The point is to understand things, but it is easy to get trapped in needing to prove the value of that understanding. Sometimes this takes decades! I am not aware of how having a deeper understanding of any system has not been a net gain for humans in the long run. So many things overlap (I mean all physics uses the same laws) that it probably shouldn't be surprising that there ends up being a relationship to how plates spin and electrons orbit. But finding that connection is far easier said than done and far more obvious post hoc than a priori.
So why not encourage these people to play? Because it's clear that "fuck around and find out" is a very successful model. It's hard to measure and noisy, but I'm not convinced it's more noisy than the current system we use.
Finding a fun way to do things does seem like it produces better solutions, though. I mean, if you dread reading about your solution, what do you expect from people who don’t even want to devote their life to it?