README.md

间隔重复的历史

已翻译为中文的章节名即润色过的文本，其余为机翻待润色。

• 1 前言
  • 1.1 真正的历史
  • 1.2 谁发明了间隔重复？
  • 1.3 不灭的神话
  • 1.4 致谢名单
• 2 1985：SuperMemo 的诞生
  • 2.1 为了更好学习的动力
  • 2.2 间隔重复诞生的日子：1985 年 7 月 31 日
    • 2.2.1 直觉
    • 2.2.2 实验
  • 2.3 1985 年 7 月 31 日的事件
  • 2.4 第一个间隔重复算法：算法 SM-0，1985 年 8 月 25 日
  • 2.5 从十年的角度看 Supermemo 的诞生
• 3 1986：SuperMemo 的第一步
  • 3.1 纸上的 SuperMemo
  • 3.2 大学：计算机科学
  • 3.3 ZX Spectrum
  • 3.4 军队生活
  • 3.5 1986 年的夏天
  • 3.6 我的第一台私人电脑：Amstrad PC 1512
  • 3.7 模拟学习过程
    • 3.7.1 学习曲线几乎是线性的
    • 3.7.2 新的项目需要 5% 的时间
    • 3.7.3 学习速度
    • 3.7.4 学习负荷
    • 3.7.5 最优遗忘系数
    • 3.7.6 记忆容量
• 4 1987：DOS 上的 SuperMemo 1.0
  • 4.1 SuperMemo 1.0: 日志 (1987)
    • 4.1.1 为这个想法做准备
    • 4.1.2 编写 SuperMemo 1.0
    • 4.1.3 第一次在 SuperMemo 中重复
  • 4.2 算法 SM-2
• 5 1988：记忆的两个组成成分
  • 5.1 长期记忆的组成成分
  • 5.2 记忆双组分模型的起源
  • 5.3 同行评议刊物 (1995)
  • 5.4 Robert Bjork 的研究
  • 5.5 算法 SM-17 中的双组分模型
  • 5.6 证明
  • 5.7 Murakowski 的证明
  • 5.8 SuperMemo 中的记忆的两个组成成分
• 6 1989: SuperMemo 适应用户记忆
  • 6.1 引入可变间隔函数
  • 6.2 僵硬的 SuperMemo 4
  • 6.3 Supermemo 4 在新的 SuperMemos 中的残迹
  • 6.4 算法 SM-4
  • 6.5 间隔矩阵的问题
  • 6.6 SuperMemo 5
  • 6.7 算法 SM-5
  • 6.8 对算法 SM-5 的批评
    • 6.8.1 Anki 手册中的错误声明
    • 6.8.2 SuperMemo5 的评估 (1989)
    • 6.8.3 新算法优越性的理论证明
  • 6.9 收敛
  • 6.10 矩阵平滑
  • 6.11 间隔的随机散布
• 7 1990: Universal formula for memory
  • 7.1 Optimum review vs. intermittent review
  • 7.2 Model of intermittent learning
  • 7.3 Past (1990) vs. Present (2018)
  • 7.4 Similarity to Algorithm SM-17
  • 7.5 Formulation of the problem of intermittent learning
  • 7.6 Solution to the problem of intermittent learning
  • 7.7 Simulations based on the model of intermittent learning
  • 7.8 Workload vs. Retention trade-off
  • 7.9 Conclusions: model of intermittent learning
• 8 1991: Employing forgetting curves
  • 8.1 Painful birth of SuperMemo World (1991)
  • 8.2 Slow start of commercial SuperMemo
  • 8.3 Origins of Algorithm SM-6
  • 8.4 Algorithm SM-6
• 9 1994: Exponential nature of forgetting
  • 9.1 Forgetting curve: power or exponential
  • 9.2 Wrong thinking helped spaced repetition
  • 9.3 Contradictory models
  • 9.4 Collecting data
  • 9.5 First forgetting curve data
  • 9.6 Forgetting curve approximations
  • 9.7 Exponential forgetting prevails
  • 9.8 Negatively exponential forgetting curve
  • 9.9 Forgetting curve: Retrievability formula
  • 9.10 Retention vs. the forgetting index
  • 9.11 Forgetting curve for poorly formulated material
  • 9.12 Power law emerges in superposition of exponential forgetting curves
• 10 1995: Hypermedia SuperMemo
  • 10.1 Birth of Algorithm SM-8 in a mountain hut
  • 10.2 Absolute item difficulty
  • 10.3 Post-lapse interval
  • 10.4 First grade vs. A-Factor
  • 10.5 Grade vs. Forgetting Index
  • 10.6 Algorithm SM-15
• 11 1997: Employing neural networks
  • 11.1 Neural Networks: Budding interest
  • 11.2 Push for neural networks
  • 11.3 Is SuperMemo inflexible?
  • 11.4 Futility of fine-tuning the spaced repetition algorithm
  • 11.5 Dreger's Neural Network Project
  • 11.6 Neural Network SuperMemo : Why memory model is essential for SuperMemo algorithms
  • 11.7 Neural Network SuperMemo: Design
  • 11.8 Neural Network SuperMemo: Implementation
  • 11.9 David Calinski and FullRecall
  • 11.10 Why is the neural network in FullRecall flawed?
  • 11.11 Future of neural networks in SuperMemo
• 12 1999: Choosing the name: "spaced repetition"
  • 12.1 Quest for a good unique name
  • 12.2 Choosing the name: spaced repetition
  • 12.3 The future of the name: spaced repetition
• 13 2005: Stability increase function
  • 13.1 Why a simple idea could not materialize?
    • 13.1.1 1990: First hint
    • 13.1.2 1993: Distraction
    • 13.1.3 1996: Venture capital
    • 13.1.4 2005: Theoretical approach
    • 13.1.5 2013: Big picture re-awakening
  • 13.2 Increase in memory stability with rehearsal
    • 13.2.1 Two-step computation
    • 13.2.2 Computing stability increase
    • 13.2.3 Symbolic formula for stability increase
      • 13.2.3.1 Dependence of stability increase on S
      • 13.2.3.2 Dependence of stability increase on R
      • 13.2.3.3 Dependence of stability increase on retrievability (2018)
    • 13.2.4 Memory stability increase formula
  • 13.3 Conclusions derived from stability increase formula
    • 13.3.1 Linear increase in value of review over time
    • 13.3.2 Expected increase in memory stability
    • 13.3.3 Memory complexity in spaced repetition
    • 13.3.4 Convergence of sub-stabilities for composite memory traces
    • 13.3.5 Composite stability increase
• 14 2014: Algorithm SM-17
• 15 Exponential adoption of spaced repetition
  • 15.1 Slow start of Algorithm SM-2
  • 15.2 1998: publishing and acceleration
  • 15.3 2008: explosion
  • 15.4 How many people use spaced repetition?
• 16 记忆研究摘要
  • 16.1 关于间隔重复研究的问题
  • 16.2 间隔重复直觉
  • 16.3 早期记忆研究
  • 16.4 20 世纪 60 年代：复兴
  • 16.5 1972 年：莱特纳盒子
  • 16.6 1980 年代：SuperMemo
• 17 The anatomy of failure and success
  • 17.1 Formula for research failure
  • 17.2 Failed experimentation
  • 17.3 Ebbinghaus experiments (1885)
    • 17.3.1 The myth
    • 17.3.2 The fact
    • 17.3.3 Saving on re-learning
    • 17.3.4 Forgetting curve (1984)
  • 17.4 Spitzer experiment (1939)
  • 17.5 Wozniak experiment (1985)
  • 17.6 Why spaced repetition idea succeeded in the end?
  • 17.7 First decade of SuperMemo: Battling skepticism
  • 17.8 The future is bright
• 18 Epilog