This hard-won knowledge was the result of decades of observation by groups like the entire armed forces of the US on how to make systems that a 20 year old non-expert in a stressful situation can reliably use. It was the result of sticking people in front of computers and watching them make mistakes and figuring out what they intended to do and happen, so we could understand how people interpret an interface, and we used this research to make interfaces more discoverable and human-friendly. It lead to things like https://en.wikipedia.org/wiki/Fitts%27s_law . All this work was used heavily for example in Windows 95 to make an interface that grandma would be able to use. This research is why my mother was able to eagerly utilize computers for her work with minimal training.
But fuck all that I guess. Why base UI and UX on empirical study when you can just make bad faith A/B testing prove whatever you want and just make a red more magenta because your company makes a billion dollars per employee and has a captured market and hasn't had to actually build anything useful in decades.