By definition, the Plate Boundary Observatory (PBO) will investigate the mechanics associated with the boundary between the North America and Pacific plates. A natural frame of reference to describe crustal deformation between these two plates would be one in which the far field appears to be stationary, which leads to two possibilities: the "stable North America frame" or the "stable Pacific frame." The theory of plate tectonics pre-supposes the existence of such entities. Within such a framework the sum of deformations across contiguous regions of the plate boundary zone are constrained by the known relative far-field plate motion. However both theory and observation complicate this simplified framework. Are "stable plate interiors" a meaningful concept at the level of anticipated geodetic accuracy? Observational evidence suggests that most of the North America and Pacific Plates appear to be horizontally stable at the level of 1--2 mm/yr over plate-wide distance scales, which is not a very stringent constraint given current and anticipated station velocity accuracies at <1 mm/yr. Where do these stable plate interiors end, and the plate boundary zones begin? The Pacific-North America plate boundary zone is so broad that gravitational potential energy and mantle dynamics must play an important role - how deep into the plates do these dynamics have a measurable effect? How should we account for glacial isostatic adjustment in the definition of a plate-fixed frame? How should we define a "vertical datum" that is theoretically useful and can be realized in practice? In this presentation I review the current status of reference frame theory and actual practice, and I outline current thinking on the reference frame needs for PBO that derives from the working group on the "Stable North America Reference Frame" (SNARF), recently formed under the auspices of UNAVCO and the IAG sub-commission on the North America Reference Frame (NAREF).