When writing a compiler/linker, when I need to know what order to put things in, I perform a topological sort of the dependency graph.
When writing a complex talk, when I need to know what order to put things in, perhaps I should perform a topological sort of a dependency graph...!
I'm writing a dot file right now to graph the relationships between all the topics for my talk to figure out how I should organize and eliminate them. Not what I expected to be doing. And unfortunately I can't just take the first topological sort and call it a day. The resulting talk wouldn't be very good.
Step one should be get back to work instead of using GNU Social as a procrastination tool.
@catonano certainly---the topological sort part was a joke (the graph has cycles, and an algorithm to choose the "correct" sort would be the human creativity).
But the graph on its own is very useful in mapping out thoughts and discovering relationships that might be suited for certain sections of the talk. Dense portions of the graph might benefit from reduction (too granular---would take up too much time). Cycles show chicken-and-egg situations that are always fun to introduce in talks/presentations. Nodes with the most edges may demonstrate core concepts, or it may demonstrate that too much research / focus is being placed on something that it shouldn't be (isn't a core topic). Etc.
Other people call them "mind maps" (though I haven't researched the formalities of mind maps, so I can't really say much about them.)