This is why I'm so thankful of being a graphics coder. Because they try this stuff in coding as well. And you know for like web apps and stuff, they can kinda get away with it, kinda make it work. But graphics coding? Nah - faceplant every time. Trivially falsified. Go home.
That's what you're missing in history and theoretical physics. A nice quick ground truth test. Which - yeah - most fields don't have them. Art? Literature? Good music? Although, you know, for physics... hmmm...
Except theoretical physics doesn't exist in a vacuum... To even be considered remotely valid, any new theory must, at the bare minimum, produce results that are consistent with existing observational physics, either by matching existing results or by explaining previously unexplained observation.
We all agree on this.