Great article! Just a note: spaced repetition is not only for rote learning, or "remembering things". When talking about spaced repetition, we need to make a d

Great article! Just a note: spaced repetition is not only for rote learning, or "remembering things". When talking about spaced repetition, we need to make a distinction between repeated studying (seeing both the problem and answer) and testing. Once you've overlearned the material, the biggest value of an SRS comes from testing yourself, even if you don't see the solution after [1]. Another potential benefit is interleaving concepts from different problems [2]. Another distinction is what is being tested. For example, testing using same math problem you learned from is repeating both the concept learned and exact content of the problem. Repeating the exact content you originally studied may dilute the benefit of testing you on the concept. However, it is clear that repeated testing on the concept is very beneficial when the content is new (e.g. spaced testing related math problems, [3]). That isn't to say repeating the concept and content can't be beneficial, I think it can be useful, but I'm less sure about how it stacks up. In the posted article, I think the type of SR will be useful because (1) recalling the pieces being tested are helpful when solving the problem (or a related one), and (2) hopefully they cue you to think of the problems domain they're related to. In a sense, interleaving the pieces can be a test of matching the piece with it's problem domain. It seems like a better solution would be to create related problems related to the ones you learned initially, to test yourself in a spaced way. For example, you and a friend could solve the same Eloquent Javascript problems, and then create a handful of new problems that are in the same problem domain. Then, you could trade problems for spaced testing. This type of spaced testing could be thought of as repeating problem domain, but not exact problem content. To the degree that the SRS interleaves testing problem domains, it will likely be especially effective for solving new problems [2]. [1]: http://pss.sagepub.com/content/27/2/223.full.pdf+html http://pss.sagepub.com/content/27/2/223.full.pdf+html (based on this classic!: http://learninglab.psych.purdue.edu/downloads/2008_Karpicke_Roediger_Science.pdf http://learninglab.psych.purdue.edu/downloads/2008_Karpicke_...) [2]: http://www.indiana.edu/~pcl/papers/carvalho_goldstone_delayedgeneralization2014.pdf http://www.indiana.edu/~pcl/papers/carvalho_goldstone_delaye... [3]: http://onlinelibrary.wiley.com/doi/10.1002/acp.1266/abstract http://onlinelibrary.wiley.com/doi/10.1002/acp.1266/abstract