Fact-checked by Grok 2 weeks ago

SuperMemo

SuperMemo is a computer-assisted learning software that employs spaced repetition algorithms to optimize long-term retention of knowledge by scheduling reviews based on individual forgetting curves.^[1] Developed by Polish researcher Piotr Woźniak in the mid-1980s, it originated as a personal tool for memorizing biochemistry and English while Woźniak was a molecular biology student, evolving from paper-based flashcards to the first digital implementation in 1987.^[2] The software's core innovation, the SM-2 algorithm introduced in SuperMemo 2.0, uses user ratings of recall difficulty to dynamically adjust review intervals, a method that has influenced modern learning tools like Anki.^[2] Over nearly four decades, SuperMemo has progressed through 19 major versions, incorporating advanced features such as incremental reading for processing large volumes of text, support for multimedia elements including images, audio, video, and HTML, and integration with AI-driven tools for language learning and content generation.^[3] Key milestones include the addition of variable intervals in 1989, hypermedia and knowledge trees in the 1990s, and priority queues for massive collections in the 2000s, culminating in version 19.1 released in April 2025, which enhances incremental learning with automated import of educational videos.^[2]^[4] Founded as SuperMemo World in 1991 by Woźniak and Krzysztof Biedalak, the company has commercialized the software, offering pre-built courses for languages, professional skills, and general knowledge, alongside customizable collections for users worldwide.^[5] The system's emphasis on sleep's role in consolidation—drawing from Woźniak's research on memory—distinguishes it from generic flashcard apps, promoting "sleep & make" cycles where learning sessions are followed by rest to maximize retention.^[1] SuperMemo is available across platforms, including a Windows desktop application, mobile apps for iOS and Android with hands-free modes, and a web-based version, making it accessible for both individual learners and educational institutions.^[6] Its algorithms continue to evolve, with recent updates focusing on neural network-based predictions for even more precise spacing, underscoring its ongoing relevance in cognitive science and lifelong learning.^[7]

History and Development

Origins and Early Versions

SuperMemo originated from the personal learning experiments of Piotr Woźniak, a Polish student at Adam Mickiewicz University in Poznań, who began developing the method in 1985 to enhance his English language studies while pursuing a master's degree in molecular biology.^[8] Motivated by inefficiencies in traditional memorization techniques, Woźniak sought to optimize retention through systematic review scheduling, drawing on the broader concept of spaced repetition as a psychological principle for combating forgetting.^[9] Woźniak's early efforts traced back to manual experiments with spaced repetition starting in 1979–1980, during his high school years, when he used paper cards and notebooks to track vocabulary and concepts, frustrated by rapid forgetting despite repeated study.^[9] By 1982, he refined this approach with active recall using English-Polish word pairs, compiling detailed notebooks that grew to 79 pages containing 2,794 words by December 1984; these manual methods highlighted the need for interval-based reviews but were limited by the labor-intensive process of tracking progress on paper.^[10] This groundwork culminated in 1985 with a focused experiment from July 31 to August 24, where Woźniak tested optimal review intervals on sets of approximately 40 word pairs per page across five pages, establishing the foundational scheduling principles for SuperMemo as a tool tailored to his university demands.^[9] The transition to digital implementation occurred in 1987, when Woźniak created the first computer version of SuperMemo, known as SuperMemo 1.0, for IBM PC compatibles running DOS, developed on an Amstrad PC 1512.^[11] Released in December after 16 evenings of programming starting in November, this initial software featured basic repetition scheduling to automate interval calculations, enabling Woźniak to memorize 10,000 English items over 365 days with just 40 minutes of daily study.^[2] Unlike later iterations, it lacked advanced modeling of forgetting curves, relying instead on straightforward postponement of reviews based on user performance.^[9] Early development faced significant challenges due to constrained computing resources, including only 360 KB of floppy disk storage, which forced simplifications like abandoning full repetition history logs until 1996.^[11] Woźniak's work remained a solitary pursuit amid the demands of his studies and Poland's limited technological infrastructure in the 1980s, prioritizing personal efficiency over commercial viability and conducting all validations through his own learning data.^[9] These constraints underscored SuperMemo's evolution from a rudimentary personal aid to a more robust system, driven by the goal of minimizing study time while maximizing long-term recall.^[12]

Evolution of Algorithms and Software

The evolution of SuperMemo's algorithms and software commenced with the introduction of the SM-2 algorithm in SuperMemo 1.0 in 1987, representing the first computerized implementation of an automated spaced repetition system. Developed by Piotr Woźniak, this milestone built upon his earlier manual methods from the mid-1980s, enabling dynamic adjustment of review intervals based on item difficulty and recall performance.^[13] The software, initially coded in Turbo Pascal for DOS platforms, marked a shift from paper-based scheduling to algorithmic optimization, laying the foundation for subsequent refinements.^[2] Key advancements accelerated with the establishment of SuperMemo World in 1991 by Piotr Woźniak and Krzysztof Biedalak, which commercialized the product and supported broader development.^[14] In 1993, SuperMemo 7 introduced Windows compatibility and a graphical user interface, incorporating support for images and sounds to enhance multimedia learning integration.^[2] This version sold over 30,000 copies by the mid-1990s, reflecting growing adoption. SuperMemo 8 followed in 1997, alongside hypermedia and a knowledge tree structure for organized content management.^[2]^[15] Subsequent releases emphasized interoperability and advanced scheduling. SuperMemo 2004 (version 12) incorporated XML for data exchange and web integration, enabling synchronization with online resources and mobile applications.^[15] A shift toward open-source elements occurred in earlier iterations, with SuperMemo 5 entering the public domain in 1993 and SuperMemo 6 in 1995, fostering community contributions while later versions retained proprietary core algorithms.^[2] SuperMemo 2006 (version 13) advanced scheduling with a priority queue to handle large-scale incremental reading workloads efficiently.^[15] Later versions, such as SuperMemo 17 (2008) with the SM-17 algorithm and SuperMemo 19 (2025) enhancing incremental learning with automated import of educational videos and AI-driven tools, continued refining the balance between algorithmic precision and user-friendly software features.^[3]^[2] These developments, driven primarily by Piotr Woźniak, continually refined the balance between algorithmic precision and user-friendly software features.^[9]

Core Principles

Spaced Repetition System

The spaced repetition system (SRS) is a learning technique that schedules reviews of educational material at increasing intervals based on the psychological principle of the forgetting curve, aiming to optimize long-term retention while minimizing the time spent studying.^[16] This approach counters the natural decay of memory by timing repetitions to occur just before forgetting is likely, thereby strengthening recall efficiency for individual knowledge items such as flashcards or facts.^[17] The foundational theory draws from Hermann Ebbinghaus's 1885 experiments on memory, which demonstrated that retention declines rapidly after initial learning but can be stabilized through timely reviews, as quantified in his seminal work Über das Gedächtnis.^[18] In SuperMemo, this forgetting curve is adapted to account for variations in individual item difficulty, recognizing that not all material forgets at the same rate; easier items require less frequent reviews, while harder ones demand more adjustments to prevent overload.^[19] This personalization shifts from Ebbinghaus's uniform nonsense syllables to real-world knowledge, enhancing applicability for diverse learners.^[20] Central to SuperMemo's SRS are three key components: item priority, which ranks material by importance to focus reviews on high-value content first; review intervals, which expand progressively (e.g., from days to years) based on successful recalls to exploit memory stabilization; and ease factor adjustments, which modify future spacing according to the learner's subjective ease of recall for each item, fine-tuning difficulty on a per-item basis.^[21]^[22] These elements work together to minimize study time while targeting near-perfect recall, typically aiming for 90-95% retrievability across a collection.^[16] Unlike methods focused on short-term cramming for exams, SuperMemo's SRS emphasizes optimization for lifetime learning, where repetitions are regulated to build enduring knowledge structures over decades, reducing the cumulative burden of reviews as stability grows exponentially with consistent use.^[23] This long-term orientation supports sustained personal development, such as language mastery or professional expertise, by prioritizing efficiency in perpetual knowledge maintenance.^[24]

Incremental Reading Technique

The Incremental Reading technique, developed by Piotr Wozniak, was introduced in SuperMemo 10 in 2000 as a method to process and learn from extensive textual materials such as books, articles, and web content by breaking them into manageable portions over time.^[25] This approach addresses the limitations of linear reading by allowing users to handle thousands of articles simultaneously, converting imported electronic texts into durable, well-structured knowledge through iterative refinement.^[26] Unlike traditional reading, it leverages interruption and prioritization to align with human memory processes, enabling deep engagement without overwhelming the learner.^[27] The process begins with importing text into SuperMemo, which can be done via copy-paste (Ctrl+N), mass web import (Shift+F8), or dedicated tools for sources like Wikipedia or local files. As of version 19.1 released in April 2025, it also supports automated import of educational videos to enhance multimedia processing in incremental learning.^[4] Users then read in small increments, marking progress with read points and extracting key snippets (Alt+X) to create focused elements. These excerpts are transformed into cloze deletions—gap-filling questions (e.g., "The capital of France is [Paris]")—using Alt+Z, which facilitates active recall.^[26] Excerpts are scheduled for review based on user priorities and integrated into a hierarchical knowledge tree, where concepts branch logically from prioritized snippets, postponing less essential parts until readiness.^[25] This builds an interconnected web of knowledge, with elements rescheduled dynamically using tools like Ctrl+J for interval adjustments or Shift+Ctrl+R for postponement.^[26] Incremental Reading offers significant benefits by reducing the cognitive load of voluminous content, allowing learners to skim and delay non-critical sections, which can accelerate overall information processing compared to linear methods.^[27] It integrates seamlessly with spaced repetition for individual elements, ensuring high retention (defaulting to about 95%) and fostering deep comprehension through gradual assembly of ideas, much like solving a complex puzzle over time.^[27] Users report enhanced pleasure in learning due to variety and reduced monotony, with the technique supporting creative synthesis across diverse topics.^[27] Key tools include article scheduling via Alt+P to set priorities (e.g., A-Factor for similarity-based queuing), extraction for isolating valuable fragments, and rescheduling options to manage overload, such as auto-postpone for spreading reviews.^[26] These features enable fine-tuned control, with the knowledge tree serving as a dynamic registry that evolves as new extractions are added and reviewed.^[25]

Algorithms

SM-2 Algorithm Details

The SM-2 algorithm, introduced in SuperMemo 1.0 in 1987, serves as the foundational spaced repetition scheduling method for optimizing memory retention by dynamically adjusting review intervals based on user performance. It operates on individual knowledge items, each assigned an initial easiness factor (EF) of 2.5, which represents the perceived difficulty and influences interval growth. The algorithm uses a 0-5 grading scale for user recall quality, where 5 indicates perfect response, 4 is hesitant but correct, 3 is correct but laborious, 2 and 1 denote incorrect responses with partial recall, and 0 signifies complete failure.^[28]^[29] The core interval calculation begins with fixed short intervals for initial repetitions to establish baseline recall: the first repetition occurs after 1 day (I(1) = 1), and the second after 6 days (I(2) = 6). For subsequent repetitions (n > 2), the interval is computed as I(n) = I(n-1) \times EF, with any fractional results rounded up to the next whole day. After each review, the user provides a grade (q), and the EF is updated using the formula EF' = EF + (0.1 - (5 - q) \times (0.08 + (5 - q) \times 0.02)); the EF is then floored at a minimum of 1.3 to prevent excessively short intervals, though it has no explicit upper bound beyond the initial 2.5. If the grade q is less than 3, the item is considered a failure, resetting it to the first repetition stage (next interval = 1 day) without altering the EF, ensuring problematic items receive intensive relearning. Additionally, within a single learning session, items graded below 4 must be repeated immediately until achieving at least a 4, promoting session-level mastery before advancing.^[28]^[29] Key parameters include the repetition count (n), which tracks progress from initial learning, and the inter-repetition interval (I), which grows multiplicatively for "young" items during the first two repetitions before fully relying on EF scaling. This structure prioritizes rapid early feedback while allowing easier items to space out over time. However, SM-2's fixed EF adjustment lacks decay mechanisms or probabilistic modeling, often leading to interval overestimation for long-term retention as user performance stabilizes, which can result in suboptimal scheduling for advanced learners.^[28]^[30]

Advanced Algorithms (SM-4 to SM-19)

The advanced algorithms in SuperMemo, spanning SM-4 to SM-19, represent iterative refinements to the foundational spaced repetition system, incorporating adaptive mechanisms for difficulty, workload management, and predictive modeling to optimize long-term retention while minimizing review overhead. Building on the baseline SM-2 approach, these versions leverage user-specific data to dynamically adjust repetition intervals, with a core emphasis on personalization derived from historical performance metrics. This evolution enabled more efficient learning schedules, particularly for large collections, by addressing limitations in static interval calculations and introducing predictive tools for forgetting and priority.^[31] SM-4, completed in February 1989, marked an early advancement as the first adaptable algorithm, enhancing difficulty estimation through dynamic updates to the E-Factor and an optimum interval (OI) matrix that allowed intervals to vary according to perceived challenge rather than uniform progression. These features improved scheduling flexibility, though empirical validation was limited to initial user tests showing modest gains in retention for diverse subjects.^[32]^[33] Subsequent iterations, such as SM-5 in October 1989, enhanced these foundations by adding a forgetting index—a metric tracking the proportion of items recalled below a quality threshold (typically set at 10%)—to calibrate overall retention and adjust global parameters accordingly. Workload buffering was introduced via interval dispersal, using a matrix of optimal factors (OF) to randomize near-optimal intervals and prevent clustering of reviews, thereby smoothing daily loads; for instance, the formula for near-optimal intervals disperses around the primary interval by a factor derived from exponential probability distributions. This reduced scheduling lumpiness and doubled acquisition rates in practice, with retention reaching 95% over 47-day intervals compared to shorter spans in prior versions.^[34]^[35] During the 1990s, algorithms SM-8 through SM-12 further refined difficulty handling with leech detection, which flagged problematic items (leeches) exhibiting repeated failures—defined as lapses exceeding a threshold like 5 in 10 repetitions—for targeted intervention or removal, preventing disproportionate time sinks. Matrix-based E-Factor adjustments were incorporated, using a grid of optimal factors interrelating repetition number, E-Factor, and interval growth to propagate updates efficiently across the database. These changes, tested in collections exceeding 10,000 items, minimized interference from outliers and stabilized E-Factors around user-specific baselines, enhancing overall efficiency without exhaustive recomputation.^[36]^[37] Priority queues were introduced in later versions to sort items by estimated importance and urgency, facilitating focused reviews on high-value content amid growing databases.^[38] The most sophisticated advancements came with SM-17 in 2016 and SM-18 in 2019, which employed neural network approximations to model forgetting curves, drawing on vast historical data from millions of repetitions to personalize stability and retrievability estimates. Central to these is the AWOT model—encompassing activation (item salience), workload (daily review burden), and optimum time (ideal review timing)—which optimizes schedules across scales from seconds to decades. Retrievability R is computed as R = 0.9^{t/S}, where t is the time elapsed since the last review and S is the predicted stability in days, enabling precise predictions of recall probability (approximating the exponential form e^{-t/S}). By integrating full repetition histories, these algorithms reduced required reviews by up to 50% relative to SM-2, achieving grade deviations as low as 18% in large-scale validations and superior R-metric performance in 12-month comparative tests.^[39]^[40]^[41] Algorithm SM-19, introduced in SuperMemo 19 in 2023 and updated through April 2025, further refines these neural approximations with enhanced stability curves and a universal metric for algorithm comparison, outperforming prior versions and competitors like FSRS in retention efficiency as of May 2025. As of October 2025, development of Algorithm SM-20—an AI-based model incorporating expert knowledge of brain and memory processes—is underway, promising even greater precision in spacing predictions.^[7]^[42]^[43]

Software Implementations

Desktop Applications

SuperMemo's desktop applications are designed exclusively for Microsoft Windows operating systems, providing a full-featured environment for knowledge management and spaced repetition learning. Since SuperMemo 7, released in 1993, the software has remained Windows-only, with the current iteration, SuperMemo 19.1 (initially introduced as version 19 in 2023), released in April 2025 and offering 64-bit support for modern systems.^[44]^[45] The core functionality revolves around an element browser that allows users to navigate and manage learning materials in table or thumbnail display modes, facilitating efficient organization of questions, answers, and multimedia content.^[46] During the learning process, users input grades on a 0-5 scale to assess recall quality, which informs the underlying spaced repetition algorithm to schedule future reviews optimally; this process supports various test formats, including multiple-choice, spelling, occlusion, and cloze deletions.^[46] A priority queue visualizes pending tasks, employing auto-postpone and auto-sort mechanisms to prioritize high-importance items based on user-defined metrics.^[46] Import capabilities enable seamless integration of content from text files, HTML sources, web pages such as Wikipedia or YouTube, and email attachments, allowing users to build collections incrementally.^[46] The user interface organizes the knowledge base in a hierarchical tree structure, enabling users to categorize elements into branches for topics and subtopics, which enhances navigability for large collections.^[46] Customization options include adjustable interface levels (e.g., Beginner or Basic modes) and the use of templates, stylesheets, and font registries to tailor element appearance and formatting without altering the core layout.^[47] Integration with email supports incremental processing of incoming messages for content import, using tools like Windows Mail or Microsoft Outlook to extract and convert articles directly into learning elements.^[46] SuperMemo operates on a commercial model, with SuperMemo 19 available as a one-time purchase for a lifetime license priced at $66 USD.^[45] Older versions, such as SuperMemo 9 and SuperMemo 15, are provided as freeware for users seeking entry-level access without cost.^[44]

Mobile and Cross-Platform Versions

SuperMemo's mobile applications were first introduced in the mid-2010s, with initial versions for Android launching around 2016 and iOS following shortly thereafter; however, these legacy apps were discontinued on September 15, 2025, with users encouraged to migrate to updated versions released around 2023 that support offline review modes allowing downloads of courses for practice without internet connectivity.^[48]^[49] The current iOS app enables similar functionality on Apple devices, though it faces limitations due to platform-specific restrictions on file handling and background processing, which restrict deeper customization compared to Android.^[50] Both apps emphasize cloud synchronization via the SuperMemo.com platform, permitting seamless progress transfer across devices upon reconnection, thus maintaining consistent repetition schedules.^[48] The user interface in these mobile versions is optimized for touch input, featuring simplified navigation with swipe gestures for card reviews, customizable daily learning plans, and AI-assisted tools like MemoChat for conversational practice, all while prioritizing portability over comprehensive content creation.^[6] Compatibility with desktop collections is achieved through XML-based data exchange, specifically the supermemo.net XML specification for commercial courses, allowing subset syncing of question-and-answer elements to minimize data usage and enable offline portability; however, full user-generated collections from the Windows desktop version require conversion due to incompatible XML formats.^[51] Web-based tools provide browser access to SuperMemo collections via the online platform at supermemo.com, functioning as a reference app for reviewing downloaded materials without installation, and supporting cross-device continuity through cloud storage.^[6] An API for third-party integrations is in development, aimed at embedding SuperMemo's spaced repetition algorithms into external applications, though as of November 2025, it remains forthcoming for broader dissemination.^[52] Despite these adaptations, mobile and web versions exhibit reduced support for advanced features like incremental reading, which is primarily available in the desktop application, shifting the focus to review-only workflows for on-the-go learning efficiency.^[51]

Impact and Applications

Adoption in Education and Self-Learning

SuperMemo has found significant adoption among language learners seeking to build and retain extensive vocabularies efficiently. For instance, it is particularly popular for English acquisition, where users report mastering thousands of words and phrases through daily repetitions, often achieving fluency without immersion in native environments. Medical students and professionals also represent a core user base, utilizing the software to memorize complex facts such as drug dosages, anatomical details, and diagnostic criteria, with medical sciences ranking as the second most common application after languages.^[53]^[54]^[54] Piotr Wozniak, the creator of SuperMemo, exemplifies its personal application in self-learning, having employed the system daily since the 1980s to pursue polyglotism and deepen his expertise in computer science. Through consistent use over more than 30 years, Wozniak achieved fluency in English solely via spaced repetitions without traveling to English-speaking countries and maintained a diary in Esperanto for two years to reinforce linguistic skills. His routine integrates SuperMemo for broader knowledge acquisition, including biology and technical subjects studied during his time at Poznan University of Technology.^[55]^[55]^[56] In educational contexts, SuperMemo supports shared collections tailored to subjects like history, programming, and professional certifications, enabling collaborative learning among students and educators. A notable integration is the partnership with Medico in Poland, which deploys SuperMemo's algorithms for preparing medical residents for the State Specialization Examination, incorporating nearly 20,000 questions optimized for individualized review schedules. Studies and user reports indicate that consistent application yields retention rates of 90% or higher at review points, with average overall retention approaching 95%, far surpassing traditional cramming methods that achieve only 20-25% long-term recall.^[57]^[53]^[58]^[59] The SuperMemo community, centered on forums and resources at supermemo.guru, facilitates the exchange of user-created decks for diverse topics, fostering a collaborative ecosystem for self-learners and educators. This platform has contributed to the software's evolution from Wozniak's personal tool in 1985 to a system estimated to have reached 5 million users by 2007, with 20,000 to 200,000 active users in the late 2000s; as of 2025, mobile apps show continued niche adoption with thousands of downloads annually, reflecting steady growth in lifelong learning despite a specialized user base. Incremental reading, a complementary feature, aids in processing text-heavy materials like academic articles, enhancing comprehension in knowledge-intensive fields.^[57]^[60]^[61]^[50]

Influence on Other Tools and Criticisms

SuperMemo's SM-2 algorithm, first implemented in 1987 and publicly detailed shortly thereafter, has profoundly shaped the development of subsequent spaced repetition software by providing a foundational, accessible model for optimizing review intervals based on user performance.^[13] This algorithm directly inspired Anki, released in 2006, which adopted SM-2 with minor modifications as its default scheduling mechanism to enhance long-term retention through adaptive spacing. Similarly, the open-source Mnemosyne flashcard program, launched in 2006, incorporated SM-2 to calculate repetition intervals, crediting SuperMemo for pioneering practical applications of spaced repetition in digital tools. Duolingo's review system, while evolving into a proprietary model, draws from the spaced repetition principles popularized by SuperMemo, as evidenced in analyses of its optimization efforts that reference SM-2 as a benchmark for memory scheduling in language learning apps.^[62] Beyond specific implementations, SuperMemo has played a pivotal role in mainstreaming spaced repetition within educational technology, influencing a wide array of platforms and research by demonstrating empirical improvements in memory retention.^[63] Its algorithms are frequently cited in cognitive science literature on memory optimization, such as studies developing adaptive scheduling frameworks that build upon SuperMemo's heuristic approaches to personalize learning intervals and reduce forgetting rates.^[62] This legacy extends to edtech innovations, where SuperMemo's emphasis on incremental, data-driven reviews has informed systems in apps like Quizlet and broader pedagogical tools, fostering a shift toward evidence-based retention strategies over rote memorization.^[64] Despite its contributions, SuperMemo has faced criticisms regarding usability and accessibility, particularly the steep learning curve associated with its incremental reading feature, which requires users to master complex workflows for processing and prioritizing information extracts.^[60] The desktop version's exclusivity to Windows has limited its reach, excluding users on macOS, Linux, or other platforms without emulation, thereby hindering broader adoption in diverse computing environments.^[1] Additionally, the system's reliance on subjective user grades for interval adjustments has been noted to potentially lead to suboptimal scheduling, as grading accuracy varies and can weakly correlate with actual retention outcomes.^[64]

References

[1]
SuperMemo method
a young, talented scientist and student of molecular ...
[2]
Development of SuperMemo (1985-2013)
Oct 15, 2013 · SuperMemo's development includes the first spaced repetition algorithm (1985), first computer application (1987), variable intervals (1989), ...
[3]
Four Decades: From SuperMemo 1.0 to SuperMemo 19.0
Apr 12, 2025 · SuperMemo 1.0 ( ) was written in 1987, using Turbo Pascal 3.0. It was written by Piotr Wozniak. It was the first ever computer ...Missing: development | Show results with:development
[4]
What's new in SuperMemo 19?
Apr 11, 2025 · 10 YouTube videos featuring American biologist and 1972 Nobel prize winner Gerald Edelman are available and selected for automatic import into SuperMemo.
[5]
Official websites - SuperMemo
Apr 30, 2020 · SuperMemo World is a software company founded on July 5, 1991 by Krzysztof Biedalak and Piotr Wozniak in Poznan, Poland.
[6]
SuperMemo app - iOS, Android, PC, online
Download the SuperMemo app to your smartphone or tablet, or learn using a web browser. Online learning is a pleasant and straightforward way to get to know a ...
[7]
The best spaced repetition algorithm (2025) - SuperMemo Guru
Jun 3, 2025 · This text will argue that SuperMemo 19 is still the best (May 28, 2025). The claim is made on the basis of universal metric contest as well as theoretical ...
[8]
http://supermemo.guru/wiki/Piotr_Wozniak
[9]
The true history of spaced repetition - SuperMemo
Jun 1, 2018 · Using an obvious intuition, Wozniak attempted to measure the retention of knowledge after different inter-repetition intervals, and in 1985 ...
[10]
http://supermemo.guru/wiki/Birth_of_SuperMemo
[11]
http://supermemo.guru/wiki/SuperMemo_1.0_for_DOS_(1987)
[12]
http://super-memory.com/english/history.htm
[13]
SuperMemo 2: Algorithm - Super Memory
Algorithm SM-2 used in the computer-based variant of the SuperMemo method and involving the calculation of easiness factors for particular items.Missing: 1988 | Show results with:1988
[14]
SuperMemo World - Super Memory
SuperMemo World is a software company founded on July 5, 1991 by Krzysztof Biedalak and Piotr Wozniak in Poznan, Poland.
[15]
History of SuperMemo
The text about Piotr Wozniak and SuperMemo World was added in 1991 when releasing SuperMemo 2.0 as shareware. The name SuperMemo was first used on Apr 27, 1988.
[16]
SuperMemo Website Contents - Super Memory
- SuperMemo 2006 and SleepChart can be used to investigate the ... Implementing neural network SuperMemo - how neural network SuperMemo will be trained ...
[17]
Spaced repetition and learning — how does it work? - SuperMemo
Aug 31, 2021 · It involves regulating repetitions so as to minimise them, and so minimise the time spent on learning, but at the same time achieving the desired outcome.
[18]
Forgetting curve - SuperMemo Guru
May 10, 2025 · This is used in spaced repetition to minimize the cost of an indefinite retention of memories. Power or Exponential? Forgetting is exponential, ...
[19]
https://www.supermemo.com/en/blog/history-of-spaced-repetition
[20]
Ebbinghaus and the forgetting curve - SuperMemo
Sep 2, 2021 · The results are shown on the table below, where the number of repetitions necessary to memorise a series comes to 32. Spaced repetition. Source: ...
[21]
Replication and Analysis of Ebbinghaus' Forgetting Curve - PMC - NIH
Jul 6, 2015 · We present a successful replication of Ebbinghaus' classic forgetting curve from 1880 based on the method of savings.
[22]
E-Factor - supermemo.guru
Apr 8, 2019 · Algorithm SM-17 does not use E-factors or A-factors. Instead, difficulty is obtained by matching item's repetition history to the stability ...
[23]
Priority - SuperMemopedia
Mar 18, 2020 · All articles in incremental reading and all items in spaced repetition are sorted by priority. Elements sorted by priority can be viewed as the ...
[24]
Theoretical aspects of spaced repetition in learning - SuperMemo
May 20, 2000 · Piotr Wozniak, 1990-2000. This text was derived from P.A.Wozniak, Optimization of learning : Simulation of the learning process conducted ...Missing: history | Show results with:history
[25]
General principles of spaced repetition - SuperMemo Guru
Apr 21, 2025 · Spaced repetition dramatically reduces the time necessary for review. This should solve quite a portion of your learning problems.
[26]
History of incremental reading - SuperMemo Guru
Apr 22, 2025 · ... SuperMemo 8 (e.g. in language learning); 1998: introducing richly formatted text that encourages copy and paste from the internet, and ...
[27]
Incremental reading - SuperMemo 19 Help
Sep 30, 2023 · In incremental reading, you read articles in small portions. After you read a portion of one article, you go on to a portion of another article, ...
[28]
Advantages of incremental reading - SuperMemo Guru
Mar 30, 2025 · As incremental reading is based on spaced repetition, all memories that you form while learning will be indefinitely protected from forgetting.
[29]
Algorithm SM-2 used in the computer-based ... - SuperMemo.com
I wrote the first SuperMemo program in December 1987 (Turbo Pascal 3.0, IBM PC). It was intended to enhance the SuperMemo method in two basic ways:.
[30]
Algorithm SM-2 - SuperMemo Guru
May 2, 2025 · Algorithm SM-2 was the first computer algorithm for computing the optimum schedule in spaced repetition. It was first used in SuperMemo 1.0 for DOS on Dec 13, ...
[31]
SuperMemo Algorithm FAQ
6 Will SuperMemo use neural networks? 7 New SuperMemo algorithms are superior, but this should not determine your choice of your learning program. 7.1 ...
[32]
SuperMemo Algorithm
Jun 4, 2019 · 2002 introduced the first SuperMemo algorithm that is resistant to interference from delay or advancement of repetitions.Missing: 3 12
[33]
Account of research leading to the SuperMemo method
Feb 2, 1990 · Account of research leading to the SuperMemo method version 3.1. This text was taken from P.A.Wozniak 1990.
[34]
SuperMemo Algorithm - Super Memory
Algorithm SM-17 is the ultimate practical proof for the correctness of the two component model. Past vs. Future. All SuperMemo algorithms before the year 2015 ...
[35]
Further improvement of SuperMemo: introduction of the matrix of ...
Mar 1, 1990 · Further improvement of SuperMemo: introduction of the matrix of optimal factors. This text was taken from P.A.Wozniak, 1990.Missing: 1988 | Show results with:1988
[36]
https://super-memory.com/english/algsm8.htm
[37]
https://help.supermemo.org/wiki/Leeches
[38]
Leeches - SuperMemo 19 Help
Aug 16, 2025 · These parameters can be defined when using View : Other : Leeches (Shift+F3). An element becomes a leech when the number of memory lapses passes ...Leech Hunting Options · Leech Alert Wizard · Defining And Viewing LeechesMissing: SM- 8 12 detection matrix- based EF adjustments<|control11|><|separator|>
[39]
Algorithm SM-17 - supermemo.guru
Jul 18, 2025 · Algorithm SM-17 is a modern version of the spaced repetition algorithm used in SuperMemo. It was developed, implemented, and tested in the years 2014-2016.Missing: AWOT | Show results with:AWOT
[40]
Algorithm SM-18 - SuperMemo Guru
May 2, 2019 · Algorithm SM-18 is the newest version of spaced repetition algorithm used in SuperMemo (2019). The main difference is the new model for estimating item ...Missing: AWOT formula
[41]
http://super-memory.com/articles/stability.htm
[42]
SuperMemo Versions - SuperMemopedia
Nov 21, 2024 · SuperMemo 9 (98/99) - This software is freeware. It's an old and no longer supported. It was a predecessor of the multimedia and pro ...
[43]
SuperMemo 19 for Windows
In stockSuperMemo 19 for Windows helps you import knowledge from the web, organize it, and keep it in your memory at minimum cost for ever.Missing: features platforms
[44]
Features - SuperMemo 19 Help
Mar 16, 2024 · efficient learning: SuperMemo started in 1987 as the first freeware and later commercial software to implement spaced repetition.
[45]
Using registries - SuperMemo Help
### Summary: Customizable Skins or UI Customization in SuperMemo Desktop
[46]
SuperMemo.com apps for Android, iOS and PC Windows
Sep 19, 2016 · With SuperMemo.com apps, you can learn both online and offline, and transfer your learning between devices at no extra charge.
[47]
SuperMemo – effective learning - App Store - Apple
Rating 4.3 (128) · Free · iOSCombine language learning with the proven method of spaced repetitions and innovative features based on AI technology.Missing: mobile | Show results with:mobile
[48]
SuperMemo data exchange via XML - SuperMemopedia
Jun 25, 2025 · SuperMemos use two XML specs that are not mutually compatible: (1) supermemo.net commercial courseware XML, and (2) SuperMemo for Windows XML ( ...
[49]
Spaced repetition, AI & RAG in publishing and education
Soon, you'll be able to integrate this scientifically proven approach into your own systems with the upcoming SuperMemo spaced repetition API. In parallel ...Spaced Repetition, Ai & Rag... · Supermemo E-Learning... · Memochat: Ai Agent...Missing: third- | Show results with:third-
[50]
Medico and SuperMemo: Transforming Medical Exam Preparation ...
Feb 24, 2025 · SuperMemo, on the other hand, is renowned for its spaced repetition algorithms, which optimize memory retention through systematic review ...Missing: studies | Show results with:studies
[51]
FAQ: Learning with SuperMemo - Super Memory
Learning medical sciences with SuperMemo comes second in popularity to learning English! ... SuperMemo will stabilize around the programmed level of retention ...
[52]
Want to Remember Everything You'll Ever Learn? Surrender to This ...
Apr 21, 2008 · SuperMemo is the result of his research. It predicts the future state of a person's memory and schedules information reviews at the optimal time.Missing: origins | Show results with:origins
[53]
Piotr Wozniak - supermemo.guru
University education: Adam Mickiewicz University in Poznan (1980-1985), MS degree in molecular biology; University of Technology in Poznan (1985-1990), ...Missing: origins | Show results with:origins
[54]
Exponential adoption of spaced repetition - SuperMemo Guru
As of 2006, Mnemosyne keeps collecting repetition history data running a mutation of Algorithm SM-2. As a free multi-platform application, Mnemosyne quickly ...Slow Start Of Algorithm Sm-2 · 1998: Publishing And... · How Many People Use Spaced...
[55]
Forgetting index in SuperMemo
Apr 6, 2019 · ... 90% of the material at repetitions. This does not imply that your knowledge retention will be 90% only. Your average retention will be nearly 95 ...Missing: studies | Show results with:studies
[56]
Incremental learning - SuperMemo 19 Help
Spreading activation. Spreading activation is an algorithm for activating nodes in a neural network, semantic network, directed graph, etc. ... SuperMemo 2006 ...
[57]
Enhancing human learning via spaced repetition optimization - PNAS
Jan 22, 2019 · In this work, we develop a computational framework to derive optimal spaced repetition algorithms, specially designed to adapt to the learners' performance.
[58]
SuperMemo is better than FSRS, by far!
Oct 13, 2025 · Algorithm SM-20 (a new AI algorithm based on SuperMemo ... learning in artificial neural networks and the first step to superintelligence.