Thought experiments, while useless for direct science, are an excellent tool for mental processing. It can often lead to falsifiable experiments, that helps tease out the nature of reality.

The amount of good science that comes out of stupid "what if" type games/discussions/thought experiments would unnerve many people. The catch is that it needs to be backed up by old fashioned slog work.

As for the many worlds Vs Copenhagen interpretation, in most ways they are impossible to separate, they look at the same data, and create the same conclusion. There are still cracks that can be pried at, however. Most will lead to nothing, a few can help understand QM better, and find its flaws. Ultimately, however, the maths and measurements win. Any understanding method must conform to those. The model just helps envisage future paths.