A commonly accepted hierarchy of cognitive skills puts analysis and synthesis near the top, with straightforward knowledge and comprehension at the bottom. A typical Computer Science curriculum, though, usually starts with programming, an activity that requires both analysis and synthesis. The student, without the right conceptual skills, attempts synthesis without analysis, resulting in poor programming skills. This paper presents a view of Computer Science curriculum, drawing on earlier work on abstraction-first learning, to propose a curriculum which starts from lower-order cognitive skills, while working up to higher-order skills in later years.