"Breakthroughs" come along regularly, but often don't hold up under scrutiny -- potentially undermining trust in science. My feature for Undark.