Collectively, we have been exploring this middle ground since the invention of the first digital computer. The history of programming languages is the history of increasingly powerful computational abstraction. Where once we manipulated binary digits in registers, we now manipulate objects, procedures, and graphics which, while ultimately still only stored binary digits, are taken to represent all manner of things from architectural designs, to mathematical theory, to small virtual worlds.
The challenge to statistical computing is to determine the appropriate abstractions for statistical design and analysis, to figure out how to implement them computationally, and to develop interactive tools which facilitate their use.
Meeting this challenge requires careful design, both of appropriate programming abstractions and of graphical user interface. The goal is to seamlessly integrate the two approaches.