Despite many accolades, I have not touched the recent sequels to Civilization. Civilization is not the same game when you play it a second time. When you already know the outcome of research combinations, the game became far too... predictable. The optimal path to win becomes clear, and before long you are so far ahead that you have modern tanks steamrolling over enemy musketeers.
When you have a single model and are unable to see combinations and interactions, your ability to make predictions is hampered. When you understand combinations and interactions, prediction of non-linear events improves and optimal paths become more clear.