Studying which prompts always fail could give us better insights into "mechanistic interpretability", or possibly lead to insights in how to train better, that

Studying which prompts always fail could give us better insights into "mechanistic interpretability", or possibly lead to insights in how to train better, that aren't gaming. Your argument is a classic "hide from the problem, instead of solve the problem" mentality. So no, please don't. Face your problems, and solve them.