Bicicleta needs incremental recomputation, like a spreadsheet, to be reasonably efficient. And I just discovered that ten years ago this guy Umut Acar (and Guy Blelloch, et al.) found a super rigorous way to derive statically-type-safe incremental versions of non-incremental (i.e. batch) algorithms, and wrote his dissertation on it. It seems like this work ought to be pretty broadly applicable right now, especially relating to incremental updates of DHTML user interfaces, but I haven’t heard of anybody doing that — which seems totally insane to me! Is something wrong with the work, or is it just that nobody has heard of it?