I've tended to explain it to people this way -- In the related computer science fields there are only three kinds of problems:
1) The problems math makes for you. These are the kind of problems lots of people think they work on, but don't. Compared to other problems, there aren't a lot of them, but solving one has a huge impact on the field as one solution can be shared across the industry.
2) The problems physics makes for you. These are the problems you encounter when you take the math and start applying it to actual machines. This is hardware design or software to directly make that hardware do what you want. This is 'applied' computer science and again, many can leverage the work of a few.
3) The problems other software engineers make for you. This is what most computer scientists/engineers/programmers actually have to do. This is dealing with making someone else's code do what you want. It's dealing with API's and layers and layers of other, primarily human-created, problems.
1) The problems math makes for you. These are the kind of problems lots of people think they work on, but don't. Compared to other problems, there aren't a lot of them, but solving one has a huge impact on the field as one solution can be shared across the industry.
2) The problems physics makes for you. These are the problems you encounter when you take the math and start applying it to actual machines. This is hardware design or software to directly make that hardware do what you want. This is 'applied' computer science and again, many can leverage the work of a few.
3) The problems other software engineers make for you. This is what most computer scientists/engineers/programmers actually have to do. This is dealing with making someone else's code do what you want. It's dealing with API's and layers and layers of other, primarily human-created, problems.