About Joshua Wong | Why I Build Evidence-Based Learning Tools

The Forgetting Curve

Hermann Ebbinghaus, 1885 · Replicated hundreds of times since

Within 24 hours of learning something new, we forget approximately 70% of it. Within a week, that number rises to 90%. This isn't a failure of effort or intelligence—it's how human memory works. The brain is ruthlessly efficient at discarding information it doesn't think you need.

This is why your child can ace Tuesday's 听写 and blank on the same words two weeks later. The problem isn't that they didn't learn it. The problem is that nothing happened between learning and forgetting to make the memory stick.

Implication: Any learning system that teaches new material without systematically reviewing old material is fighting biology. Most tuition works this way—new content every week, minimal structured review.

Spaced Repetition

Piotr Wozniak, 1985 · Cepeda et al., 2006 · Kang, 2016

The direct antidote to the forgetting curve. Instead of cramming (massed practice), you spread reviews over increasing intervals. You see a word after 1 day, then 3 days, then 7, then 16, then 35, and so on. Each review happens just before the memory would fade, and each successful recall makes the memory stronger and the next interval longer.

The research on this is not ambiguous. A meta-analysis by Cepeda et al. (2006) across 254 studies found that spaced practice produces significantly better long-term retention than massed practice in every case. It's one of the most robust findings in all of cognitive psychology.

Implication: Anki's algorithm (SM-2, developed by Piotr Wozniak) automates spaced repetition. It calculates the optimal moment to review each card based on your child's actual performance. This is why 15 minutes of daily Anki practice can outperform hours of traditional revision.

Active Recall (The Testing Effect)

Roediger & Karpicke, 2006 · Dunlosky et al., 2013

Retrieving information from memory—actively trying to remember something—strengthens that memory far more than passively re-reading or re-studying it. This is counterintuitive. It feels harder. Students rate themselves as learning less. But in study after study, the testing group dramatically outperforms the re-reading group on delayed tests.

Dunlosky et al. (2013) reviewed decades of learning research and rated practice testing as one of only two strategies with "high utility" for learning. (The other was spaced practice.) Meanwhile, the strategies students rely on most—highlighting, re-reading, summarising—were rated as "low utility."

Implication: Flashcards are one of the purest forms of active recall. When your child sees the Chinese character and has to produce the meaning (or vice versa), they're practising retrieval. This is fundamentally different from looking at a word list or re-reading a textbook.

Mastery Learning

Benjamin Bloom, 1968 · Kulik, Kulik & Bangert-Drowns, 1990

Benjamin Bloom's insight was simple but radical: most variation in student achievement isn't caused by talent—it's caused by allowing students to move forward before they've actually mastered the prerequisite material. In conventional classrooms, everyone moves on after the same number of hours, regardless of whether each student has truly learned the material.

Bloom's mastery learning framework requires students to demonstrate mastery of each unit before proceeding to the next. When this approach was tested, the average mastery learning student performed better than 84% of students in conventional classrooms. Bloom called this "the 2 sigma problem"—showing that the right conditions could move average students to exceptional performance.

Implication: In language learning, this means a child shouldn't move on to P3 vocabulary while P2 words are still shaky. Spaced repetition naturally enforces this: words that aren't mastered keep appearing until they are. There's no moving on until the foundation is solid.

Desirable Difficulties

Robert & Elizabeth Bjork, 1994 · Bjork & Bjork, 2011

Robert Bjork's research revealed a paradox: learning conditions that make performance worse in the short term often make memory better in the long term. He called these "desirable difficulties." Spacing out practice (instead of cramming), interleaving different topics (instead of blocking), and testing yourself (instead of re-reading) all feel harder in the moment—but produce dramatically better retention.

The implication is unsettling: the activities that make children (and parents) feel the most productive—re-reading notes, highlighting textbooks, doing 50 of the same problem type in a row—create an illusion of learning without the durability. Real learning feels effortful.

Implication: When your child struggles to recall a character during Anki practice, that struggle is the learning happening. The moment of difficulty—the "tip of the tongue" feeling—is precisely when memory is being strengthened. Effortless review means nothing is being consolidated.

Interleaving

Rohrer & Taylor, 2007 · Kornell & Bjork, 2008

Mixing different types of problems or topics during practice leads to better learning than practising one type at a time (blocking). When students study AAABBBCCC, they perform well during practice but poorly on delayed tests. When they study ABCABCABC, practice feels harder but long-term retention is significantly better.

Implication: Anki's algorithm naturally interleaves vocabulary from different lessons, levels, and contexts. Your child doesn't drill all of Lesson 5 vocabulary, then all of Lesson 6. Instead, words from across the entire syllabus appear in a mixed order based on when each individual word needs review. This mirrors the conditions research shows produce the most durable learning.