> It is not very common to get good results from applying a coarse tool to a fine problem. From the blog post, that isn't really what happened here -- the prob

> It is not very common to get good results from applying a coarse tool to a fine problem. From the blog post, that isn't really what happened here -- the problem was accounting for reflections, and a technique was needed to estimate such effects. It doesn't matter that the anomalous acceleration was small compared to the total acceleration. It was of the same order as these reflection effects, and so only a "coarse" accounting of them was needed.