HELP CENTER & FAQ
What type of algorithm does Memory use?
Memory uses a proprietary Spaced Repetition algorithm behind the scenes to calculate when you should review each question. It does not use a traditional off-the-shelf 'Supermemo' or 'Leitner' algorithm; it instead uses a new proprietary system to serve you questions based on AI and tested algorithms. The fundamental principles are still based on spaced repetition learning and its foundational concepts are still from SM-2 and Leitner, but the Memory method is unique. Fundamentally, the system to calculate practice intervals is closer to SM-2, whilst the initial learning is more like Leitner. Unlike the original formulas, Memory takes various inputs such as your hesitation time, backspace hits, flashcard flips, average typing speed, and other variables to automatically calculate your performance to know the ideal time for you to repeat questions and set them due for practice at the correct interval. As well as this, the algorithm learns your style and adapt to your abilities to ensure it is fair and optimal for every user. From 2021, we offer the ability to configure variables in the algorithm. You can configure your own spaced repetition algorithm in your user settings. Rather than a basic 5 point accuracy system like other sites, with Memory's algorithm, you should note that if you are accurate and quick for a given question (compared to your typical speed), you will find yourself having to practice it less frequently. We measure your accuracy with higher granularity, saving you less time and improving your memory. Unlike other systems, Memory will not ask you to assess your own performance to calculate when to next show you a question - that is the job of our algorithm. If this all sounds complex, the best thing to do is create a course and get started and experience it for yourself. You don't need to worry about the details because we do. New for 2021: You can now "mark yourself" 1 - 5 in flashcard mode if this is your desired behaviour.
If there is a question and answer that you recommend we add, please let us know below: