Ask an ordinary person what, say, an institution is and you are likely to get a fairly inarticulate answer, maybe a couple of examples of things that are institutions. Ask a social ontologist, and you are likely to get a considerably more articulate answer (although different ones depending on who you ask). Or take more concrete social phenomena, like money and baseball (favorite examples among some social ontologists) or race and gender (favorite examples among other social ontologists). Here the ordinary person might have more to say, but it would probably still be quite disorganized, while some social ontologists could give you sophisticated accounts of the nature and existence conditions of those phenomena.

At first sight, this is hardly surprising, and not necessarily problematic. It is difficult to see what social ontology should be about, if not to sharpen the contours of the social realm: organizing and systematizing our understanding of social phenomena. That there might be a discrepancy between how laypeople understand certain phenomena and how those same phenomena are explicated in developed ontological or scientific accounts is not automatically worrisome. Yet the social case might stand out as an exception here. After all, a very common idea in social ontology is that our ideas play a constitutive or causal role in making entities in the social realm come into existence.Footnote 1 If laypeople have indeterminate and disorganized ideas about various social phenomena, then these will accordingly go into the processes that either constitute those phenomena or cause them to come into existence, and presumably the results would then often be relatively messy rather than neat and precise.

In what follows here, the notion of social indeterminacy will be used to refer to the various ways in which social phenomena might be vague, fuzzy, unsettled, and unclear. The case that will be made in this paper is that we have reason to think that social indeterminacy is pervasive. There are then further questions to be raised about the more exact nature of this indeterminacy. In the general debate about vagueness and indeterminacy, a common view is that all cases of vagueness or indeterminacy are really semantic or perhaps epistemic in character, never ontic or metaphysical. For many, this is even ‘the only intelligible account’ (Lewis, 1986: 212). But if the ways in which we understand or interpret objects in the social world will not just be about how we decide to represent a pre-existing world, but instead play a constitutive or causal role in making entities in that realm come into existence, matters become more complicated. The relevant kind of social indeterminacy might perhaps still be acceptable even to skeptics about metaphysical indeterminacy, since the ultimate source of indeterminacy is not worldly.Footnote 2 But exactly what to say on such matters is something that will be set aside for now.

The simple argument provided above for why we should expect social indeterminacy to be pervasive is really very simple, while the sources of indeterminacy that eventually will be identified here will be much more specific. But the simple argument does still raise something of a puzzle: why is the phenomenon of social indeterminacy hardly discussed at all in the social ontology literature?Footnote 3 The working hypothesis for this paper is that this is due to presuppositions about how theorizing in social ontology is to be done, where certain complicating factors tend to be put to the side in order to facilitate a specific kind of theorizing. It then becomes important to critically discuss those presuppositions, not just because there is a question about whether we need to change how we do social ontology in order to account for social indeterminacy, but also because by identifying features that might contribute to making it less visible, we might get a path to making it more visible.

This paper has two main sections. The first one attempts a diagnosis of how certain tendencies in analytic social ontology might have contributed to obscuring the phenomenon of social indeterminacy. It will by necessity be a broad-brush picture, but hopefully it can still help in trying to explicate different possible sources of indeterminacy, as well as potentially facilitating a discussion about how social ontology can do justice to social indeterminacy. The second part will then discuss three possible sources of indeterminacy which have been identified on the basis of the analysis in the first part.

1 The creationist approach to social ontology

While different kinds of social theory have been around for a long time, social ontology as a distinct philosophical field is still relatively young. In an overview, Guala (2007: 961) highlights two influential works published in the 1990s, Searle (1995) and Hacking (1999),Footnote 4 as important in making social ontology come together as a field. One possible worry whenever a field of inquiry comes together is that which ideas that become central, and which ideas that become marginal or even left out, will largely be a contingent matter, depending on which particular works that just happen to become influential in shaping the debate. This is indeed a worry raised by Guala in his overview of what he calls the Standard Model of Social Ontology, and more recently by Burman (2023) with respect to what she identifies as a dominant branch of ideal social ontology.

One thing that arguably sets analytic social ontology apart, at least compared to analyses of social construction in critical theory and continental philosophy (which are often genealogical in character), is its ahistoricity. Of course, we all know that the ways in which the social world is structured are products of complex historical processes, where things have mainly changed through gradual modifications of what was already in place. And yet, many (analytic) social ontologists tend to abstract from such things, offering logical or rational reconstructions of how social facts or entities could come into place through a kind of ahistorical acts or events. We might call this type of social ontology creationist. It is an approach that is similar to contract theory in that it works by postulating a kind of pre-institutional original situation, where an institutional solution then comes to be created, and where the dynamics of that original situation is supposed to throw light on actual institutional arrangements. This methodology should not be dismissed out of hand. For instance, institutions are typically quite stable, and game-theoretical modeling can elegantly make sense of how there can be stability even without strong systems of enforcement. At the same time, the ahistoricity and level of idealization that is needed to make such modeling work might also come with significant costs, and it is important to consider these as well. The argument here will be that one such cost is to render social indeterminacy less visible.

If we look at the trajectory of analytic social ontology, the most influential agenda-setter is arguably Searle (1995, 2010). Focusing on institutional facts, his theory of how this part of social reality is integrated into the rest of reality is that certain social statuses are imposed on different material objects through collective acts of accepting that such statuses are in place. More specifically, the relevant social statuses can be understood in terms of constitutive rules that have the general form X counts as Y in context C. Unlike regulative rules, which regulate already existing activities, constitutive rules create the possibility of new types of behavior.Footnote 5 Just like rules of chess play a constitutive role in defining what the different pieces are and can do, e.g., that a certain physical movement is not just moving a piece of wood three inches to the left but also counts as making a check-mate, something like rules of payment can play a constitutive role in making it the case that certain transfers of pieces of metal or paper notes count as making a payment. What the relevant constitutive rules typically put into place is also a number of corresponding deontic powers, rights and duties, that can then script our interactions, e.g., ‘if I hand you this paper note, you are obligated to accept it as payment’. In a later work, Searle develops this part of his account and proposes the following formula (2010: 101–102):

We (or I) make it the case by declaration that a Y status function exists in C and in so doing we (or I) create a relation R between Y and a certain person or persons S, such that in virtue of SRY, S has the power to perform acts (of type) A.

The idea here is that even if institutions emerge over time and gradually, there is something like ‘the logical form of the creation of socially constructed reality’ (Searle, 1995: 191) and that the establishing of institutional facts can always be logically reconstructed in terms of declarative speech acts. Although not that many follow Searle in making this particular move, it still provides a good illustration of what is here called the creationist approach to social ontology: the idea that even if real-world institutions always have a history and are entangled with other institutions, we can (and should) understand their nature by reconstructing them in terms of the conditions under which they can arise at an ahistorical point of creation.

Although Searle is an especially clear example of a creationist approach, he is far from unique in understanding social ontology as a project of logical or rational reconstruction, and where that reconstruction takes the form of laying out certain ahistorical conditions of possibility. There are other authors, such as GilbertFootnote 6 and Tuomela,Footnote 7 who also propose accounts where institutions are understood in terms of reconstructive schema where collective intentionality plays an important role, but there is also another important type of creationism that can be found among authors who model institutions as equilibria in a game-theoretical sense. This approach goes back at least to Lewis (1969) and his work on understanding conventions, but the first major statement of this approach was by Schotter, who suggests (1981: 20) that we ‘start our analyses in a Lockean state of nature in which there are no social institutions at all, only agents, their preferences, and the technology they have at their disposal to transform inputs into outputs.’ This is an approach coming out of economics, with Aoki (2001) and Greif (2006) being two other important theorists, but in recent years it has also been introduced into social ontology by Guala and Hindriks, who have proposed what they call a rules-in-equilibrium account, which they have defended and elaborated on both jointly (Guala & Hindriks, 2015; Hindriks & Guala, 2021) and separately (Guala, 2016; Hindriks, 2019). Another recent example is Hédoin (2021), who departs somewhat from especially Guala and instead suggests a beliefs-rules-equilibrium account of institutions.

These equilibrium accounts do not feature joint acts of creation in the way envisaged by someone like Searle. Rather, their main strength lies in being able to show how institutions can arise through how individuals come to coordinate their behavior, even in the absence of being socially integrated with each other. But the situations that are studied are typically ones which cry out for the creation of an institutional solution.Footnote 8 These theorists are of course aware of how the real-world counterparts to the institutions that they study have usually evolved over time, but the accounts that they offer are decidedly ahistorical. As Hindriks & Guala (2021: 2029) put it, ‘[t]he point of departure of game-theoretic accounts of institutions is a game form that specifies the preferences that agents have and the actions or strategies that are open to them. Within a particular game, agents maximize their utility by taking into account how others are likely to behave.’ The social circumstances being considered in this type of theorizing are accordingly ones that we do not tend to face as real-world agents, because we are always already living our lives within given institutional frameworks, ones that are not only often overlapping and entangled with each other, but into which we have typically also been socialized from an early age, and where shared ideas and norms have already been integrated into our preferences.Footnote 9 Rather than positing ahistorical acts of creation these accounts accordingly posit something like ahistorical events of creation: a plurality of agents all adopting certain interlocking strategies given certain circumstances. Such accounts are clearly different from Searle-style ones, but the suggestion here is that they are still similar in how the kind of ahistorical analysis that is adopted abstracts from many features of actual social life, pushing potentially important complications to the side.

With respect to social indeterminacy, there are at least three interrelated features of creationist approaches that appear troubling. To begin with, when it comes to the relevant agents, these tend to be (basically because they have to be) very thinly characterized, either just as parts of a homogenous ‘we’ or as agents with certain clear-cut preferences and very limited sets of well-defined actions available to them. Typically, the agents under consideration are also all unambiguously part of the relevant situation and are aware of each other as being in that situation. Similar to work done in the contract-theory tradition, the relevant communities and situations tend to be modeled as closed systems, whereas in actual social life it might often be unclear which people that count as members of a certain community or group, and which people that are insiders or outsiders can also change over time. There is of course a reason for making such simplifications: they facilitate the relevant forms of theorizing. But this also means that issues about drawing the boundaries of communities or groups are being abstracted from, and a potential source of social indeterminacy is thus left to the side.

Additionally, because of how simplified the original situations get construed in creationist modeling, the kinds of collective acceptance, agreement, or forms of coordination and cooperation that come out of them tend to be complete. There is nothing left to determine (or at least: there being something left to determine is not part of the picture). This is not to say that these theorists actually think that institutions or social practices really are that simple. These models are perfectly compatible with there being various complexities that are not covered by them, and the idea is presumably that working at a high level of abstraction allows us to get to the core of things and neatly capture the central aspects of the social world. But again, a potential source of social indeterminacy is then being left to the side, in that actual social life can at times be very much characterized by partial or temporary agreements, by uncertainty about what to do in many situations, and by always ongoing processes of negotiation and bargaining, the outcomes of which might often be unclear.

Finally, when it comes to understanding the objects of agreement or acceptance, there is a strong tendency to conceptualize these in terms of rules. Again, this is a feature that makes sense in the light of the other ones: given that we are reasoning in terms of complete agreements or convergences, then why should these not take the form of relatively clear-cut rules? And while rules-in-equilibrium theorists rely on a technical notion of games, many authors in the literature tend to rely on much more concrete analogies with games in developing and explaining their accounts, with two especially prominent examples being chess and baseball. Already when Rawls (1955: 16) first articulated his influential account of social practices, chess and baseball were his two main examples. Someone like Searle points to chess (1955: 27–28) and football (1955: 66–69), but also baseball (2010: 103). North suggests that the rules that make up institutions ‘are perfectly analogous to the rules of the game in a competitive team sport’ (1990: 4). Indeed, certain examples are so ingrained in the literature that when discussing the idea of constitutive rules, Guala (2016: 69) sees the need for admitting that he personally has ‘no idea how the game of baseball works.’ Again, a worry here is that an important source of social indeterminacy is being left to the side with this heavy focus on rules, especially when the relevant rules are understood on analogy with the kind of carefully crafted rule sets that we find in games like chess and baseball.

The argument here is not that creationism is a tightly defined research program. The term is merely intended to refer to something like a family of theorists and theories, where there are certainly a lot of differences between them, but where the resemblances can still be worth bringing out in order to try to make sense of why social indeterminacy has been such a neglected topic. It should also be recognized that in recent years, many social ontologists have started to problematize the standard ways of doing things. There are theorists, like Epstein (2015); Haslanger (2012, 2018a,), and Jenkins (2023), who all very much emphasize the complexity and messiness of the social realm, and Burman (2023) problematizes the kind of idealized theorizing found in Searle, Tuomela, and Gilbert. However, precisely because creationism is not a tightly defined program, but rather a set of more loosely interconnected ideas, it still seems to be able to exert a kind of gravitational pull, where even authors who are exploring other ways of doing social ontology are sometimes drawn into its orbit. For instance, while someone like Ásta clearly seeks to understand more subtle aspects of social life, strongly emphasizing the role of negotiation in context and developing a distinct account of communal social statuses, she still ends up using baseball (2018: 10–13) and chess (2018: 118–119) as guiding examples and her account of institutional statuses still has many similarities to the one proposed by Searle. And when a metaphysician like Schaffer (2019: 749 − 50) ventures into social ontology, seeking to understand the grounding of social facts, he too ends up emphasizing the role of rules (and using chess as a central example):

[S]ocial facts can be built through two factors—which I label ‘rules’ and ‘moves’—operating together. There are background social rules which determine what counts as what, just as the background rules of chess determine what counts as a checkmate. Then there are foreground social moves that thereby count as social outcomes, just as a particular movement of a plastic figure may count as a checkmate.

To be sure, partly influenced by (but also disagreeing with) Epstein,Footnote 10 Schaffer works with a much more complex picture of how social facts are grounded than what can be found in Searle. There are no acts of creation involved. But at the same time, there still remains something that kind of looks like the output of such acts of creation, and in setting his task, Schaffer still maintains that ‘the question is how best to understand the metaphysics behind these “rules-and‐moves” operations, by which something “counts as” a social entity’ (Schaffer, 2019: 750). One worry here is accordingly that especially Searle’s prototypical creationism is casting a long shadow in the sense that even authors who seek to explore other kinds of theorizing still get stuck in a problematic path-dependence, where by engaging with previous theorizing, they end up affirming some of its presuppositions.

Of course, unless we just want to say that the social realm simply is too messy for philosophical theorizing, we will need to engage in at least some abstraction or idealization. We are accordingly faced with having to strike a balance between respecting the phenomena, messy as they are, and sharpening the contours of the social realm enough so that we can (hopefully) capture its core characteristics and dynamics, thereby increasing our understanding of the social realm as a whole. The argument in the present paper is that this balance should be struck differently than what is often the case in analytic social ontology: that what is here called creationism involves sharpening the contours of the social realm in ways that render a central characteristic of it, social indeterminacy, much less visible than it should be.

2 Sources of indeterminacy

Based in the analysis above, we will now proceed to look into three possible sources of social indeterminacy, ones that might give us reason to reconsider how we as social ontologists theorize the social realm. These sources will often be co-present and co-reinforcing in actual practice, but at least analytically they can still be distinguished from each other and we will discuss them one by one. The first two both concern how certain statuses might not be fully settled (either in terms of whether something has a certain status or what is involved in having that status), while the third is about how even what we agree or converge on might be indeterminate in character. I will provide examples for all three, but these are mainly intended as illustrations. Nothing hinges on the particular details of these examples, but hopefully they at least illustrate how these sources of indeterminacy can work. Neither is the point merely that one can find some examples of social indeterminacy, but that it is a pervasive phenomenon. Things in the social realm might often be determinate enough for indeterminacy not to cause any problems or tensions – but determinate enough is still a form of indeterminacy.

(i) Untidy communities. Part of the creationist approach is the idea that there is a ‘we’ or at least a specific set of participants accepting certain rules. Some social ontologists offer detailed analyses of how things like collective intentions can be built up from individual attitudes, whereas game-theoretically oriented theorists need not assume that any form of strong collective acceptance is in place. But these various approaches all tend to work with toy examples, abstracting from how real-life communities are almost always much messier, with unclear boundaries, possibly divided or disunited in many ways, as well as overlapping and crisscrossing with each other – characteristics which idealization typically removes from the picture.

One example here is how, even at a given point of time, there are often both insiders and outsiders to practices and institutions, and that both groups contribute to making the relevant social entities into what they are. Take churches as an example. The concept of church is arguably vague, with unclear boundaries to sect or cult, and possibly an unclear boundary to business as well. Part of something being a church is about what goes on within it, part involves relations to the outside world. What happens within them is largely the product of insiders, such as clergy and worshippers, but their place in the overall social landscape is also very much determined by different outsiders: governments, peer organizations, as well as lay people. These different groups or entities can very well operate on different conceptions of what a church is and which entities that are churches. In some cases, such differences might not matter, at least not for the question of whether some X belongs to a kind Y. The Catholic Church is very clearly a church. But is Scientology a church? Scientologists would say so. But some governments regard it as a sect or a cult. Other people might say that it is a business. If we look at the general question of whether Scientology is a church, then given the divergence in beliefs in different groups it seems reasonable to regard it as having an indeterminate status, as being a borderline case. Of course, if everyone shifted to the same view on whether Scientology is a church, the indeterminacy could be removed (or at least become a marginal matter). But this will be the case for all forms of social indeterminacy: if we sharpen the contours of the relevant notions or agreements, the contours of the relevant entities will be sharpened as well. The reverse process can however also happen: we start with agreement about well-defined ideas, constituting fairly determinate social entities, and over time they become increasingly fuzzy around the edges.

Another untidiness typically removed in creationist idealizations involves the temporal dimension. While many real-world social entities might have founding moments, with an exact date where they come into existence, they then still evolve over time and might differ between different times. Take complex institutions or organizations like universities.Footnote 11 Already at a given point of time, even faculty and students will have fuzzy ideas about the university where they work or study. No-one ever interacts with the university as a whole, and there might be many activities where it is unclear if they take place within the university or not. It might also be unclear where any specific university is located more precisely.Footnote 12 But if we look at universities historically, things become even more complicated. The university is originally a medieval institution which has then evolved and grown more and more complex. For instance, an entity like the University of Cambridge has existed since 1209 and gone through very many changes. On a contemporary understanding of a university, the institutions in place back then would hardly even count as universities. There will not be a determinate understanding of what a specific university is that will have been shared across time by all the different groups who have related to this entity, but rather partly overlapping ideas. And yet all of the activities and ideas of these communities are arguably part of what makes something like the University of Cambridge into what it is.Footnote 13 Now, any fuller account of the temporal dimension of entities would inevitably have to consider questions about three-dimensionalism vs. four-dimensionalism (cf. Sider, 1997), but to the extent that the ideas and practices of communities at different times together make a university into what it, it would presumably have to be an entity whose nature is quite fuzzy in character.

One possible response here might be that what these cases highlight is not so much indeterminacy as the importance of context-specificity. Perhaps there is no determinate answer to the completely general question about whether Scientology is a church or what a university is, but within more particular contexts it might still be clear what counts as what, especially if we relativize to more particular groups. This would be similar to how in different jurisdictions, the same basic act can be allowed or forbidden. We should not ask general questions about the social status of different things, but should always specify the relevant contexts. Up to a point this is quite reasonable – context is certainly very important. But emphasizing context-specificity does not automatically mean that we end up with determinate statuses, because the boundaries between contexts can also be untidy. When it comes to laws, there are typically relatively clear jurisdictions, but for many other social phenomena the difficulty of demarcating relevant contexts is part of the problem of untidiness. Indeed, recognizing the potentially messy character of contexts is important for understanding how the statuses of certain things can be contested and unclear.

(ii) Incompletely worked-through agreements. Many creationists share with social-contract theorists the methodological idea that if you show how institutions can arise even under conditions where people are self-interested and there is no prior social basis for the unions in question, you can throw light on real-world counterparts to those institutions.Footnote 14 The tendency is often then to see any institution as an isolated response to some One Big Problem. There is however another possible starting-point: that what we face in our daily lives is rather an endless stream of many small problems, and that the solutions that we develop in order to handle these are more like a patchwork quilt, where new patches are added to old patches, partly by different people, where already existing institutions might take on new roles when new needs arise, and where different institutions are always entangled with each other and constrain which new additions that are feasible.

This kind of patchwork conception of institutions can be reasonable even for highly formalized institution like the law. A leading example here is Sunstein’s (1995) account of law in terms of incompletely theorized agreements. Sunstein takes issue with the faith that many legal scholars, and surely many philosophers as well, have in the idea that when we are in disagreement about something, we move upwards in abstraction in order to find agreement on some more general principle or theory, and then work out its implications for the issue we started in. Another way of working however, and for Sunstein this is both descriptive and at least weakly prescriptive, is to move to a level of greater particularity and find a shared solution there instead. The scope of the agreement will instead be relatively narrow. We will typically have different reasons for finding it acceptable, and many issues that lie in the vicinity of the one that we agreed on will be left unresolved, to be dealt with at a later date (maybe).Footnote 15

For other institutions than law, it is not clear that theorization is what is at stake, so in generalizing Sunstein’s account we can perhaps talk about incompletely worked-through agreements instead, where the defining feature is that we are in agreement about certain core cases, but that then there will also be many unsettled cases. These will be ones where our previous agreements, sometimes made by previous generations, function as precedents and typically have some bearing, but where deciding on the status of these unsettled cases is not just a matter of working out the implications of principles or rules we already accept – there will instead be a need for renewed negotiations. This kind of incompleteness should be expected to be prevalent in informal institutions and practices, since even while there is negotiation going on there as well, communicating with far-off others will be highly indirect, moving via complex chains of interaction. Shared clear-cut standpoints might then be better thought of as a kind of waypoints by reference to which we can navigate and negotiate our social relations with concrete others. A simple example here is how borders between communities mostly functioned historically. Communities A and B might agree on, say, the fields on one side of the forest belonging to A and the fields on the other side belonging to B, but have no agreement on the forest itself, although they agree that going really close to the fields of the other is to overstep the boundary. The fields constitute relatively clear-cut waypoints, but the status of much of the forest is unclear and the geographical boundary between the two communities is indeterminate.Footnote 16

Or take what is arguably the favorite example in social ontology: money. As Frasser & Guzmán (2020: 25) have put it, ‘Although we do not usually ask ourselves what money is, we are all competent in the practice of using banknotes, coins, checks, etc. Despite the ease with which we master the practice of using it, explaining money has proved to be an arduous task.’ In much of everyday life, we do not have to worry about exactly what money is, we can have many unclear or indeterminate ideas about the nature of money and still be in agreement about how to conduct certain commonplace transactions. But if our agreement is not fully worked-through, then one cannot assume that there is a latent set of necessary and jointly sufficient conditions that can be extracted from our ongoing monetary practices. We might certainly opt for a more determinate conception depending on some specific purpose,Footnote 17 but the general nature of money could still be indeterminate.

A consequence of this kind of indeterminacy is that when a novel form of payment appears it is quite possible that it is unclear whether it should be counted as money or not. A recent example is bitcoin, and similar cryptocurrencies (Smit et al., 2016; Passinsky, 2020). These share some features with what we typically regard as money, such as being used as means of payment, but on the other hand they typically do not have the status of legal tender, and they are not issued under a central authority, like the currencies that we standardly regard as clear-cut examples of money. Bitcoin accordingly constitutes a borderline case, and arguably does so because our shared understanding of money is incompletely theorized or worked-through. We then have reason to think that money is an indeterminate kind.

This is not to say that we cannot sharpen our understanding of money so that bitcoin comes to clearly count as money (or not). Indeed, this is par for the course when it comes to incompletely worked-through agreements: we deal with indeterminate statuses when they become pressing enough so that a decision needs to be made: are they Y or not? We make things up as we go along. Again, note that the idea here is not that there are these rare occasional indeterminacies, but rather that indeterminacy is pervasive in the social realm, and that even some of the things (like money) that we might think are determinate are really better understood as being determinate enough for current practices – until something changes and the indeterminacies that were already there become highlighted. If we take the physical world, it can be fully determinate even if we have not fully worked out our conception of it. But for the social realm, given that our notions and beliefs play a central role in making it come into existence and shaping it into what it is, there is reason to think that there are substantial limits to how determinate it can be. Different parts of it might certainly be more or less determinate, but that is just to say that we are clearly in the realm of degrees of determinacy or indeterminacy.

Things are further complicated by (as argued above) there often being many different communities and organizations involved in social matters, making it even more challenging to come to determinate agreements or to converge on clear-cut patterns of behavior. If we take the case of money, then for some things that count as money there might be more or less global consensus, but it is quite possible that something like bitcoin might come to count as money in some jurisdictions, but not in others, or for some purposes, but not for others. As argued by Passinsky (2020: 288 − 90) the decision about whether or not to count something like bitcoin as money is a normative choice. It is a choice about moving towards a conception of money that is more determinate in certain aspects, but it is ultimately a choice that will have to be made on pragmatic grounds, such as how to best deal with things like tax evasion and money laundering. It is not about deciding once and for all what money really is, but merely about making some specific precisifications in the face of a pressing need for greater determinacy in certain respects. And different communities might very well have different pragmatic reasons for taking this or that approach.

(iii) Similarity-based categorizations. As pointed out above, one feature of the creationist approach is an emphasis on rules. Especially if we understand social rules by analogy to rules in various competitive games, then these are typically quite well-defined and would ideally sort cases as either falling under the rule or not. Different moves that we make within the context of our institutions and social practices would accordingly fall into relatively determinate categories. Of course, as long as we are dealing with rules expressed in natural language, there will be always be some indeterminacy involved. But at least for relatively well-defined rules this might arguably be seen as a marginal phenomenon and perhaps just the kind of thing that we can reasonably abstract from in order to facilitate more developed theorizing. However, it is far from clear that the analogy with competitive games, with their well-defined rule sets, is representative of what the social realm largely looks like. Especially when it comes to many informal social structures, these could be interpreted as often involving similarity-based categorizations instead, where different individuals might occupy informal social roles in different degrees rather than in an either-or way. When one’s belonging to certain social categories is set up by practices of categorizing by degrees of similarity to relevant standards or conceptions, such as exemplars or prototypes,Footnote 18 there being borderline cases and gray areas will be important characteristics.

One example here is the role of informal hierarchies in social life. In a formalized game like baseball, an umpire is an authority, and the rules of baseball describe who counts as umpire and what the umpire can authoritatively do. But in other parts of the social world there are people who are recognized as authority figures in their social setting, but where this can instead be understood in terms of how they resemble prototypes about what an authority figure typically is like.Footnote 19 For example, on the question of whether a person is to be deferred to on certain matters there might be a range of features that serve as partial cues for which response is felt to be appropriate: gender, skin color, age, height, body shape, hairstyle, clothing style, vocal timbre, dialect, sociolect, and so on. Some of these might have more weight than others, but whether a certain response is seen as appropriate or not will be a function of how many of the relevant cues are present (and to which degree, given that some of them might themselves come in degrees). In some societies it might be the case that the prototypical authority figure is a tall white man, middle-aged, not working-class, and with a deep voice. The extent to which one comes to count as an informal authority figure in such a society will then be a matter of degree, depending on how closely one matches the prototype (Brännmark, 2021: 239 − 40).

If social kinds are at least partly constituted by how we categorize things, typicality effects and borderline cases will be endemic to them.Footnote 20 Take contested social kinds such as gender and race. These are categories which also exemplify the issues of untidy communities and incompletely worked-through agreements. Not only is it difficult to clearly identify the relevant communities in which statuses of some individual counting as belonging to a certain race or gender are grounded, these statuses might partly be legal statuses, which might be more clearly defined, but partly also be about informal and situational social statuses, which to a large extent turn around how people categorize people that they face and interact with. These are also categories where there are disputes about whether they are biological or social, or whether there might be distinctions between different senses of them, which can then be in play at the same time in different contexts. How they have been understood has also varied over time. These are quite clearly messy categories.Footnote 21 There is no room here for considering all the relevant dimensions of such kinds, but one aspect which also contributes to this situation is that, if we look at our everyday practices of categorization, they seem to involve typicality effects and borderline cases.

Let us look at a specific example, namely the phenomenon of passing, where situationally a person who is typically or often categorized as belonging to one race, e.g., being black, is categorized as belonging to another category, e.g., being white. As pointed out by Mallon (2004: 648), ‘[p]assing is problematic for constructionists, since it seems to involve a person objectively belonging to one race while being believed to belong to another.’ However, the very way that Mallon states this challenge for constructionists about race assumes that there are clear-cut belongings in play, one objective and one situational. But if we accept that the folk concepts involved in constructing certain social categories can be messy and vague, then this need to be the case. If situational racial categorizations in a community largely operate on prototypical images of what members of different races look like, e.g., in terms of skin color, eye color, hair, facial features, etc., then some individuals will be in closer vicinity to those prototypical images, and will more or less always be categorized in a certain way, others will have more racially ambiguous features, and depending on the situation might be classified either way. Visual appearance will not be all that counts,Footnote 22 but as long as it plays an important role in our everyday practices of categorization, and hence in setting up race as a social category, there is reason to think that some people can have an indeterminate racial belonging that often gets worked out situationally.

There are certainly other ways in which passing can be accounted for. An example here is Ásta who proposes an account where some social statuses are situationally conferred through the actual beliefs that other people have with respect to us, but where they at the same time track certain base properties. Situational category belongings can then be shifty. Her account represents an important step away from the creationist approach,Footnote 23 clearly recognizing that there are many social matters that get negotiated in context: ‘what roles there are to play, who plays what role, and what the expectations are of each role’ (2018: 121). On her account, someone might then have the base property, and yet this not be recognized situationally. One needs to distinguish, however, between other people believing that one belongs to a certain race, and people simply not forming any race-related beliefs. In many settings, a certain racial belonging will be the default. When a person passes as belonging to that default, then this might not be because this has been negotiated in context but rather in the absence of negotiation. The experience of the person passing would then rather be of occupying uncertain ground (e.g., Piper, 1992). There is something tentative to the status. Passing as, say, white in a situation might then be better conceptualized not as simply being white in that situation, but rather as occupying an indeterminate position.Footnote 24

Of course, while the position that one occupies in any concrete social space will largely be determined by others and how they relate to one’s being there, one’s own sense of identity can be much more determinate. People can have a clear sense of who they are even if the social status that they tend to be accorded is more indeterminate or ambiguous, or even runs counter to how they understand themselves.Footnote 25 Issues about how an individual’s sense of identity relates to social category belonging are however too complex to address here, where the main point is simply to stress the need for recognizing the possibility that the latter can be indeterminate.

3 Concluding remarks

A central objective of this paper has been to make social indeterminacy more visible, partly by trying to identify tendencies in much of analytic social ontology that might contribute to making it less visible, partly by discussing a series of examples where the presence of indeterminacy in the social realm is highlighted. The argument here is not that creationism prevents us from accepting that there is social indeterminacy, but that it risks making it seem like a marginal phenomenon. Of course, choosing a theoretical approach always involves a cost-benefit analysis, where different approaches will have different strengths and weaknesses. Adopting a non-creationist approach to social ontology might make it easier to highlight social indeterminacy as a phenomenon, but it is certainly possible that such an approach would ultimately score lower in terms of theoretical virtues like simplicity and unification. How such balances should best be struck is a matter that lies well beyond the scope of this paper. But the extent to which an account can deal with social indeterminacy should be a factor in deciding how to best theorize the social realm, and the worry that has been raised here is that if we labor under certain standard simplifying assumptions, it will be more difficult to do justice to social indeterminacy.

In recent years, a number of social ontologists have been arguing for a shift towards nonideal theorizing in social ontology. Burman (2023) has provided the most developed argument for this move, but authors like Haslanger, 2018b); Ásta (2018); Brännmark (2019); Jenkins (2020) have also been taking steps in that direction. This is a type of theorizing that seeks to respect the messy, contested character of the social realm, and one feature often characterizing nonideal social ontology is an emphasis on how communities are not homogenous or merely differentiated into just a couple of neatly defined sub-communities. Instead, social positions and identities are complex and have interconnections that are infused by power differences and hierarchical structures. This feature of nonideal social ontology is certainly in line with the argument in the present paper, although it should be recognized that much of nonideal social ontology also tends to intersect with political theory, seeking to address important structural injustices. This might very well be a reasonable direction to take, but the worry about creationism that has been discussed in this paper has simply been about its descriptive adequacy, about how it pushes to the side what is arguably a central characteristic of the social realm.

Additionally, as already pointed out, the picture painted of creationism is by necessity a broad-brush one. Creationism is not an explicit research program, so even to the extent that it is fair to describe much of mainstream social ontology as creationist in character, what we have is rather something that serves as a further illustration of how categories are often based in family resemblances and similarities rather than strict membership criteria. While someone like Searle can arguably be taken as a prime exemplar of creationism, other authors or accounts might exhibit considerably lighter creationist tendencies. This also means that even if matters of social indeterminacy have largely been ignored, it is quite possible that some of these theories can be more easily modified than others in order to account for indeterminacy. But exploring such possibilities would seem to be best left to the adherents of those theories. Hopefully, the present paper has however at least made the case for the need to take social indeterminacy seriously, as well as having taken a few first steps in accounting for it by identifying some of its sources.