While this algorithm is very simple it doesn't seem to take into account critical periods in memory formation.
Just choosing which card to show next is only half the battle. If you're learning X cards you need to review all X within the first hour, then the next period of 24-48 hours, etc. The forgetting curve which underlies spaced repetition has some biological tipping points (did you sleep between your last session and this one) which are not easily approximated.
I strongly support software like this, but would prefer a complex accurate system, to a simple elegant one.
If anyone is interested in this sort of thing, another application which uses a slightly more complicated (but open) algorithm is "mnemosyne" (mnemosyne-proj.org). Its algorithm is based on an older algorithm from SuperMemo.
While I agree with iandanforth that you're missing an important aspect of this, I'd like to commend you on what you have. My first attempt at writing a spaced-repetition algorithm turned out with the same results as yours, however your algorithm is a much neater way of looking at things. If you take iandanforth's comment to heart try to incorporate it into what you already have, instead of rewriting.
Just choosing which card to show next is only half the battle. If you're learning X cards you need to review all X within the first hour, then the next period of 24-48 hours, etc. The forgetting curve which underlies spaced repetition has some biological tipping points (did you sleep between your last session and this one) which are not easily approximated.
I strongly support software like this, but would prefer a complex accurate system, to a simple elegant one.