In 2001, in an essay called "The Other Road Ahead," Paul Graham described a future where nobody owns a computer, only data; where the client is a telephone or something that used to be one; where your files live on a server maintained by people who are good at that sort of thing, and a disk crash happens to someone else's hardware instead of your life. He was describing what would later be called the cloud, five years before the term existed, and he got the architecture almost exactly right. That's the strange thing about rereading this book now: the prediction isn't a curiosity that missed. It landed. You are, in some sense, living inside "The Other Road Ahead" as you read a review of it.

What the essay doesn't mention — not omits accidentally, just never raises as a question worth asking — is who pays for the servers, and how. Graham's whole pitch is convenience: less to install, less to lose, less to fix. He's right that users would trade ownership for ease. He doesn't ask what the company on the other end of that trade gets in return for being "the kind of people who are good at that kind of thing." The answer turned out to be your attention, sold twice — once to you as a product, once to advertisers as a resource — and Graham, a man who spent an entire chapter cataloguing the ways moral consensus gets manufactured by groups "poised halfway between weakness and power," never turns that instrument on the industry he was describing. He built the plumbing. He didn't ask who owns the water once it's flowing.

The wealth chapters are where the book's confidence is least defended. "How to Make Wealth" runs the arithmetic: a good hacker at a big company is worth maybe $80,000 a year; strip out the middle manager, the interruptions, the averaging-down that comes from working alongside people who are merely competent, and that same hacker working for himself is worth something closer to three million. Graham calls this "the technical sense" of wealth — not money, but the stuff people will actually pay you for — and insists the arithmetic is basically fair, a market correctly pricing effort. He wrote this before he became the man who funded the arithmetic at scale. Y Combinator isn't a sequel to this book; it's the same claim, capitalized. There was no gap between the diagnosis and the practice for the market to absorb, no warning that got neutralized by being turned into a business plan — the essay already was the business plan. That makes it a different kind of document than the ones that get co-opted after the fact. This one shipped with its own execution built in.