Everything you're saying is reasonable but this phrase reminded me, one of my favorite profs used to say, "Knowing about computational complexity won't ever get you a job, but it will keep you from being fired." Several times I've started a job and discovered my predecessor had coded himself into a corner because he didn't see the combinatorial explosion on the other side of what he was doing.
Hard saying in much detail; after all, I showed up after they left. Factually at least twice they were highschoolers or college dropouts. In another case he had a history degree, and a business degree in another. I now work for the NRAO, so I can tell you any level of education in a physics or math track is not necessarily sufficient to learn how to program--the physicists in particular seem wholly unable to perceive code quality or understand what might affect performance and what wouldn't. This, of course, doesn't stop them from writing their own code and giving opinions about what's wrong with your code, sight unseen. ("I think your program is calling malloc and free incorrectly."--scientist, about my Java code).
Everything you're saying is reasonable but this phrase reminded me, one of my favorite profs used to say, "Knowing about computational complexity won't ever get you a job, but it will keep you from being fired." Several times I've started a job and discovered my predecessor had coded himself into a corner because he didn't see the combinatorial explosion on the other side of what he was doing.