Can Great Programmers Be Taught? Insights from a Software Design Workshop

For over 80 years people have programmed, yet there is no consensus on what constitutes good software design. The author questions the most fundamental concept in computer science and argues that problem decomposition—breaking complex problems into independent, solvable pieces—is the core principle that is rarely taught.

Drawing on a conversation with a Google VP, the author notes that an elite programmer is worth replacing 300 average ones, highlighting the huge productivity gap and prompting the question of whether such talent can be taught.

Referencing Geoff Colvin’s *Talent Is Overrated*, the author suggests that deliberate, focused practice, rather than innate talent, is key to mastery, and proposes teaching software design through extensive coding experience.

The author created Stanford’s CS190 software‑design workshop, modeled after high‑school essay writing: students write large projects, receive detailed peer and instructor feedback, and iterate over three three‑week units, reviewing tens of thousands of lines of code each quarter.

The workshop emphasizes two design principles. First, classes should be "deep": they provide many functionalities behind a simple interface, reducing learning cost while increasing leverage. Second, classes should be generic rather than overly specialized, avoiding the proliferation of tiny, purpose‑specific methods that lead to "classitis" and tangled code.

The author contrasts tactical programming—making code work quickly at the expense of long‑term complexity—with strategic programming that invests time up front to produce clean, maintainable designs. Tactical shortcuts accumulate technical debt, while strategic approaches keep complexity manageable and improve future development speed.

The author observes that companies that prioritize rapid, tactical delivery (e.g., early Facebook) often end up with highly complex, hard‑to‑maintain codebases, whereas firms that foster strong design cultures (e.g., Google, VMware) sustain faster, higher‑quality development.

Students in CS190 gradually adopt the deep‑class and generic‑design mindset, leading to cleaner implementations and richer discussions about software design throughout the semester.

In conclusion, while the author believes great programmers can be taught, a complete curriculum does not yet exist. The emerging principles—problem decomposition, deep and generic class design, and strategic programming—offer a philosophical framework for improving software design practice.