Spaced repetition systems have gotten way better

Summary

Spaced repetition recap Mastering any subject is built on a foundation of knowledge: knowledge of facts, of heuristics, or of problem-solving tactics. If a subject is part of your full-time job, then you’ll likely master it through repeated exposure to this knowledge. But for something you’re working on part-time—like myself learning Japanese—it’s very difficult to get that level of practice. The same goes for subjects in school: a few hours of class or homework a week is rarely enough to build up an enduring knowledge base, especially in fact-heavy subjects like history or medicine. Even parts of your life that you might not think of as learning-related can be seen through this lens: wouldn’t all those podcasts and Hacker News articles feel more worthwhile, if you retained the information you gathered from them indefinitely? Spaced repetition systems are one of the most-developed answers to this problem. They’re software programs which essentially display flashcards, with the prompt on the front of the card asking you to recall the information on the back of the card. You can read more about them in Andy’s notes, or get a flavor from the images below drawn from my personal collection: What gives these programs their name is how they space out repeatedly prompting you to review the same card, depending on how you self-grade your response. Increasing intervals after correct answers prevents daily reviews from piling up. This is how you can, for example, learn 10 new second-language words a day (3,650 per year!) with only 20 minutes of daily review time. (If you’re still unconvinced and have some time to spare, I suggest Michael Nielsen’s post Augmenting Long-term Memory.) Improving the scheduling algorithm So far, this is all well-known. But what’s less widely known is that a quiet revolution has greatly improved spaced repetition systems over the last couple of years, making them significantly more efficient and less frustrating to use. The magic ingredient is a new...