1. Basic Realism

You and a friend are on a hike. At the crest of a hill, you pass a ponderosa pine. You reach down for a pine cone that stands out to you, but your friend stops you. ‘Maybe we should leave those things around the tree alone,’ she says.

In a perfectly good sense, the things around the tree are real. This isn't just an unsupported intuitive judgment. The things around the tree satisfy two basic criteria for reality: they exist, and they do so mind-independently (e.g., Brock and Mares Reference Brock and Mares2007). However, the metaphysical status of things around a tree is not obviously robust. In particular, things around a tree is not a natural kind, on dominant operationalizations of the term ‘natural kind’. There are not, for example, many properties that things around trees as such have in common and that set them apart from all the other things in the world (Mill Reference Mill1959: book I, chapter 7; book IV, chapters 6 and 7). Nor do they have some ‘theoretically important property’ (Dupré Reference Dupré1981: 68) or explanatory value (LaPorte Reference LaPorte2004: 19).

Of course, ‘natural kind’ is a term of art, and not everyone understands natural kinds as kinds that have many distinctive properties or that bear some theoretically important property. But even if we take natural kinds to be kinds that simply play any indispensable role in the natural or social sciences, things around a tree is not a natural kind in this sense, either. Most basically, having the property of being a thing around a tree does not facilitate induction or generalization, the sine qua non of science. And specifically, things around trees do not as such constitute a biological kind. There is no biological property that they have as such that sets them off from other things; the category things around a tree is arbitrary as far as the biological facts of the world are concerned. Nor is things around a tree a social kind. On one tradition, some people wrap up presents and put them around trees that they temporarily put inside their houses. Those gifts are given a special social status, so arguably Christmas presents or things around the Christmas tree counts as a real social kind. But the things around that ponderosa pine do not occupy the category of Christmas presents or any other social kind. We don't make them or confer upon them any ontological status, including their being things around the tree. And they could perfectly well be sustained even if all humanity vanished or never existed, a sure sign that they are not a social kind.

In short, things around trees do not jointly constitute a kind that is useful for or presupposed by any of the sciences. Things around a tree is not what we will call a scientifically relevant kind. But again, there is still pressure to say that things around a tree is nevertheless a real kind. There are other things around other trees, so things around a tree is a general category. Moreover, the things around the tree appear to be real, not just as individuals, but as things around the tree—again, things around a tree exist mind-independently as things around a tree. When we say that the things around the tree are real in this way—real in the sense of existing mind-independently as such—we maintain that ‘real’ operates as an ordinary-language term. These points suggest that there is a perfectly ordinary sense of ‘real’ in which kinds can be real even when they are not natural or scientifically relevant kinds.

At this point it is worth drawing some distinctions. Call a group a minimal group if its essential properties have no direct scientific relevance. There are at least two kinds of minimal groups. The most minimal groups are mere sets, defined exactly by their membership. A more substantial but still minimal group has its membership determined by some property that goes beyond trivial group membership but is not directly scientifically relevant: its membership is fixed by a scientifically irrelevant similarity. Proximity to the pine tree, for example, is the respect in which the rocks, stick, and pinecone around that tree are members of the minimal group things around a tree. We will refer to these two minimal groups as follows. A kind is a basic kind if its members are united merely by sharing a similarity, that is, if they have something in common that is nontrivial but also not directly relevant to science. A group is an unkind set if its members have nothing in common other than, trivially, being members of that set. Quine made an observation that we honor here. In an ordinary sense of ‘kind’, kinds are not just bare sets, but rather sets whose members are similar in some way: for two things to be kindred, they must be connected by something like similarity (Quine Reference Quine1969: 117). Thus, unkind sets are not kinds, even if basic kinds are. (Because two individuals can be more or less similar, basicness can come in degrees. To simplify discussion, however, we proceed to talk about kinds as either basic or not. We also simplify by not trying to sort out a workable analysis of ‘similarity’. See Quine [Reference Quine1969] and Audi [Reference Audi2013] for more.)

As it will be important, let us take a moment to explain why we use the cumbersome phrase, ‘directly relevant to science.’ Basic kinds may be quite relevant to science in an indirect way. Things around trees, for example, might be relevant to the study of tree disease or fire science—things around trees might significantly overlap with fire fuel, and fire fuel is scientifically relevant if fire science is a science. But things around trees are not always fire fuel, and so their connection to science is at best indirect. Another way of saying this is that we might use the category things around trees as an imperfect proxy for a category that is scientifically relevant, fire fuel, but the category things around trees is not itself scientifically relevant. Its essential properties are not perfectly coextensive with the properties that are useful to science. Moreover, there is a deeper explanation for this imperfect overlap: basic kinds (and unkind sets) as such lack causal powers, and so (on some accounts) they will not be part of the fundamental level of things, even if they are grounded on the fundamental level (Schaffer Reference Schaffer2004).

With those distinctions drawn, we can name some theories. Basic realism is the view that basic kinds like things around a tree are real, and unkind realism is the position that unkind sets are real. In what follows, we will have little to say about unkind realism as we focus on basic realism. But both of these brands of realism pack a lot into the ontological suitcase. Those who oppose these views argue that the realm of the real is more exclusive, only admitting certain elite properties. We call this view elitism. We hope that this section has convinced you that basic realism has plausibility as a good, intuitive theory. In particular, we hope to have made a case that for basic kinds like things around a tree, existence and mind-independence are sufficient to earn the title ‘real’ on one unpacking of that ordinary term as it works in ordinary language. After examining the application of basic realism to race in section 2, we argue in section 3 that it can meet objections from the elitists.

2. Basic Racial Realism

Many people once thought that races were robust kinds. According to that way of thinking, members of the same race share not only distinctive visible traits like skin color but also some sort of deeper essence (based in ‘racial blood’) that at least partially determines cognitive and behavioral traits, such as intelligence or a virtuous disposition. Among those whose job it is to study race, very few, if any, believe this anymore. No doubt this consensus is partly due to the ascendance of antiracism. But it is also due to the fact that we know more about human variation, inheritance, and behavior than we used to know, and the verdict is in on that one: races are not robust kinds (e.g., Appiah Reference Appiah and Appiah1992; Appiah Reference Appiah, Appiah and Gutmann1996; Zack Reference Zack2002).

Nonetheless, races might be something more austere than those robust kinds, and it is here where there is a lively and interesting discussion afoot among the Big Three camps noted earlier. One reason why many racial antirealists believe that race is not real is that there appears to be no biologically principled basis for race. Constructionists, though, argue that this inference is too quick: they maintain that even if race is not biologically real, it is still real in another, socially constructed way. On this view, a race like white people is just like graduate students or oenophiles: though races are not biological kinds, they are kinds of people that we literally make real when we classify each other in certain ways or engage in certain other practices. These social kinds are properly studied by social science, not natural science. Finally, the third position denies the premise shared by the first two: race, according to biological racial realists, is biologically real after all. The details vary within this camp, but as a whole its members tend to agree that there is a nonracist, nonrobust, and scientifically sound way of talking about race in biological terms.

The exchange between these three positions has mostly failed to note a fourth position that is not only viable but can even exploit a dialectical advantage over the two dominant realist theories and neutralize common antirealist arguments. This fourth position is basic racial realism. It holds that races are real, basic kinds: races are unified by a similarity that is not directly relevant to any natural or social science. On our particular specification of basic racial realism, races are groups of people who are distinguished from other groups by having certain visible features (like skin color) to a significantly disproportionate degree. But as with all basic kinds, the points of similarity are not directly scientifically relevant.

It is important to see that basic racial kinds might be—we think are—indirectly related to scientifically relevant kinds. Consider links between certain racial populations in certain social contexts and disease. It is well-established, for example, that people racialized as black who live in the United States (but not always those living elsewhere) have a host of diseases that are exacerbated by stress, such as heart disease, likely due to living in an antiblack, racist society. Other racialized populations might experience unique morbidity rates due to disproportional exposure to environmental toxins (Clark et al. Reference Clark, Millet and Marshall2014). Medical science, then, will study races and their links to disease. But this does not mean that race is a medical or biological kind. As with things around trees and fire fuels, so with races and heart disease populations: the two are not perfectly coextensive, because the unifying properties of the two kinds—similarity of visible traits and disease properties—are not identical, and only the latter has the causal power to generate consequences relevant to science. (Again: many groups of people racialized as black outside the United States do not have similar rates of disease.) Similarly, social science will study a large number of social facts about races. But for reasons we will see shortly, the racialized populations that social science studies and race are arguably not coextensive. So these connections between race and science are proxy connections at best. Strictly speaking, science studies not races but other populations that partially but not perfectly overlap with race.

Why basic kinds, rather than unkind sets? It's embarrassingly simple, really. Names such as ‘white’ and ‘black’ don't just coincidentally refer to colors as well as races; these names were chosen because race is tied to visible traits such as skin color. So, we believe (though we do not decisively argue for the claim here) that races are basic kinds rather than mere sets because we think that the very concept of race dictates that racial groups are supposed to be distinguished by certain visible traits, such as skin color, that their members share to some significantly disproportional extent—that is, their members are supposed to be similar in some way. If so, then each race is not a mere set; it is at least a basic kind.Footnote ¹

Turn now to how basic racial realism enjoys a substantial advantage in the race debate. The common problem infecting the debate as a whole is the elitist premise that to be real, race must be scientifically relevant. This presupposition is dubious, given our arguments in section 1.

To expose this advantage, we begin with antirealism. As noted above, many antirealists maintain that race is an illusion because it is not vindicated by the biological sciences (e.g., Zack Reference Zack2002). Others argue that races aren't real because they fail to be either biological or social kinds (Appiah Reference Appiah, Appiah and Gutmann1996; Glasgow Reference Glasgow2009). Both approaches presuppose the elitist premise, leaving an opening for basic realism. The category, white people, for example, might for basic realism include Bill Clinton and Francis Ford Coppola, but not Yukio Hatoyama or Spike Lee. (We choose these examples only for illustration; if you think that we have misplaced any of them, then go ahead and reclassify accordingly.) In holding this, basic realists can agree with antirealists that Clinton and Coppola are not white by virtue of any social practices, so they are not (as such) members of a social kind. We can also stipulate that they are not white by virtue of any biological principle, in which case they are not (merely by virtue of being white) members of a biological kind. And they are not white by virtue of sharing theoretically important properties or properties that enable much induction and generalization, so they are not part of a natural kind just by being white. Nonetheless, they could comprise a real grouping, for it does not follow from the stipulation that races are not biological or social or natural kinds that races are not real. They might be basic kinds. The antirealist arguments appear to fail because they include the undefended elitist premise that races must be scientifically relevant to be real. In fact, that premise appears indefensible.

A similarly hasty presupposition of the exclusionary, elitist premise can be found in representatives from the constructionist and biological racial realist camps, too. In all cases, the ground shifts from needing to show that race is real to trying to show that it is real in a scientifically relevant way. And in all cases, this jump puts racial realism in a dialectically weaker position than it needs to be in.

Consider first biological racial realism. Some, like Robin Andreasen (Reference Andreasen1998, Reference Andreasen2000) and Philip Kitcher (Reference Kitcher and Harris1999), argue that races are biologically real because they exhibit certain breeding patterns—races are breeding populations. (Kitcher [Reference Kitcher2007] has since updated his view to hold that races are pragmatically defined kinds.) One common counterargument to this position, the ‘mismatch objection’, is that these populations aren't racial groups: the two don't sufficiently match up because races are necessarily organized according to visible traits and breeding populations are not. Consider some details of the mismatch objection. Some of the ancestral groups identified by science as constituting reproductive populations, such as the Amish, do not appear to be ordinarily identified racial groups; and some ordinarily identified racial groups, such as ‘Asians’, either are not ancestral populations or are organized differently than biological populations on at least some of the data that are coming out—a project still in progress, to be fair. To turn from biological facts to thought experiments, if two ancestral groups looked exactly alike but were reproductively isolated, populationists would have to separate them ‘racially’. But that seems to work with an objectionably idiosyncratic notion of race, if we are attempting to address the entrenched debate over the commonsense concept of race (Glasgow Reference Glasgow2009, Reference Glasgow and Hazlett2010). Imagine that God instantaneously created an exact duplicate of earth, so that we have apparent racial duplicates whose ancestral lineages are separate from ours. This view would have to say that, because our Al Gore and twin earth's Al Gore have different lineages, twin-Al is not white! In addition, if two populations engaged in thorough interracial reproduction but still produced what looked like very different races (say that a quirk in recessive genes made inheritance of the relevant traits decidedly binary), populationists would have to say that the offspring eventually constitute one race, even though ordinary intuition would sort them into different races. For all of these reasons, it looks like race and breeding population are not sufficiently coextensive.

Of course, more must be said to make this argument fully compelling (see, for example, Glasgow Reference Glasgow2009: ch. 5 and Glasgow Reference Glasgow and Hazlett2010). We have more modest aims. We merely want to point out that the mismatch objection can be avoided entirely if we don't try to vindicate races as biologically real. If a realist position on race is what we want, a more austere basic racial realism can do the trick; and its trick will be more dialectically compelling than new wave biological realism, for it can unproblematically tie race to visible traits, which will protect it against the mismatch problem. That is, basic racial realism can say that races are basic kinds organized along visible-trait lines, allowing the view to track the ordinary concept of race.

In fact, as long as there is some organizing similarity, basic realism can pick out pretty much whatever races commonsense wants! Biological racial realism has a harder time vindicating common sense because commonsense racial categories evolved under local social and political pressures that frequently had little to do with good biology. It was always unlikely that racial discourse would coincidentally be vindicated by the natural sciences. Basic realism, though, doesn't face the limitations of science—it can get kinds much more cheaply than biology can. And it can get us much of what we want out of biological race. In particular, as we have seen, it can make sense of using race in science indirectly, due to the imperfect overlap between scientific populations—including both disease populations and breeding populations—and basic races. As long as we don't commit to any more robust connection than that proxy connection, we can avoid the predictable worry that our concept of race and the biology of humanity come apart.

Let's turn to the constructionist camp. Sally Haslanger (Reference Haslanger, A. Koenig, Soo-Jin Lee and Richardson2008: 60) comes very close to articulating a basic realist position when she writes,

But Haslanger rejects basic realism in favor of constructionism. On her interpretation of racial discourse, we refer to something social by ‘race’. Races are more than mere basic kinds, she maintains—they are real social kinds.

But although Haslanger lays out the linguistic framework for her view, she gives no argument for moving from the more modest proposal of basic racial realism to the harder-to-defend interpretation of racial discourse that leads to constructionism. And in fact, as with biological racial realism, constructionism's weaknesses are no problem for basic realism. One compelling argument against constructionism is that the concept of race seems to be structured such that race is not subject to social forces (Glasgow Reference Glasgow2009, esp. ch. 6). Fully demonstrating this point requires working through the various social forces that different constructionists place at the center of race, but for the sake of illustration imagine that the constructionist says that the force in question is the mere practice of engaging in racial classification. Calling people ‘Asian’ makes people Asian, then. The objection to this view hangs on a thought experiment: what if, for an hour, the world was collectively struck with racial amnesia, and consequently we stopped calling people ‘Asian’? Would the people we currently classify as Asian thereby stop being Asian? On the ordinary concept of race, when we forget how to classify people racially for that hour, does Al Gore stop being white for that hour? Does he then magically re-become white after the hour is up and we return to normal practice? The most intuitive answer to these questions seems to be no: Al Gore stays white even if we happen to forget that he is white. On the ordinary concept of race, race is not socially contingent like that. The same argument can be run, mutatis mutandis, for other brands of constructionism. Moreover, what does seem to determine race—what keeps race around, even when we forget to classify one another racially—is certain visible traits like skin color. That is why Al Gore remains white. In short, then, race and racialized social group are not sufficiently coextensive.

If that criticism succeeds, then race is not a social kind. Now, as with the mismatch objection, much more needs to be said to make good on this argument. (For example, perhaps more entrenched commitments about race should push us to give up our intuitions about the amnesia case. More generally, a host of methodological and semantic assumptions underlie the foregoing argument against constructionism—see Glasgow [Reference Glasgow2009, Reference Glasgow and Hazlett2010, forthcoming, and ms.] for defense.) But we can make theoretical progress by going in another direction. If what we want is a defensible form of realism, a less burdensome approach would be to say that races are basic kinds. This approach allows us to say that races are what they are by virtue of their visible traits and not by virtue of social practices. Indeed, a strength of basic realism is that this similarity can just define each race, and so Al Gore retains his race even if we forget about race for an hour. Thus, the antirealist arguments against constructionism won't work against basic realism. Again, it avoids a key landmine in the race debate.

And as with biological realism, basic realism can secure the main benefits of constructionism. According to the anticonstructionist argument, we know that basic races and racialized social groups are not sufficiently coextensive only from certain unlikely cases. So basic realism can allow that basic races and racialized social groups are significantly coextensive. In the actual world (as opposed to the counterfactual worlds imagined above), they are perfectly coextensive! Accordingly, when social scientists study, say, racial discrimination, they can easily use proxy connections between racialized social groups and basic races.

In addition, basic racial realism can help secure the ideas that our racial practices have made a profound impact on lives and that we should attend to that impact. This impact is the very foundation of some constructionist views of race, and race theorists worry about ontologies of race that cannot capture the import of this impact. For example, some maintain that antirealist views of race prevent us from talking about race, which is unacceptable, since talking about race is an important tool in combating injustice and making sense of people's experiences and identities (e.g., Du Bois Reference Du Bois, Blight and Gooding-Williams1897; Hardimon Reference Hardimon2003; Outlaw Reference Outlaw1996; Sundstrom Reference Sundstrom2002; Taylor Reference Taylor2004: 126).Footnote ² In a similar fashion, some might worry that basic racial realism also disables the fight for justice if it says that race is not something social. But this concern would be misplaced. Basic racial realism can directly legitimize a racial discourse that enables justice and makes sense of racial identity, experience, and history. According to basic racial realism, races are real kinds, so it is kosher to talk about race, and it is fine to use proxy connections between races and racialized groups. And this means that basic realism can license talk about the social significance and impact of race, and about social identities that are built upon race. This can be preserved even if we don't want to talk about there being some social reality to racial groups (as opposed to, say, racialized groups or groups of people with shared racial identities). In other words, basic racial realism takes away nothing about the lived, social reality of racial experience; it just classifies that reality as part of practices involving race rather than as part of race itself. On this front, the difference between constructionism and basic racial realism is a bookkeeping difference, not a substantial gap in theoretical resources.

It is important, then, that basic racial realism secures much of what we want from a good theory of race, without committing to the more onerous ontology of race put forth in constructionism and biological realism. Of course, we haven't argued that the objections to biological racial realism and constructionism are decisive. Rather, we’ve only recapitulated the broad outline of those arguments in order to show that basic realism has certain advantages: whether or not biological realism and constructionism can ultimately be successfully defended against these objections (objections that we believe to be powerful), basic realism doesn't even face those challenges. It just avoids the objections by having weaker commitments. So embracing basic realism is rewarded with a dialectical advantage.

And recall that in securing this advantage, basic realism neutralizes racial antirealism, too. Extant arguments for antirealism rely on the negative premise that if neither biological racial realism nor constructionism succeeds, antirealism is the last contender standing. As is now obvious, we hope, this leaves basic realism unaddressed—antirealism is not in fact the last contender standing. For these reasons, at the very least basic realism deserves serious consideration for placement alongside the Big Three in any discussion about the reality of race. More enticingly, basic racial realism may be leading the pack, as it looks unchallenged and retains strong intuitive credentials.

But is basic realism a little too convenient? Does it mean that naming anything as any kind of thing makes it real as that kind of thing? For example, do we now have to say that witches were real basic kinds in colonial Salem because some people there were called ‘witches’?

Basic realism does not have that result because it maintains that our semantic intentions provide application conditions for terms, and these conditions preclude the reality of witches, mermaids, and the like. We will put this in terms of conceptually nonnegotiable commitments. A commitment is conceptually nonnegotiable if abandoning it means abandoning the concept in question. For instance, if you say that bachelors need not be humans, you fail to use the (or anyway our) concept of bachelor. Some similarly nonnegotiable commitments attach to the concept of witch, in the colonial sense. If the people called ‘witch’ in Salem couldn't cast spells, then they simply were not witches in the operative sense. Similarly, if the horselike animal doesn't have a horn, then it is not a unicorn. It's part of the concept unicorn that it has to have (or be part of a species that typically has) a horn to count as a unicorn. We can't make unicorns real by calling Secretariat a unicorn, and we can't make witches real by calling Martha Corey a witch, because Secretariat and Martha Corey fail to have a property that is nonnegotiably embedded in the predicate applied to them. So basic realism need not say that we can make anything a real f just by giving it a name ‘f’. For any f to gain entry into the real world, some extralinguistic facts must be in place, namely, there must actually be something that has the properties that are nonnegotiably embedded in the concept of f.Footnote ³

Our question, then, is whether race fails to exist in even a basic way because nothing in the world satisfies the conceptually nonnegotiable requirements for counting as a race. But far from being a liability, it is on this question that basic realism about race appears to occupy a position of dialectical strength. The arguments briefly canvassed above against constructionism and biological racial realism hinged on two key claims. The first is that races must, as a conceptually nonnegotiable matter, be organized around certain visible traits in a way that maps onto our racial categories. New-wave biological racial realism struggles to comport with this requirement by dissolving the necessary connection between race and those visible traits. Constructionists don't have a problem with that fact: they maintain that we do organize races around visible traits, but it is we that do this, and it is our very doing of this that makes race real. But the problem with this move is that, second, races must not, as a conceptually nonnegotiable matter, be determined by social facts (because, again, when we imagine removing those facts, racial facts appear unchanged). Basic racial realism satisfies both of these conceptual requirements: it says that races are unified by visible traits, not social traits.Footnote ⁴

Given the dialectical advantages explored here, basic racial realism seems an especially powerful view. But the fact that it has gone unaddressed raises other questions. In particular, some might think that basic realism has been neglected because race theorists have been searching for some more robust, scientific, or deep reality for race. In that case, doesn't basic realism simply fail to share their project? Perhaps basic realism is not really a rival to constructionism or biological racial realism or antirealism, because it's an answer to a question that is different from the one they are asking.

We do recognize that robust races are what some race theorists have been investigating (e.g., Appiah Reference Appiah, Appiah and Gutmann1996; Blum Reference Blum2002). Nevertheless, to whatever extent this might reflect the motives of those in the race debate, it is an overly narrow approach to fulfilling the broadest aims of race theory. The ultimate goal is not to limit ourselves to the science of race but, more expansively, to track the truth about race accurately. That, after all, is the only reason why it would be worth investigating whether race is an illusion. This goal serves a policy purpose, too: whether we eliminate, conserve, or rebuild our race-talk depends on what the nature of race is, in particular on whether race is real. And these aims of finding the truth about race and deciding what path we want for our race-talk are no less served by basic realism than by any of the other three approaches to the ontology of race. When it comes to the race debate that has actually occupied the pages of academic journals, popular press, and mainstream thought, basic realism is a relevant entrant and, in fact, one that looks especially promising.

Basic realism is exciting not just because it occupies a strong position in the debate. Basic realism is also appealing because it very nicely reflects a certain commonsense way of thinking about race. Permit us an unstudied anecdote: when ‘civilians’ hear academics debate the reality of race, it often comes across as patently disconnected from obvious truth. Sometimes this is because those civilians aren't fully informed about the important science and sophisticated arguments that pressure us to think that race is a fiction. But other times it is arguably because academics are working with a notion of race that is not as fluid and flexible as the ordinary concept of race. Academics glue race to a certain commitment about ‘real biological kinds’ or to certain social facts, and that glue threatens the reality of race. But if ordinary use of race-talk is anything, it is malleable. Civilians can hear the academic race debates and reasonably respond: ‘Yeah, but we all look different, right? Why isn't that just race?’ Basic realism's austerity allows it to answer with a straightforward ‘Yes, that just is race!’ Nothing more need be said. The other three views commit to a more elitist notion of race, and that is where they court trouble.

Moreover, basic realism can recognize that our race-talk evolved not only as part of a rudimentary (and bad) science, but also and arguably more centrally as an artifact of geopolitical and economic pressures that were, in many cases, morally disastrous. One widely noted result of these pressures is the phenomenon that race doesn't travel, that is, that one's race can change as one moves between societies with different racial classification systems (Root Reference Root2000; Mallon Reference Mallon2004; Glasgow Reference Glasgow2007). Basic realism explains this well. Consider being mixed race. At times, the United States has followed the ‘one-drop’ rule that any black ancestry makes a person black—although in some respects the rule's influence is waning today (Glasgow, Shulman, and Covarrubias Reference Glasgow, Shulman and Covarrubias2009). It is tempting to point out the absurdity of such a rule. It is absurd! But, absurd (and morally dubious) as it is, it also sets humans off from one another. That is, for all of its warts, it does succeed as a (crude) principle of demarcation. And it succeeds in demarcation, arguably, not because social forces create races (see above), but because it demarcates real, socially independent basic kinds. Importantly, societies that allow for mixed-race identities also recognize basic kinds, just different ones than those recognized by the one-drop rule. And a society that followed a reverse one-drop rule, where any white ancestry suffices to make one white, would also be tracking basic kinds, just different ones than we track now.

This is why race doesn't travel, according to basic racial realism: race comes so cheaply that different societies can recognize different sets of human categories while remaining true to the facts about human variation. We each have, then, multiple properties, appropriately called ‘races’ that are indexed to different social classification schemas—a person might be Black_UnitedStates and simultaneously White_Brazil, for example. Our view is that this is best understood as a ‘modest ontological localism’ (Glasgow Reference Glasgow2007). The view is not that one has a race simpliciter that changes from location to location (that seems inconsistent with the ordinary concept of race in the United States) nor merely the anthropological observation that different communities use different racial frameworks (that is too weak). Rather, modest ontological localism says that we have multiple races, different ones of which become uniquely relevant at different locations.

The overall view, then, is simple. The basic kinds are out there. The metaphysics of race stops with that point. The practical questions, the questions of social justice, psychological health, and ethics, then become which basic races we want to recruit into our social practice.

But does this expose the basic realist position to a problem comparable to the problems facing constructionism and biological racial realism insofar as it fails to match the ordinary concept of race? Doesn't it imply that some people who are black in our society are white in another society, clearly violating ordinary race-talk? In fact, doesn't our view imply that we might have 40 races based on different shades of skin color or other visible features? And if basic realism does have these implications, doesn't it commit the very sin of violating the ordinary concept of race that we used as grounds for rejecting basic racial realism's realist rivals?Footnote ⁵

We acknowledge (indeed, embrace) the fact that basic racial realism will have some implications that are surprising to many competent users of ordinary racial discourse. But we believe that these surprises are consistent with the ordinary concept of race in a way that the radical implications of populationism and constructionism are not—those views violate the concept of race, while basic realism does not. The surprise of basic racial realism is essentially that there are many more races than you might have thought and that you may belong to many more races than you thought. Barack Obama, for example, has at least three races (and almost certainly many more) on this account. This will be revealed in different societies, where different schemas of race will be operative, bringing different racial identities to the fore: in a one-drop-rule society, his relevant race will be black; in a reverse-one-drop-rule society, it’ll be white; and in a mixed-race-accepting society, it’ll be mixed race. This may be surprising, but we believe it is consistent with ordinary racial thinking. If you’ll permit us one more anecdotal observation—and, of course, more robust data would be better—we’ve found that when people learn that there are different racial classification systems in different societies, they tend to accept it (and do not insist that their own is somehow more privileged). We think, then, that there is an implicit willingness to accept this surprise. If so, then this particular surprise is consistent with whatever is conceptually nonnegotiable about race. And if the foregoing is right, that consistency makes sense, because on that account the main nonnegotiable propositions are that race must be tied to visible traits and not tied to social facts. These propositions say nothing about having one or several races. (That said, we acknowledge that this argument may not convince everyone. Mallon [Reference Mallon2004: 657] argues that the idea that race does not travel departs from the commonsense idea of race.)

By contrast, constructionism has races doing things they simply cannot do according to the ordinary concept of race, at least if the argument given above is correct. Constructionism implies that we can lose our races—even within one society's racial scheme—and then regain them, just based on amnesia. We believe this violates the concept of race. It's an imperfect analogy, but consider the following. Basic racial realism's surprise is akin to learning that you have a bunch of cousins you did not know about. Constructionism's surprise is akin to learning that your cousin can go from being your cousin to not being your cousin if we just forget she's your cousin and become re-related to you again if we remember. The former is a big surprise; the latter violates the concept of cousin.

Similarly, populationism says that groups that are exactly visibly identical but reproductively separated would be of different races, such that the doppelgängers of the Gores on twin earth would not count as white. We believe that this implication (and others) of populationism violates nonnegotiable commitments of ordinary race-talk.

We conclude, then, that though basic racial realism has unexpected implications, its surprises are consistent with the ordinary concept of race in a way that the surprises of populationism and constructionism are not. It appears, then, that basic racial realism is in a very strong position.

3. The Realm of the Real

Basic realism is not a new position, and even unkind realism enjoys strong pedigree. Mill (Reference Mill1959: book I, ch. 7) holds that, while natural or ‘real’ kinds are differentiated from other ‘classes’ (like white things) by whether they are distinguishable by a large number of properties, both natural/real kinds and other classes are real in the ordinary sense of being ‘made by nature’. Likewise, David Lewis (Reference Lewis1983, Reference Lewis1984) grants the reality of sets, even though he wants to make sure that some of them, the ‘natural’ ones, are set aside to do some metaphysical heavy lifting. Catherine Elgin (Reference Elgin1995) also comes very close to endorsing unkind realism,Footnote ⁶ and T. E. Wilkerson (Reference Wilkerson1995) holds that there are ‘real but superficial kinds’, which are very close to what we are calling ‘basic kinds’.Footnote ⁷ This pedigree notwithstanding, there are two reasons that one might want to be an elitist and exclude basic kinds from the realm of the real.

The first elitist move says that only certain properties are metaphysically privileged, and only privileged categories are real. For example, consider the claim that the disjunctive properties that unify unkind sets are ‘gimmicky and fishy’, as contrasted with ‘natural or genuine’ properties (Shoemaker Reference Shoemaker1988; Vallentyne Reference Vallentyne1998). We take no issue with this assessment: we offer no comment on such technical categories as the gimmicky, the fishy, the natural, or the genuine. But some go a step further and suggest that if kinds are not natural or fundamental in this deep sense, then they are not real (Hardimon Reference Hardimon2012). (Vallentyne (Reference Vallentyne1998: 172) seems not to go this far; Shoemaker (Reference Shoemaker1988: 219–220) sympathizes with this kind of elitism, but he acknowledges having difficulty arguing against permissive views.) We disagree with that kind of elitism. In claiming that basic kinds are real, basic realism claims the mantle of reality for basic kinds. As argued above, because basic kinds exist mind-independently, we capture a legitimate usage of ‘real’ when we say that basic kinds are real.Footnote ⁸

We also claim that basic kinds and unkind sets are objective. Many, including John Dupré (Reference Dupré1993: 5), Philip Kitcher (Reference Kitcher1984: 330), and Ted Sider (Reference Sider, Chalmers, Manley and Wasserman2009: §7), disagree, at least when it comes to unkind sets. And although we are focused on basic realism, it is instructive to note that even unkind realism can take a stand here. Objectively, nature does have what we can call ‘sundogs’: there are, mind-independently, things that are either suns or dogs, even if there is no utility in our recognizing them as such. And there are things around trees, even if that basic kind directly serves no scientific purpose. Thus, it is hard to see any good reason why we should say that basic kinds (and unkind sets) are not objective. That is, unless ‘objective’ is just used as a sneaked-in stipulated synonym for some property that is privileged by the rival theory (such as utility), in which case it can't really do any non-question-begging work. To be non-question-begging, ‘objective’ seems to have to refer to something like mind-independence, which is possessed by some basic kinds and unkind sets.Footnote ⁹

So far we have just appealed to intuition about how to understand such honorifics as ‘real’ and ‘objective’: we think that on a plausible, ordinary way of thinking, ‘real’ refers to existence (and possibly mind-independence) and ‘objective’ refers to something like mind-independence. But there is another argument against elitism. Any position that makes scientific relevance a condition of being real or objective exceedingly exposes reality and objectivity to the vagaries of whatever science focuses on, and that seems problematic. We cannot offer a full theory about what makes a kind scientifically relevant, but we want to illustrate this point with one possible way in which a kind might be scientifically relevant: entering into causal relations.Footnote ¹⁰ If sundogs (as such) suddenly were the only things able to kill insects, sundogs would spontaneously become scientifically relevant, on this possible theory of scientific relevance. That sounds sensible enough—sundogs can do certain things now, and they could not, as such, do anything before. So science now needs to attend to them. But those who think that scientifically relevant kinds alone are real or objective would be committed to saying that sundogs go from being illusory and nonobjective to real or objective. That seems much harder to accept. Sundogs’ newfound power might not have even come from a change in suns or dogs; it might originate in some change in insects. Surely such a fluke change in the causal nexus of the universe cannot make an otherwise unchanging category go from illusory and nonobjective to real or objective; it just makes for a change in causal powers—a change in scientific relevance, arguably. We suspect that the same argument, mutatis mutandis, can be run for any other reasonable theory of scientific relevance since it seems generally possible for things to become more or less relevant to science without that affecting their status as real, unless one begs the question by stipulating that what is real just is what is scientifically relevant. When reality and objectivity are characterized in non-question-begging terms, permissiveness seems to be in order, so that basic kinds and possibly even unkind sets can count as real and objective.

Finally, we want to register one last note on the claim made by Lewis, Sider, and Vallentyne that their privileged divisions uniquely carve nature at the joints. We think that it is fair to say that unkind sets and basic kinds also carve nature at the joints. But this is probably just a hopeless semantic dispute over ‘joint carving’, since there is no antecedently given, ordinary concept of ‘jointcarving’. We focus on reality, leave carving to the butchers, and follow Lewis (Reference Lewis1984: 231–232) in understanding realism as the view that we don't make the world's divisions. If that is realism, then the realist must accept the reality of basic kinds (and unkind sets). We think it is natural to understand them as objective, too.

To make a long story short, then, you need not be inclined to reject basic racial realism if you think that basic kinds are not among the natural or elite or privileged kinds. You can privilege whatever kinds you want. We can even agree that elite existents are plausibly the grounds of abundant basic kinds (Schaffer Reference Schaffer, Chalmers, Manley and Wasserman2009: 353). Just don't on that basis exclude basic kinds from the realm of the real and objective. They exist, and they do so mind-independently.

A second possible reason for rejecting basic realism is parsimony. As we have seen, basic realism hangs its case on a commonsense understanding of ‘real’ according to which existence and mind-independence are the most that can be required for entry into the realm of the real. (We actually prefer an even more permissive realism that eschews the mind-independence requirement and keeps only existence as a requirement for reality, so that social kinds like philosophy journal are real, as they indeed seem to be. That said, we’re willing to limit the discussion to mind-independent kinds.) But that conception of reality makes the number of real categories expansive. Is it an excessive ontology?

One answer is that in the end, the twin pressures to admit everything that appears to exist and to simplify as much as possible might not have to be traded off against one another. We might just be able to say that the catalog of what is real exceeds the most reduced catalog of what is real. After all, it is not obvious that reduction entails elimination of that which is reduced. As Jonathan Schaffer points out (Reference Glasgow2009: 361), if we ground basic kinds on more fundamental entities and if our level of fundamental entities is itself parsimonious, then we haven't really violated Occam's Razor. Race-talk operates at a pretty high and very abundant macro level, which might be wholly grounded on a very sparse fundamental level.

But say you remain unconvinced. You only want the most reduced, fundamental level; everything else is ontological waste in your view. In that case, our answer returns to the race debate specifically, where appeals to this radical kind of eliminativism are not relevant. All sides in the race debate have accepted that if racial facts are either biological or social facts, then race would be real. (The rival views just disagree about the truth of the antecedent.) But neither reproductive facts nor social facts are going to be included in the most reduced, parsimonious account of what there is. If there are races—the question we are pursuing—they obviously will not be found at the most reduced level of description. They are not atomic facts, nor features of quarks, nor of whatever might lie beneath quarks. So as far as the race debate is concerned, appeals to radical eliminativism are beside the point. If we wanted only the fundamental bits of the universe, we could have abandoned the race debate before it got started—everyone knew from the beginning that races aren't among those bits. The whole question of race would be uninteresting. (Cue the howling traditionalists!) So our conversation is concerned with another question—whether race is real in some other, more abundant sense. In that conversation, if extreme reduction is not a strike against biological racial realism or constructionism, it cannot be a strike against basic racial realism, either.