High-quality, near-source seismic data are key to resolving whether differences exist between the physics of (1) large versus small earthquakes, (2) human-induced versus natural earthquakes, and (3) earthquake nucleation in laboratory experiments versus in natural faults. The need for near-source data has been widely recognized, as evidenced by the many recent projects designed to access fault zones at seismogenic depths. In 2005 the Natural Earthquake Laboratory in South African Mines (NELSAM) project was established in TauTona Mine, South Africa, to take advantage of the direct access and high rates of seismicity at depths approaching the seismogenic zone on natural faults. With data from the NELSAM project, my collaborators and I have shown that the magnitude-frequency distribution [Boettcher et al., 2009] and source mechanisms [Boettcher et al., 2015] of earthquakes in the magnitude range -2 < MW < 3 remain constant, such that tiny and moderate sized earthquakes appear to be controlled by the same source processes. We have also demonstrated that the characteristics of mining-induced earthquakes are similar to laboratory friction experiments and to earthquakes on the San Andreas Fault [McGarr et al., 2009a; 2009b; 2010; 2013]; that seismicity patterns correlate with in-situ geogas concentrations [Lippmann-Pipke et al., 2011]; and that physics-based models of fault friction and stress transfer developed for large earthquakes also apply to tiny mining-induced earthquakes (down to magnitude -3.4) [Kozłowska et al., 2015]. Our on-going projects in the South African mines include determining how source parameters vary with source type and proximity to faults. I have also been collaborating as a proponent to an International Continental Drilling Program (ICDP) project to drill into the rupture zone of recent M5.5 mining-induced earthquake, where the fault core was just recently brought to the surface. Pamela Moyer, Dr. Bill Ellsworth and I have inverted for the source parameters of the target earthquake and have encouraged others to do the same [Moyer et al., 2017]. We plan to compare multiple methods of seismic slip inversion with in-situ measurements of stress and observations of faulting obtained directly from the rupture zone. Constraining the source properties will allow us to better address fundamental questions in earthquake physics, such as how and why earthquake ruptures stop.