Not yet reviewed by Pith; the record is open.

Advances in data acquisition and numerical wave simulation have improved tomographic imaging techniques and results, but non-experts may find it difficult to understand which model is best for their needs. This paper is intended for these users. We argue that our notion of best is influenced by the extent to which models satisfy our biases. We explain how the basic types of seismic waves see Earth structure, illustrate the essential strategy of seismic tomography, discuss advanced adaptations such as full-waveform inversion, and emphasize the artistic components of tomography. The compounding effect of a plethora of reasonable, yet subjective choices is a range of models that differ more than their individual uncertainty analyses may suggest. Perhaps counter-intuitively, we argue producing similar tomographic models should not be the goal of seismic tomography. Instead, we promote a Community Monte Carlo effort to assemble a range of dissimilar models based on different modeling approaches and subjective choices, but which explain the seismic data. This effort could serve as input for geodynamic inferences with meaningful seismic uncertainties.