Yes, and she says it quite clearly: > Now, what we do when we want to explain what a galaxy does, or a galaxy cluster, or even the whole universe, is not to pl

Yes, and she says it quite clearly: > Now, what we do when we want to explain what a galaxy does, or a galaxy cluster, or even the whole universe, is not to plug the matter and energy of every single planet and star into the equations. This would be computationally unfeasible. Instead, we use an average of matter and energy, and use that as the source for gravity. > Needless to say, taking an average on one side of the equation requires that you also take an average on the other side. But since the gravitational part is non-linear, this will not give you the same equations that we use for the solar system: The average of a function of a variable is not the same as the function of the average of the variable. We know it’s not. But whenever we use general relativity on large scales, we assume that this is the case. > So, we know that strictly speaking the equations we use are wrong. The big question is, then, just how wrong are they? [emphasis added] It's not a "hunch". It's just how the math works. average of 0-9 = 4.5 4.5^2 = 20.25 average of squares of 0-9 = 28.5