For example lets say you're learning calc and you have a perfectly good book but the author's explanation of Green's theorem simply doesn't resonate with you. Yes, there's always wikipedia or mathworld or one of the online lecture series... but, if its free and fast and searchable, may want to try another book's perspective.
I took a EE type microwave class in the 90s where the instructor provided a lot of value by not specifying one textbook but gave us the best one or two photocopied pages from perhaps a hundred widely varying books. Sometimes the ideal textbook for a rare enough niche simply doesn't exist, or something like it existed but its 30 years out of date, or...
You may not like this, but I suggest another complicated dimension be applied like learning style. A concrete example of what I'm taking about is some people learn the quicksort best in prose / analogy, some learn it best in psuedocode, some learn the best in whatever language they are most comfortable in, and some would learn the best from graphical animation flowcharty type things. So its entirely possible the path to mastering the quicksort would be four different books for four different people. Of course you can "fix" that by simply having different paths to the same goal for different learning styles.
Even worse is people playing their own multidimensional games when mushed up against a multidimensional book. So is the little schemer series really a book about learning scheme or a book about learning programming or purely educational learning how to think? One book could quite easily end up in three totally different linked list / tree diagrams that you're proposing.
Purely from a startup business strategy idea, I suppose you have a guaranteed sale to some bookseller or another.
Referenced table on page six of the PDF.
I recommend the Dover Books on Mathematics series:
"Click here if you prefer a categorized directory of mathematics books."
My guess is that not all of the books here are free. Some might just be available online if you're willing to perpetrate copyright infringement.
a really good list for machine learning and the prereq linear algebra and prob/stats: http://www.reddit.com/r/MachineLearning/comments/1jeawf/mach...