<p>The study of glasses and glass formation is a rich and technologically important area of research. This Perspective reviews recent progress in the field from the different points of view of chemistry and physics. At the heart of glass science are the super-Arrhenius temperature dependence of molecular relaxation processes and the nonexponential nature of relaxation functions. We discuss three questions motivated by these features and describe the progress of the past decade in answering them. The paper reviews experiments and simulations that probe growing and possibly diverging length scales associated with supercooling, the possibility of a thermodynamic transition to an ideal glass state underlying the glass transition, and new indications for universal relaxation functions describing molecular motion in deeply supercooled liquids. For each of these still largely unsolved problems, we illuminate the interplay between the universality often favoured by physicists and the specificity typical of the chemist approach.</p>