I'm not sure I can make sense of that question.
No one can know everything, but if your specialty is, for example, virtual memory paging algorithms you should be able to write programs that implement and comprehend those algorithms, even if it's not production-quality code. If you are on the more practical side you should be able to read and hack on the existing code, even if it's not the central focus of your job.
I'm sure it's possible to find people who can be considered computer scientists but don't code very well or at all. But those people are rare exceptions, and aren't evidence that programming isn't an important tool in computer science.
There is a place for academia and mathematic approaches, and there is a place for real world experience and technical abilities. Ultimately someone who is good at being programmer does need a leavening of academic background, and someone who is a successful academic needs to know something about real computers instead of the academic over-simplified illusory world that contain friction-less pulleys and massless ropes.
I really don't think Knuth of all people can be held up as an impractical egghead. Compared to many other academics, he has always maintained a view towards real computers.