The Ten-Year Programmer, Reimagined
Peter Norvig’s essay, “Teach Yourself Programming in Ten Years,” is a foundational text. It’s a reality check for anyone hoping to shortcut their way to expertise. His central thesis is simple: mastery demands time, deliberate practice, and a deep, holistic understanding of the craft.
Twenty years later, his advice is more relevant than ever. The only thing that’s changed is everything.
We now live with an ever-present pair programmer in the form of agentic AI. This has led to a quiet panic in some corners: Have the ten years been compressed into ten weeks? Is mastery now obsolete?
The answer is an emphatic no. The ten thousand hours are still required. The journey is still long. But the nature of those hours has fundamentally changed. The goal is no longer to be the lone craftsman who builds everything from scratch. The new goal is to become the conductor of an orchestra of intelligent agents.
Norvig’s core argument was about rejecting shortcuts. The AI is the ultimate shortcut, but it’s a shortcut to typing, not to understanding. The ten years are still the time required to build the mental model, the gut instinct, the judgment. That part of the journey is non-negotiable.
So, what does the augmented path look like? Let’s reimagine Norvig’s advice for our new reality.
- Program. (But differently.) The old way was to write every line yourself. The new way is to be in a constant dialogue with a creation engine. Your “programming” is now a cycle of high-level instruction, critical analysis, and precise refinement. You spend less time on syntax and more time on strategy. The deliberate practice is no longer just in writing code, but in critiquing it. Is this generated function secure? Is it efficient? Does it handle edge cases my human brain would have caught? Your time is spent elevating the AI’s output from “correct” to “excellent.”
- Talk with other programmers (and share your prompts). This is more critical than ever. The most valuable conversations are no longer just about algorithms and data structures. They’re about prompt engineering, about how you framed a problem to an AI, about the unexpected outputs you received and how you debugged them. The new tribal knowledge is a shared library of effective human-AI interactions. Your ability to articulate a problem clearly to a machine is a skill you learn by seeing how others do it.
- Work on projects with other programmers… and AIs. The old advice to be the best programmer on your team still holds, but I’d add a corollary: learn to be the best conductor. Use AI as a tireless junior partner to prototype ideas, explore unfamiliar domains, and accelerate the “boring” parts of a project. The real learning happens when you have to take the AI’s naive implementation and shape it into something robust and maintainable. You are still the one with the vision; the AI is just the one with the fast hands.
- Learn at least a half dozen programming paradigms. This is no longer a niche recommendation; it’s a survival skill. To effectively guide an AI that can write in any language or style, you must have a diverse mental model of what “good” looks like. You need to be able to ask, “Rewrite this in a functional style,” or “How would this look as an actor-based system,” and understand the trade-offs of each. The AI is the polyglot, but you must be the linguist.
- Remember there is a “computer” in “computer science.” This is perhaps the most timeless piece of advice, and AI makes it more important, not less. When the AI can generate a sorting algorithm in a second, the value isn’t in knowing the algorithm. The value is in knowing why you’d pick a quicksort over a mergesort for your specific data set and hardware constraints. The fundamentals of performance, memory, and networking are the very things that allow you to ask the AI the right questions and validate its answers.
The ten thousand hours are still the price of admission. But what you’re buying with that time has changed. You’re no longer just paying to become a great programmer. You’re paying to become a great thinker, a great communicator, and a great conductor. The AI handles the notes; you have to spend the decade learning how to make the music.