Earthquake observations contributed by human observers provide an invaluable source of information to investigate both historical and modern earthquakes. Commonly, the observers whose eyewitness accounts are available to scientists are a self‐selected minority of those who experience a given earthquake. As such these may not be representative of the overall population that experienced shaking from the event. Eyewitness accounts can contribute to modern science only if they are recorded in the first place and archived in an accessible repository. In this study, we explore the extent to which geopolitics and socioeconomic disparities can limit the number of earthquake observers whose observations can contribute to science. We first revisit a late nineteenth‐century earthquake in the central United States in 1882 that provides an illustrative example of an event that has been poorly characterized due to a reliance on English‐language archival materials. For modern earthquakes, we analyze data collected for recent earthquakes in California and India via the online “Did You Feel It?” (DYFI) system. In California, online data‐collection systems appear to be effective in gathering eyewitness accounts from a broad range of socioeconomic groups. In India, however, responses to the DYFI system reveal a strong bias toward responses from urban areas as opposed to rural settlements, as well a bias with literacy rate. The dissimilarity of our results from modern earthquakes in the United States and India provides a caution that, in some parts of the world, contributed felt reports can still potentially provide an unrepresentative view of earthquake effects, especially if online data collection systems are not designed to be broadly accessible. This limitation can in turn potentially shape our understanding of an earthquake’s impact and the characterization of seismic hazard.