I’m awed by Scott Young’s MIT Challenge for its simplicity and audacity. In 2011, he gave himself one year to complete the full four-year curriculum in Computer Science from MIT. MIT made most of its course material available for free online, including the tests and the answer keys, so Young could work at his own pace, at his own home, and not spend any tuition money on this experiment. He successfully finished this project in 2012. In 2019, he published the book Ultralearning, a book that helps people plan big learning projects of their own.
Young writes that for any ultralearning project, you should budget about 10% of your time on metalearning — making a plan to identify what you need to learn and how you will go about learning it. Furthermore, he advises you to break down the what you need to learn into three buckets:
- New concepts that you need to understand
- New facts that you need to memorize
- New skills that you need to practice and acquire
You do this because the techniques to efficiently learn things are different for the different categories. I’m shocked I’d never thought about this before! After reading this, I understand my own learning shortcomings much better than I did before. I love love love learning that falls in “understanding new concepts” category. I gravitated to subjects like math, physics, and computer science that are rich in first-principles conceptual understanding. However, the learning tools that help me pick up new concepts don’t help me pick up new skills or memorize things, so I struggled in subjects like foreign languages and art. I wish I’d had this book back in my high school years to know that I needed to use different tools to learn different things.
Ultralearning also makes another important argument. As much as possible, you should structure your learning project around doing the thing you’re trying to learn how to do and you need a way to get feedback on how you are doing. Do you want to learn a language so you can speak to locals when you travel? Then you should be speaking to locals as much as you can as early as you can. (The reaction of the native speakers gives you real-time feedback!) If you want to learn jazz guitar, you need to spend a lot of time playing guitar.
The book devotes a few pages arguing against the effectiveness of my employer, Duolingo, because it is a very indirect way to learn a language. At the same time, though, Young writes of the importance of using drills to isolate and improve specific skills for your learning project. Someone trying to improve at tennis will do more than play games; forehand / backhand / serving drills will help you isolate and improve the building block skills for the game faster than you can just from games. Duolingo plays a similar role for serious language learners. It’s not a substitute for talking to native speakers, but the app does help you drill on vocabulary and grammar. (Also! Duolingo provides way more than app-based translation exercises. You can use https://events.duolingo.com to find groups to practice speaking and listening. You can use Duolingo Podcasts for practice understanding native speakers. And perhaps most importantly, Duolingo offers learners of all levels motivation to keep learning — the hardest part of learning a new language.)
Anyone who is interested in learning, and in particular anyone who is interested in self-directed learning, should read Ultralearning. You’ll find a ton of helpful material. For those interested in educational technology, Ultralearning suggests two things where technology seems uniquely positioned to help people learn faster and better: In providing material to practice and in feedback. All of the influential educational software I can think of — from Duolingo to Anki to Kahn Academy to experimental efforts like the “mnemonic medium” — deliver in both of these dimensions. However, I think because Ultralearning already assumes its reader is highly motivated to learn, it doesn’t say much about one of the most interesting contributions of educational technology. Successful technology makes learning fun and contains mechanisms to help sustain motivation over time.