Spaced repetition is about graph connectivity

It would be useful to do any connecting activity when faced with a card; testing recall for real, while valuable, is only one of the ways to prompt salience scores being assigned. Think for example of typing, and the way character frequency relates to grapheme frequency and up to word frequency. The muscle memory is trained in terms of the extant relationships, and recall is made immediately relevant by usage (and therefore, proof of utility). Not everything can have a learning structure that pretty, of course, or that reflective of chunking as a process; life and most subjects have a higher cardinality than typing. (what is the cardinality of knowledge atoms in most subjects relative to typing? investigate). But the algorithm that tries to abstract out knowledge in terms of fact-atoms, should probably reflect this? Or somebody should write the algorithm that does.

Related:

Backlinks