How to Learn to Code I: Use SRS and Anki

Summary

The narrator argues that spaced repetition systems (SRS), especially Anki, can help people learn programming more quickly by moving important knowledge into long-term memory. Drawing on nearly a decade of using Anki, he describes how it helped him progress from learning basic web development to doing paid programming work.

How spaced repetition works

• People tend to forget newly learned information over time.
• An SRS schedules reviews at expanding intervals, prompting recall before knowledge fades.
• In Anki, learners try to answer a prompt, check the answer, and rate how well they remembered it. The system uses that feedback to schedule the next review.
• The approach is intended for knowledge worth retaining long-term, not as a replacement for programming practice or understanding.

Benefits the narrator observed

• Chunking: Familiarity helps programmers recognize patterns and combine individual facts into larger concepts, much as chess players recognize board structures. This frees attention for higher-level problem solving.
• Fluency: Regular recall can make concepts and coding choices feel more immediate, reducing the need to consciously reconstruct everything from scratch.
• Consistency: Reviews accumulate when skipped, creating motivation to study regularly—especially useful when learning to code.

Six rules for using Anki effectively

1. Continually add cards about difficult or memorable problems. Adopt a “collector’s” mindset: when something confuses you or causes a close call, turn the lesson into a card. Examples include distinguishing similar Git expressions or remembering a security risk involving login tokens sent over non-HTTPS links.

2. Choose knowledge with long-term value. Adding and reviewing cards takes time, so prioritize fundamentals, ideas shared across languages, and stable technologies. The narrator suggests using the “Lindy effect”—the idea that something’s longevity may indicate it will remain useful—as one guide.

Prefer broadly useful data-structure concepts, standards and protocols such as SQL, HTTP, and JSON, computer science fundamentals, common Unix tools, and widely used language features. Be cautious about framework-specific APIs, proprietary tools, exact syntax or function signatures, niche languages, and new or little-used features that may soon become obsolete.

1. Build cards from real problems and practical use. Learn by making things rather than collecting facts “just in case.” When actual work exposes a confusing concept or part of the technology stack, consider making a card about it.

2. Run code before adding it to your deck. Executing examples helps catch incorrect or outdated snippets and confirms that you understand what the code does. Memorizing a code fragment without understanding it can create false confidence.

3. Record sources and tag cards. Note where a fact or example came from—such as a book, blog post, or Stack Overflow answer—so you can revisit the explanation later. Tags help organize cards, run focused reviews, and remove material that no longer seems useful.

4. Keep refining the cards and your understanding. Split cards that require remembering too many separate facts, and revise or delete cards that are unclear, outdated, incomplete, or no longer needed. As your understanding improves, update cards to be more accurate and connect ideas together.

Closing message

Anki can accelerate learning, but the narrator stresses that it is not a painless shortcut: it requires effort, discipline, and repetitive review. He encourages viewers to try it for 30 days and judge its usefulness for themselves.

Speakers and sources mentioned

• Speaker: The video’s narrator; no name is given in the subtitles.
• People cited: Derek Sivers, Gwern, Michael Nielsen, Piotr Wozniak, Damien Elmes, and Pavel, a Stack Overflow contributor.
• Other sources, tools, and services mentioned: Wired magazine’s interview with Piotr Wozniak; Anki; Duolingo; Execute Program; the article 20 Rules for Formulating Knowledge; and Stack Overflow.