But what WAS possible, was to boil everything down to very exclusive scenarios - like if you were 100% stationary and the mob you were fighting was absolutely stationary - and you knew exactly what buffs/debuffs came into play - and you could rely 100% on everyone else doing exactly what was required for the conclusion to be true - then yes, you could come up with a correct answer for that tiny and completely useless (in a pragmatic sense) scenario. That sort of answer would then be translated by the uninformed masses into some kind of universal truth - and a "cookie cutter build" was the result.
I don't know about WoW much, but it's not a bad example.
All you have to do to get a better model is to add one more variable to it. So add movement or buffs/debuffs to it.
Now you wouldn't get a model that worked 100% of the time but let's say 96% of the time. Now the question is if that's acceptable or not ?