With respect to spaced repetition algorithms, if anyone is interested in using something other than SuperMemo's SM2 algorithm, I devised the following for my JL

With respect to spaced repetition algorithms, if anyone is interested in using something other than SuperMemo's SM2 algorithm, I devised the following for my JLDrill program: http://jldrill.rubyforge.org/Strategy.html http://jldrill.rubyforge.org/Strategy.html It has a few advantages over SM2 from my perspective. Instead of trying to time reviews, it simply tries to order the vocabulary by likelihood of remembering it. You then drill the vocabulary until you show that you are getting somewhat over 90% chance of remembering the vocabulary. It then presents new vocabulary. You can also set a cutoff point where the ratio of actual elapsed time is over ideal review time is over a certain point. It can then "forget" these items, which handily helps people who have spotty review habits. Unfortunately, JLDrill currently suffers badly from bitrot and doesn't really work very well any more. I will admit that it wasn't particularly well written in the first place ;-) However, I recently dusted it off and if you are enterprising, you can probably get it mostly working: https://github.com/mikekchar/JLDrill https://github.com/mikekchar/JLDrill