the main reason is because the real world does not really work well with theory, so you end up significantly adding to your already rough calculations to make something reliable.
Even if you do advanced mathematical analysis on electrical circuits, once you include mechanical and thermal effects on a circuit, and the effects they have on the electronics, and the effects those effects have on the thermal and electrical effects and so on (to get something exactly) it becomes extremely unweidly to build something more then a single resistor standard (and thats just a resistor). Then factor in a HUMAN and you find out the biggest problem is the buttons are too weak because someone is slamming on it like balboa
So 1) you simply the math by getting rid of all the 'non-affects (they really effect things but its just sub PPM and there is so many variables it essentially looks like its random, another factor is dominating, etc.. its basically hunting a fart in a tornado. and its all time consuming and expensive to measure. oh yea and simulators go out the window, you will need to make your own simulator algorithm that might link like 3 different simulation programs together to form a experimental guideline, that you need to test, that there wont be cheap equipment or trustworthy equipment or equipment with bandwidth required or stability required etc)
2) you make the circuit tougher then it needs to be to circumvent a potential defect
3) you know 1) and 2) so you save yourself the time and just aim high with a guess, if it fails you use some math to aim a bit higher.
4) worry if pursuing this relentlessly will make you feel accomplished in life, if it will help anyone, etc. is it just a psychotic obsession at this point? is this what you want? did you play any videogames, smell some flowers, watch some TV, bake a cake recently?
If you do high-order analysis of circuits and parts and stuff you will get insight but for many things that are being sold, its not difficult to figure out how to make it better (but it wont fit a PRICE or development time model that's considered realistic). perhaps you over heard the board room talk about baseball going on for 4 hours and you realized... i should take it easy rather then crush my brain for these guys... lets take it easy on youtube for a while.. is my hair turning grey??
Even with the art of electronics, if you studied something like Sedra Smith in school, then you read AOE after (since its not part of corriculums) you will likely want to go back to the 'classic hard' literature to follow up on something, then that might make you go back to the AOE and back and forth and back and forth... you kind of just get interesting ideas you want to follow up on