It just depends on the logic of your form and your requirements.

If the SmartObject has the data you want, you can use it.  But if the data doesn't exist there yet (for example, your submit view that hasn't saved to the SmartObject yet), or if your form controls manipulate the data somehow (expressions, rules, etc) then you might want to use the control instead.

Does that make sense?

The smartobject is your dataset whether retrieved or being populated by activity on the form it is the smart object that is where all the data will reside in the end.

Controls are bound to smart objects to be prepopulated with data for whatever reason. Smart Objects are always needed for persistence of data, controls are not. Controls are focused on presentation and manipulation. Smart Objects, purely storage.

For example, you may not realize it but you can bind a smart object to a list view and not have one single control on the view, but iterate through the view to do something with the dataset (bound smartobject).

The actual use case of when you want to pick a smart object property and when you can to choose a control depend entirely on what you are doing with the data at that moment and whether or not that control is already bound to a smart object property. 

Hope this helped shed some more light on the subject.

S.