Hi AthanasiusKircher, I wouldn't worry too much about kids learning bad physics from Minecraft. They probably aren't learning anything other than how to play Minecraft.
So, they're claiming that Minecraft is useful for achieving educational learning outcomes. Where's the evidence? What are the effect sizes, what are they for, and under what conditions? I've managed to find two papers that discuss effect sizes; one simply claims that proficiency at Minecraft is a good indicator of already developed spatial temporal reasoning, the other was for teaching spatial-temporal reasoning but for a single, simple instance, and under very specific conditions that are unlikely to be replicated in real-world classrooms. Of course, the media are going to continue to run with stories about how Minecraft are great for education without doing any fact-checking, just like they did with Fold-It, which although it was excellent for performing chemical manipulations that computers couldn't do for researchers, turned out not to teach anyone playing the game anything useful.
I foresee a generation of college graduates with Minecraft skills on their resumes.