In the 18^th century, research in petrology was done by two approaches: field observation and experiment. Friction existed between the proponents of these approaches, given the long tradition of field work and skepticism surrounding experiments. As new laboratory techniques were developed and added to the observation-based petrologist’s repertoire, distrust of experiments persisted through the 19^th and early 20^th centuries. However, the field of petrology was changed irreversibly by advancements in experimental petrology in the 20^th century, with crucial new insights into phase equilibria that remain foundational to petrological research today. Observation and experiment remain central to modern petrology, but look very different than they did in the 18^th century. Although field work is still important, observation has evolved to include increasingly powerful microscopes effective down to the atomic scale and chemical analysis of a wide variety of isotopes and sub-μg/g levels of trace elements. Experiments can now be done virtually, and phase equilibrium modelling has become ubiquitous in petrologic studies. Today, novel observation-based advancements in petrology typically require data from expensive analytical instruments and can be time-consuming to collect and interpret. However, only a computer is required to conduct virtual experiments (models) with relative ease. Although both approaches contribute to high quality, high impact research, there is an inherent order in petrology: experiments must follow observations. Without grounding in the geological record, experiments—by necessity, simplified and idealized representations of nature—have little meaning. Modelling in particular can easily outpace observation-based work and laboratory experiments, despite these being crucial to reliable and useful model results. As it becomes easier than ever to conduct petrological research that is disconnected from primary observations, bringing mindfulness to modelling has never been more important.