I'm sure there are circumstances where there's good reason to firehose thousands of lines of complex unmaintainable code. But how common is such "throwaway" projects?
Usually when you are starting an actual new project that kinda throwaway is a very good way to understand business needs and problem domain.