I used to do that, but now I try to think in terms of the abstraction "list" and "array", and not worry about the underlying data structure unless I'm specifica

I used to do that, but now I try to think in terms of the abstraction "list" and "array", and not worry about the underlying data structure unless I'm specifically thinking about efficiency. For example there are multiple ways of implementing lists and arrays that have similar interfaces but different efficiencies for insertion/sort/search, etc. So basically the dichotomy you propose is just "too simple," it doesn't cover all the options and combinations that a programmer may need to think about.