<p>I think a mixture of theory and practice, more inclined towards theory is most appropriate for students. Especially in subjects such as computer science, one cannot really understand a concept well unless they are able to program it. For example, one cannot really understand the fine details of a file system without really implementing at least part of one. Top schools such as Stanford and UCB require OS students to construct these components and I would certainly consider them to be “practice”. Similarly, one cannot be a CS major knowing all the theoretical foundations (lambda calculus, turing machines, computability) without knowing how to program!</p>
<p>Similarly, in EE, I feel people should know how to construct circuits from breadboards and should have some exposure to soldering and using circuit components. Without some degree of practice, theory can be easily forgotten. Also, theory often simplifies aspects of how devices work in the real world and exposure to practice helps students to understand the limitations of the theory model. But yes I agree that all practice without theory is a step to failure and that students must understand how to think generally and that the theoretical foundations apply similarly for most vendor products. Math and physics (even higher level ones) are central to being able to innovate.</p>