We construct models to explain the observations, and these models predict that to some extent, dark matter interact with normal matter. There is no way out if one wants to reproduce data. Those interactions are very weak, but they must be there.