Then I thought about it seriously, and the answer unsettled me. The promotion I was being prepared for was real. I no longer wanted the life that came with it.

Three different problems: work that changed nothing, a sense of value that school had trained into me, and a ladder leading somewhere I didn’t want to go. What they share is how ordinary they are. Each can go on, in many lives, for years. And each is kept going, in part, by a sense that something is moving: the report gets longer, the grades keep coming, the title gets bigger.

I’ve come to believe that most of us spend a large part of our working lives inside that sense of motion. We do work that produces the evidence of work, and we mistake it for results. We learn in ways that produce the evidence of learning, and we mistake it for understanding. I’ll call these fake work and fake learning, and they are connected. Long before our first job, school trained our sense of what counts as good work, and we carry it into the office. That is a large part of why fake work is so hard to see from the inside: it looks exactly like what we were taught good work looks like. Both fool the same thing, our feeling of progress, and that is why they last so long. This book is about what happens to them now, and about what will last when AI can blow it all up.

Why now

For a long time, the evidence of work was worth something, because it was expensive to make. It at least proved that someone had spent the time and the effort. Managers learned to read it. Schools learned to grade it.

Artificial intelligence breaks that arrangement. AI can now produce something that looks like those thirty-six pages in a minute or two. It can write the code, the report, the strategy deck, the polished answer to the exam question. It makes the feeling of progress nearly free. When work-shaped output costs almost nothing, the shape of the work stops proving much. A polished document no longer shows that someone thought. A working prototype no longer shows that someone learned.

The subtitle of this book has a double meaning, and I intend both. To blow something up can mean to inflate it. It can also mean to destroy it. My prediction is that AI will do both to fake work. First it will inflate it: more reports, more decks, more prototypes, more evidence of effort than anyone can read. Then, over a decade or two, and unevenly, because institutions change slowly, it will destroy much of what that evidence was worth.

What keeps its value is harder to produce, and it is where this book is going. Real work is work that could fail, at something that matters to someone who depends on it. It starts with a question worth asking, because the question decides more than the effort. It is made with craft: the feel that tells an engineer, a product manager, or a writer that something is wrong before they can say why. That feel can’t be written down, so no one can download it, and no machine can acquire it on your behalf. It grows slowly, through work that answers back, and it belongs to the person who grew it. And at its best, real work compounds. What it leaves behind keeps paying off long after the day it was done.

Learning and doing call for different standards. When you need to get something done, use every honest help you can find, including AI. When you’re trying to learn, keep the part that matters in your own hands, because the feel grows only in the person who does the work.

What this book is

This is a book of ideas. Most of the evidence for them comes from what happened to me, because that is the evidence I can vouch for. The stories are there to show each idea at work: what it looked like from the inside, before I had a name for it.

Part One is about fake work: what it is, and why it survives so well, from a single report to an ordinary working day. Part Two is about AI and work: how AI will first inflate fake work and then empty it of value, and what that means for a career. Part Three goes back to school, where I learned the habits that make fake work feel natural, looks at what AI is now doing to learning, and follows those habits into work. Part Four is about what will last: work that can fail, the question worth asking, the craft nobody can hand you, and work that compounds.

I’m not writing as someone who has escaped the game. I did a great deal of fake work, some of it very well, and I was rewarded for some of it. I still catch myself doing it. What I can offer is an account from inside: from someone who got good at the game, kept collecting its rewards, and only slowly noticed that the scoreboard and the purpose had come apart.

Each chapter ends with a short takeaway: the idea I would most want you to keep from it.

The way this book was made puts some of its ideas into practice. If you’re curious, I describe it at the end, in the Afterword. I’d suggest reading that last. By then you’ll have your own judgment of the book, and what I say there will mean more.

The first chapter goes back to that document, and to the question I couldn’t answer.