Some new research indicates that teaching kids all these word problems about speeding trains and slices of pie may be a mistake, and that children actually learn math better if you just teach them abstract equations from the get-go. Obviously, I'm in no position to judge (my boring guess is that different approaches work best for different people and there should be some diversity of classroom methods available for different students), but it's a bit bizarre how little effort we put into developing serious research-based pedagogical methods. You'd think this would be a major component of federal education policy, but it's really not.