I studied organics chemistry during five years. All those years I just made reactions to made them, the only moments where I had to perform a synthesis on paper was during exams or exercises but all steps were given. So, even if, it was very interesting, there were no really fun to apply all this knowledge to think about the synthesis of a molecule. Obviouly with all these lessons I'm not that bad and I'm able to do a little bit of retrosynthesis.
However I know that total synthesis of natural origin molecules is perform a lot for some decades, we can find a lot of examples here, and sometimes it looks pretty weird to think that human people were able to think on how to make a synthesis in more than twenty steps.
For example, the weirdest synthesis I found, is the synthsis of Azadirachtin "only" perform in $79$ steps. That is abosolutely insane.
I'm wondering, for every molecules in general, how do chemist do to think about the retrosynthesis of such big molecules which contain in general a lot of stereocenters which then add more difficulties? I mean without to use a computer, is there some rules, such like theorems in mathematics, to think in the best possible way? Because with a bit of luck for $79$ steps at $90$% yield it makes a final yield of $0.024$%...