You should never allow effects to define the way the view will look. Effects should happen between the changes of the view aspect. This is where states makes sense. You define a view that looks like A, and B. And then you define that for going from A to B, some kind of animations will occur. You should never say that “for the view to look like B, the effect needs to happen”. This creates a dependency on the effect (and a lot of code) (and problems refactoring).