I find the most valuable point is the last: whereas, sure, you can go on wikipedia and find dependencies of an area of knowledge, that three of prerequisites makes more sense to be presented as a flow converging on a point, rather than having to dig deeper until you find a node you already understand.
A suggestion for your product: build these dependency graphs automatically, and let users mark subjects as already satisfied. That way you can omit digging into dependencies that are satisfied. If you really want to be righteous, recursively mark the dependencies of the satisfied dependency as also satisfied. That way a user can pick a topic and find everything that still remains to learn to get to that point.
And you can have the best of both worlds. Generating a linear progression of things to learn becomes a simple matter of printing the dependency tree in postorder.
A suggestion for your product: build these dependency graphs automatically, and let users mark subjects as already satisfied. That way you can omit digging into dependencies that are satisfied. If you really want to be righteous, recursively mark the dependencies of the satisfied dependency as also satisfied. That way a user can pick a topic and find everything that still remains to learn to get to that point.
And you can have the best of both worlds. Generating a linear progression of things to learn becomes a simple matter of printing the dependency tree in postorder.