Social ontology is ontology. This might seem too much a truism to be worth stating, but its consequences are far-reaching. On the one hand, its methodology is completely on a par with other fields of ontology, like the ontology of abstract objects, midsize objects, the mind, etc. The consensual methodology in these fields is to postulate ontology as the best explanation for some successful human practice, most commonly science (Barceló forthcoming). It would be hard to make sense of the continuous success of science in providing knowledge, explanations, “predictions, and guidance of our actions and decisions” (Potochnik 2022) if they were not getting at something real. Thus, what ontology must we adopt depends on what ontology makes the best contribution to the explanation of this success. But, of course, science is not our only successful practice and thus not all ontology ought to be naturalistic in the sense of relying only in science as a reliable guide to reality. Furthermore, ‘success’ can be understood both as descriptively as a characteristic that some of our practices already show and normatively as something our practices ought to aim at, hence the distinction between descriptive and revisionary ontologies. A revisionary ontology, hence, would be one that would explain not why an actual practice is currently successful, but one that would contribute to the future success of such a practice. Again, this is as much true in science that in other fields.
Social ontology ought to recognize the heterogeneity of practices that include human classifications (in medicine, social science, politics, everyday conversation, etc.), and determine, on the one hand, which ones of them are already successful, and what makes them so; and on the other, which ones are not successful and what would their success would look like. This would gives us the prime matter from which to build a social ontology. Only then can we answer questions like which ones of these categories latch onto real distinctions and which ones do not, who inhabits them (what are their extensions), how do we think about them (what are their intensions), how can we know this (what are their epistemologies), and how does someone inhabit them, etc.
Thus characterized, social ontology is a genuine ontological enterprise and not a semantic one. Our job is not to determine what it really means to be poor, or mestizo, or bisexual, or any other predicate we currently use to talk about others and ourselves. It is not directly concerned with words or already-existing concepts and distinctions, but with the independent question of what distinctions we need to make (and which we need to avoid) in order to bring about justice. These distinctions might be similar or close to the ones we already use, but this might not be so. Of course, words and their meanings are part of the social world we are interested in, but they play no more central role than other factors that also constitute the social world. Therefore, it is important to distinguish between this ontological project from semantic projects close by.