Well, one reason is that we always have to work with the quirks of each model. So, a know model is often preferred over a new/unknown one because we have to be vigilant again. (Negative) surprises are mentally exhausting in the long run. IMO, you can work much better when you know the model.