Most of he problems we solve with computers are mundane - from batch programs to update files from data tapes in the 1960s to making a phone UI talk to a CRUD back-end in the present day. This kind of ordinary programming doesn't really require much computer science and never has. However, there will always be more interesting applications (or interesting sub-problems of otherwise mundane applications) that stretch the limitations of current knowledge and technology, and these will benefit from advances in computer science and from developers who keep up with CS research. But it has never been the case that every programmer needed to be a computer scientist.

There is a tension to use your time in the best way. Recently someone asked in reddit Haskell what's the best language to learn to obtain some statistics and finance, my answer was R. Computer Science should be related to some field in which you are an expert or know about. One extreme side of the coin is being too pragmatic the other extreme is wasting your time in the hippies things.

The difficulty is putting the real amount of time in research and in getting the tools that will me your goal became a reality, that's a touch problem.