[orbifold]

To keep in your analogy working in mathematics is more like working with a codebase that is ~100 years old and that still contains all the dead ends, unused code and false starts, because no code ever gets deleted. It is basically impossible to navigate modern research mathematics without a team of professionals guiding you for quite some time. Some of the most successful research mathematicians manage to either find a largely unexplored niche (Jacob Lurie with Higher Category Theory) or manage to work their way backwards to a research level understanding of a specific already established field (Peter Scholze, he claims that he basically worked his way backwards from the Langlands Conjectures and never took a course in Linear Algebra, but just picked it up on the way).