As a programmer working on difficult and abstract problems, his method is not that usefull. The example was good for a single simple problem. But that is not really the problems you have when working on difficult problems. And writing visual-feedback software for complex problems would be like writing book authoring software that creates a movie from the book you are writing. It is a good and obvious idea. But absolutely unrealistic for real world problems.
Where it is *very* useful though is for programmers designing software. If we can make our software work like that in their limited domains, then it is a terrific way to solve our users problems.