Within the plane of a discrete fracture, gravity-driven fingers will travel faster and further than an equivalent stable infiltration front, raising interesting questions regarding prediction of flow and transport in unsaturated fractured rock. Here, we summarize and abstract current understanding of gravity-driven instability during infiltration into a dry, or partially saturated fracture. Recent literature, analogies taken from publications in other fields (petroleum, soil, thin film), and unpublished experiments that we have performed over the past several years are employed to help identify research needs, and the potential impact of unstable flow on predictive modeling. We then discuss the importance of specific research needs. We begin at the single fracture scale by considering gravity-driven instability with respect to: aperture heterogeneity, film flow, surface chemistry, precipitation/dissolution, and initial moisture content. We then conclude by considering phenomena that will influence unstable flows at the network scale, specifically, behavior at fracture intersections and matrix imbibition.