Radical alterity is just another way of saying “reality”

As a response to Eduardo Viveiros de Castro’s critique of my essay “Fetishes are gods in the process of construction,” this paper enters into critical engagement with anthropological proponents of what has been called the “ontological turn.” Among other engagements, I note that my own reflections on Malagasy fanafody, or medicine, are informed by just the sort of self-conscious reflections my informants make on epistemology, something that anthropologists typically ignore. After making note of the arguments of Roy Bhaskar that most post-Cartesian philosophy rests on an “epistemic fallacy,” I further argue that a realist ontology, combined with broad theoretical relativism, is a more compelling political position than the “ontological anarchy” and theoretical intolerance of ontological turn exponents.

anthropologists do engage in polemics nowadays, they more often than not seem to be talking past each other. If not shouting.
In this sense, it might be said that Eduardo Viveiros de Castro, in his recent Marilyn Strathern lecture, "Who's afraid of the ontological wolf?" (2015), is proposing we revive the old grand tradition. In singling out some of the arguments in an essay I wrote on fetishism as examples of what he suggests should be "inadmissible moves" in anthropology (Graeber 2005), he is, at least as I understand him, calling for a response; throwing down a gauntlet, as it were, but doing so in such an unusually gracious, good-natured, and friendly way that it strikes me his challenge does give us an opportunity to revive an old tradition in a new, more gene rous spirit. I must say I feel a little honored by the opportunity. I am a deep admirer of Eduardo Viveiros de Castro's work, and I also see him as something of a fellow spirit and political ally, in that we both are activists who feel that the discipline of anthropology is not only uniquely positioned to answer questions of universal philosophical import, but also has something crucial to contribute to the cause of human freedom.
In other words, we definitely share enough common ground that an exploration of our differences could be instructive.
Finally, I think the immediate topic of our disagreement-what is and what is not permissible for an ethnographer to say about a Malagasy hail charm-however apparently specific, does indeed open up questions our discipline would do well to address, questions that do indeed have larger political implications. *** Let me state the matter in brief. Viveiros de Castro has over the last decade become something of a standard bearer for what has come to be known in anthropology as the "ontological turn" (hence, OT; see, among others: Candea 2011; Henare, Holbraad, and Wastel 2007;Holbraad 2007Holbraad , 2008Holbraad , 2009Holbraad , 2012aHolbraad , 2012bPedersen 2001Pedersen , 2011Pedersen , 2012Viveiros de Castro 2003, 2015cf. Heywood 2012;Laidlaw 2012;Salmond 2014). In his Strathern lecture, he singles me out as an example of an old-fashioned anthropologist who still clings to the old habits of breezily dismissing what used to be called "apparently irrational beliefs"-in this case, that a charm called Ravololona can stop hailstorms from falling on farmers' crops-as untrue in the literal sense, and therefore, having to be explained as a projection of social relations of some sort. Such an approach, he suggests, has really not advanced in any fundamental way since Evans-Pritchard (1937) argued that Zande ideas about witchcraft cannot be literally true, and that rather than simply compile apparently contradictory statements and try to imagine what these people would have to think in order for all these statements to be consistent, the real task of the ethnographer is to understand how society is organized in such a way that no one ever notices the statements are contradictory in the first place. Now I must admit that, for my own part, I don't find an affinity with Evans-Pritchard particularly shameful. This is because I've always believed that his exposition of Zande witchcraft is one of the most brilliant analyses of ideology ever written. Evans-Pritchard's central point was that, when talking in the abstract, Zande would almost invariably make statements (e.g., witches are a small collective of self-consciously evil agents; sons of male witches were always witches too) that obviously contradicted everyday practice (everyone had admitted to unconscious witchcraft at some point in their lives, nobody speaks of witch lineages). Why does no one seem to notice this contradiction? Evans-Pritchard's answer is that their society is arranged in such a way that the two are never juxtaposed. Zande aren't sociologists. They do not generalize from what they say about individuals to think about what it would mean for society as a whole. But is it not exactly the same in our own society, where it's commonplace to make equally absurd generalizations ("anyone who's sufficiently determined and genuinely believes in themselves can become successful")-despite the obvious day-to-day reality that, even if every single person in the country woke up one morning determined to become the next Sir Richard Branson, society is so arranged that there would still have to be bus drivers, janitors, nurses, and cashiers?
If so, the question: "What would Zande have to believe for these statements to be consistent?" is exactly the wrong one to ask. OTers of course would agree, but for the opposite reason. They would argue that the question does not go far enough: the real question should not be "What would Zande have to believe?" but "What would the world actually have to be like for these statements to be consistent?" It is then incumbent on the ethnographer to write as if, for the Zande at least, this world actually does exist; to recognize its radical alterity, accept that we could never entirely understand it, but nonetheless allow the concepts that underlie it "unsettle" our own theoretical beliefs.
Hence Viveiros de Castro's objection to my argument about the Malagasy hail charm.
What I will do in this essay is, first of all, address the specific charge leveled against me by placing my remarks in full ethnographic context. Doing so, I believe, shall reveal what's really at stake in the rift between OT approaches and those of anthropologists such as myself. The essay will then explore some of the founding texts the current OT tradition so as to argue, among other things, that, despite protestations to the contrary, OT does not abandon the traditional philosophical quest for a universal ontology, but rather proposes its own tacit universal ontology, which is essentially a form of philosophical Idealism. In contrast, I put forward a case for combining ontological realism with theoretical relativism, and suggest that, far from trying to impose my own theoretical views on my Malagasy interlocutors "behind their backs, " this approach is far closer to the way Malagasy tended to think about such questions, and therefore, more conducive to a meaningful dialogue about those things they considered most important.

The debate
There's no need here to summarize my paper on fetishism in any detail since Viveiros de Castro addresses only very little of it, but suffice it to say that it is largely an essay on double-think. The word "fetish" is ordinarily invoked when people seem to talk one way and act another. The surprising thing is that this can happen in entirely contrary ways. In the case of the African objects that came to be labeled "fetishes" by European merchants and other travelers, those who employed them insisted that the objects were gods but acted as if they did not believe this (such gods could be created, or cast away, as needed). In the case of contemporary commodity fetishism, it's quite the opposite: the average stockbroker will insist he does not really "believe" that pork bellies are doing this or securitized derivatives doing that-i.e., that these are just figures of speech. On the contrary, he acts as if he does believe they are doing these things. I remarked that, in my experience, Malagasy ody-usually translated as "charms"-were quite similar to African fetishes in this respect. Finally, I suggested that this kind of double-think is typical of moments of social creativity. Here, a classic Marxist approach that sees "fetishism" as just a matter of confusing one's individual perspective with the nature of a social totality cannot be applied, because the relevant social totality does not yet exist-in fact, that totality is in the process of being created by exactly such apparently "fetishistic" acts. 2 Viveiros de Castro bypasses all this and focuses, instead, on a single paragraph that appears toward the essay's end. It used to be, he notes, that anthropologists could declare their informants' understanding of the world to be wrong, as Evans-Pritchard, for example, did when he informed his reader that "Witches, as the Azande conceive them, cannot exist" (Evans-Pritchard 1937: 63). But one does not have to go back to Evans-Pritchard. Some still speak this way. As an illustration, he cites the following passage from my essay: Of course it would also be going too far to say that the fetishistic view is simply true: Lunkanka cannot really tie anyone's intestines into knots; Ravololona cannot really prevent hail from falling on anyone's crops. As I have remarked elsewhere, ultimately we are probably just dealing here with the paradox of power, power being something which exists only if other people think it does; a paradox that I have also argued lies also at the core of magic, which always seems to be surrounded by an aura of fraud, showmanship, and chicanery. But one could argue it is not just the paradox of power. It is also the paradox of creativity. (Graeber 2005: 430; quoted in Viveiros de Castro 2015: 15) This is his response: "It was already decided from the very beginning, " as Deleuze and Guattari might have said, that fetishes could serve only to represent necessary illusions conjured up by living in society. Marcio Goldman, in an article from which I stole this passage as well as the general spirit of the commentary, observes that Graeber's effort to save the Marxian notion of "fetishism, " namely, that fetishes are "objects which seem to take on human qualities which are, ultimately, really derived from the actors themselves, " is somewhat misplaced. Graeber does try somehow to reconcile the Merina to Marx, arguing that fetishes only become "dangerous" when "fetishism gives way to theology, the absolute assurance that the gods are real" (real as commodities, one might say). The problem, says Goldman (2009: 114ff), is that this brave effort to save the natives' [sic] face is undertaken behind the latter's back, so to speak. One wonders, firstly, if the conversion of fetishism into "a will to believe" that is at the root of (real, social) power would be accepted by the natives. And secondly, if such a reduction, which sounds more like an essay at reconciling one explicit Western ontology (to wit, dialectical materialism) with the Merina's implicit one, rather than an effort to problematize our own assumptions, does not end up, more than simply leaving untouched, reinforcing our own ontological framework. Magical power, as the Merina conceive it, cannot exist. (ibid.) What to say about this passage? Viveiros de Castro, following an earlier essay by Marcio Goldman (2009), appears to be making the following arguments: 1) that there are a people called "the Merina" to whom can be collectively ascribed a certain "implicit ontology, " which includes a certain conception of magical power; 2) that I am denying the legitimacy of this "Merina" conception when I say Ravololona (a hail charm) can't "really" stop hail; 3) that I am substituting for the explicit Merina theory a different theory, derived from Marx, that holds such illusions to be projections of human qualities onto material objects; 4) that while I claim the "natives" tacitly recognize that this is happening, I am nonetheless trying to make my argument "behind the native's backs" by means of statements (such as #2), and theories of social power, that they would be unlikely to agree with; 5) that in doing so, I fail to problematize my own (Marxist) theoretical assumptions in response to that tacit "native" ontology of magical power.
Let me take the last of these, #5, first. Granted, it seems unreasonable to expect the author to have made a detailed study of my earlier ethnographic writings concerning Malagasy fanafody or "medicine" (e.g., Graeber 1996a: 15-19;1996b [see also 2007b: 226-34, 241-43 andpassim];2001: 108-14, 232-45;2005: 421-26;2007a: 35-39, 73-86, 139-82, 185-87, 232-36, 242-50, 261-308, 320-23, 338-47 and passim;2007b: 165, 195, 278-79;2012: 36-39). Still, since he has presumably read the paper he is critiquing, he must be aware it is explicitly an attempt to employ ethnography to problematize Marxist theoretical categories. The essay concludes by arguing that African "fetishes" are not fetishes in the Marxist sense at all; that a classic Marxist analysis of fetishism cannot be applied to any context involving dramatic social creativity; 3 and that, in such circumstances, what we call fetishism or even magic, can in a certain sense be said to be true. The essay even suggests, in the second half of the very paragraph Viveiros de Castro cites in his lecture, that what Marx would consider a free society would be at least in certain ways more fetishistic than our own! Presumably, then, Viveiros de Castro's objection is not that I fail to use the ethnographic material to problematize my theoretical assumptions, but rather that I fail to do so in the way he believes I should. What I should have done was examine "magical power, as the Merina conceive it, " and then treat it not as a theory or belief but as reality, one which "we Westerners" will never be able to completely understand, and one to which our own familiar categories like the fetish do not apply. In other words, there are only two permissible ways to "problematize our own assumptions": either one can accept and try to come to grips with the radical alterity of "native" concepts, and consider the implications of treating them as a form of reality (but a reality that exists only for this one particular group of "natives"), or one can come to accept the general theoretical framework promulgated by proponents of the "ontological turn. " And indeed it is true that I've done neither. Instead, I concluded that the examples of BaKongo nkisi and Malagasy ody can teach us something unexpected about humans everywhere: not just Malagasy farmers and astrologers, but "Western anthropologists, " Amazonian shamans, Egyptian shopkeepers, Mexican poets, and nineteenth-century German revolutionaries as well.
In this sense, the objection is not that I do not problematize my own assumptions; but that I problematize them a little too much.
This strikes me as important, and we might do well here to pause a moment and consider what's at stake before proceeding. We appear to be in the presence of two quite different conceptions of what anthropology is ultimately about. Are we unsettling our categories so as (1) to better understand the "radical alterity" of a specific group of people (whoever "we" are here taken to be); or (2) to show that in certain ways, at least, such alterity was not quite as radical as we thought, and we can put those apparently exotic concepts to work to reexamine our own everyday assumptions and to say something new about human beings in general? Obviously I am an exponent of the second position. In fact, it strikes me that the greatest achievements of anthropology have come precisely when we are willing to make that second move: to say, "But are we not all, in a certain sense, totemists?" "Is not war a form of ritual sacrifice?" "Does not knowledge of the logic of Polynesian taboo allow us to look at familiar categories like etiquette, or the sacred, in a different light?" I should emphasize: carrying out this sort of analysis is not simply a matter of "Westerners" exploiting "native" wisdom to better understand themselves. Admittedly, we live in a violently unequal world, and existing structures of power will often ensure that things will turn out that way. But this is true of any intellectual project conducted within structures of violent inequality (including projects for the recognition of radical alterity, which can easily slip into becoming charters for some sort of moral or political apartheid: see Leve 2011;Graeber 2007b: 288-90). Anything can be made to serve the purposes of power. The political question (at least for me) is: which is the approach best suited to support those who are trying to challenge those structures of power and authority, and in what ways? *** "Ethnographic theory, " as Giovanni da Col and I (2011) have termed it, is an attempt to make this latter approach explicit. There are, effectively, two steps involved, repeated endlessly ("recursively" as the OTers would put it): (1) an attempt to grapple with the internal logic of an apparently alien concept or form of practice (bearing in mind here that concepts are always the other side of a form of practice-i.e., numbers are not prior to but an effect of the practice of counting, taboo is an effect of the practice of tabooing, etc.); then, (2) an effort to reexamine less apparently exotic and more familiar practices in the light of this analysis to see if our common sense notions are in any sense partial, inadequate, or wrong. The history of the concept of fetishism is actually an excellent illustration of this. As William Pietz (1985Pietz ( , 1987Pietz ( , 1988 famously argued, European merchants operating in West Africa in the sixteenth and seventeenth centuries invented the word "fetish" because they lacked a language with which to talk about many of the practices they encountered among their African counterparts; once they had done so, the concept gave European thinkers a kind of conceptual pivot that allowed them to see some of their own familiar practices (commercial and sexual) in a radically new light. The resulting theories allowed others to return to the African material and discover the original conception of "fetishism" had been wildly inadequate, which in turn allows us to rethink our own theoretical assumptions about commodity "fetishism" . . . and so forth. My own essay was just another moment in that ongoing exchange.
Some proponents of OT are willing to state explicitly that one of the major advantages of their own approach is that it protects us from the discomfort of making that second move. Here, Martin Holbraad: A corollary of this is remarked upon more rarely and has to do with the way the ontological move actually protects both sides in the putative disagreement. Proponents of the move usually emphasize how it gets us out of the arrogance of thinking that the people we study are silly when they say and do things that to us seem irrational. But equally it gets us out of the relativist impulse to say that what we consider rational is "just as" open to question (equally "situated, " "constructed, " and so on). Our anthropological desire to give credit to those who seem to be saying (because we misunderstand them) that stones are people has no bearing on our own commonsense understanding that they are not: again, what counts as a stone in either case is different. The ontological turn, in other words, protects our "science" and our "common sense" as much as it protects the "native. " 4 (Holbraad in Alberti, Fowles, Holbraad, Marshall, and Witmore 2011: 903;italics mine) This passage is crucial because it lays bare the ultimately conservative nature of the ontological project-at least, in this particular iteration. Western science and common sense are "protected" from challenge-which of course, necessarily, also means the protection of those structures of authority that tell us that there is something that can be referred to as "Western science" or "common sense"-and what it consists of-in the first place. Since after all, if our interlocutor were, say, a Theravada Buddhist reformer (e.g., Leve 2011) or Naxalite revolutionary (e.g., Shah 2013Shah , 2014) who claimed to have a message for all humanity, the response would presumably be to tell her to pipe down and speak for herself. Any would-be Zarathustras will just have to go back up their mountains. The ontologist is effectively declaring: I will not challenge the authority of a Cuban diviner who tells me that "powder is power, " within that space I have allotted for Cuban diviners to speak with authority about such matters; but by the same token, that diviner has no business challenging a Western scientist operating within what I consider a Western scientist's appropriate sphere of authority. Neither, for that matter, shall I myself challenge any commonplace assumptions about the nature of time, objects, change, subjects, consciousness, creativity, or action that might be prevalent in that sphere I have defined as "the West, " for those people I define as "Westerners, " on the basis of anything the Cuban diviner might say.
In other words, the diviner cannot tell us anything about human beings in general; neither can the anthropologist. We must all leave the world, as Wittgenstein once said, precisely as we found it. *** Now let me turn to the particular sentence that Viveiros de Castro's critique turns on: "Ravololona cannot really prevent hail from falling on anyone's crops. " 5 It might help to explain here that "Ravololona" is the name of a famous Malagasy ody, or "charm. " A charm of that name was once part of the official pantheon of sampy, sometimes called "royal palladia" (Berg 1979), that protected the Merina kingdom in the nineteenth century; one manifestation of which was kept as a hail charm across the mountain from Betafo, the community where I did my doctoral fieldwork in Madagascar between 1989 and1991. 6 One finds it hard to imagine, Viveiros de Castro suggests, that "the Merina" would have agreed with such a statement. For this reason, my comment is the very definition of what should be an "illegal move" in anthropology, since I would appear to be appealing to a form of Western knowledge about reality, rooted in science, that makes universal claims and holds itself as necessarily superior to the understandings-or indeed, the realities-of those we study. 7 5. Lunkanka, which is also cited, is not in fact a Malagasy ody at all but a BaKongo nkisi that had been discussed in a long quote earlier in the essay (Graeber 2005: 417). The author mistakenly assumes both were Malagasy.
6. The sampy were officially destroyed on Queen Ranavalona II's conversion to Christianity in 1869; the word was chosen to translate "idol" in the Old Testament and fanampoana sampy ("serving the idols") became the word for "heathenism. " As a result no one nowadays would admit to having anything to do with sampy. Contemporary hail charms (ody havandra)-whose powers are by no means limited to hail-are, however, clearly latter-day versions of the same thing and often have the same names and powers. In the nineteenth century, Ravololona was one of the royal sampy, albeit a minor one (it doesn't make all the official lists); it was said to have escaped the royal purge and to have been taken off in secret to become the official guardian of a powerful ancestry called the Zanak' Antitra in the region of Arivonimamo, the same area where I did my fieldwork (Clark 1896: 455-56;Renel 1915: 142, 158-59;Domenichini 1985: 694-96). There were any number of incarnations of Ravololona in the region when I was there, as there were too of its rival charm, Ravatomaina. The story I was told in Betafo was that the current holder of Ravatomaina had chased Ravalolona over a nearby mountain to a town called Ambatomivolana; to this day, the two rival charms try to push the hail onto one another's fields (see Graeber 2007a: 277-79, and my discussion of "The weather as a domain of political struggle, " ibid.: 282-92).
7. Admittedly, the term is used with a certain ambiguity: Viveiros de Castro is ostensibly saying that if OT is a Wittgensteinian "language game, " this would be an illegal move.
There are a number of problems with this line of critique. One is the very existence of a group of people who can be referred to as "the Merina. " In the piece I actually carefully avoid using the word "Merina" in this way. 8 There's a simple reason. While "Merina" does seem to have been sometimes used in the nineteenth and early twentieth centuries as a generic term for the inhabitants of the northern part of the central plateau of Madagascar, and has since become established in the anthropological literature, not a single person I encountered during my fieldwork ever referred to him-or herself as "Merina. " They called themselves all sorts of other things: by their status group (andriana, hova, or mainty), geographically ("people from here in the center of the country . . . "), or many other ways beside. If they were speaking about fanafody or medicine, they almost invariably referred to themselves simply as "Malagasy, " as if to suggest that in this context, social or geographical differences were pretty much irrelevant.
It made sense that they should do so, as ideas and practices concerning fanafody were indeed largely uniform across the island. But it raises some sticky questions for the ontologist. Medicine is always treated as a pan-Malagasy phenomenon; if there was a tacit ontology underlying it, presumably, it must be island-wide. Ideas about ancestors, on the other hand, varied considerably in different parts of the island. Does this mean reality was layered? And if someone living in Betafo might be in a different reality from someone in Tulear when dealing with ancestors, but in the same reality when it came to fanafody, then why can't both of them be in the same reality as New Yorkers or Londoners in some third way, say, when it comes to epidemiology? This is not just idle musing. It's directly relevant to the concept of fanafody, which was also used to refer to the kind of medicine one might have prescribed at a clinic or pharmaceutical dispensary. It was commonplace to juxtapose anything considered "Malagasy" with something else, that was considered "Vazaha"-a word which can be translated "French, " "of European stock, " or simply "foreign. " There were both Malagasy and Vazaha ways of doing most anything, from eating breakfast to engaging in political debate. This was true of medicine as well. But it's important to emphasize that this habit is not just a product of colonialism. Madagascar was, from its initial settlement, a center of trade and migration, and there is reason to believe that the habit of juxtaposing "Malagasy" and foreign ways of doing things goes back to long before the colonial period, perhaps, even to the earliest days of human occupation (Graeber 2013a)-though presumably, at first, the paradigmatic foreigners were not Vazaha but Silamo ("Muslims").
What I'm getting at here is that it's simply impossible to think of "magical power, as the Merina conceive it"-or even as Malagasy would conceive it-as existing in any sort of conceptual bubble, in which those ideas define their own reality. Fanafody has always been a form of engagement with a larger world. This is in part because it has always been seen as somehow quintessentially Malagasy and But since I am obviously not playing that particular language game, and he is still criticizing me, he presumably holds those standards should apply to all ethnographers. 8. I do use the term as an adjective, especially when referring to the nineteenth century. I also refer once to "Imerina, " the territory of the nineteenth-century Merina kingdom.
In what follows I will use "Merina" mainly for the nineteenth century kingdom.
defined against the outside world; but it is also because, despite that, it has continually incorporated foreign techniques, objects, and ideas. In the seventeenth century, fanafody often seems to have involved bits of Arabic writing. In the eighteenth and nineteenth centuries, at the height of the slave trade, charms were typically composed of two elements: rare woods, and trade beads or silver ornaments (the latter melted down from Maria Theresa thalers or similar coins). Both the beads and silver originally came into Madagascar as foreign money (Edmunds 1896;Bernard-Thierry 1946;Bloch 1990;Graeber 1996: 141.). This sense of dialogic confrontation inherent to the very constitution of fanafody was also reflected by the way people talked about it. Ways of talking about medicine are-and by all accounts have always been 9 -marked by an endless diversity of often contradictory perspectives, including expressions of dramatic skepticism. These contradictions are not incidental but constitutive of the nature of fanafody itself. *** This, in turn, leads to the most important point of all. Would a Malagasy informant object to the statement "Ravololona cannot really prevent hail from falling on anyone's crops"? As someone who spent over a year living in a community once protected by a charm called Ravololona, and with neighbors that still were, I can assure the reader: people said things like that all the time. Of course it all depends on who you ask. Many inhabitants of Betafo were quite insistent that Ravololona could not prevent the hail under any circumstances, it was simply a fraud-and so, for that matter, was their own local hail charm, Ravatomaina, owned by a very ancient and venerable but highly controversial astrologer named Ratsizafy. Most of them were careful to add there were other hail charms that could stop hail, or that they were pretty sure there probably were. But a few denied the efficacy of hail charms of any sort. Arguments about the efficacy of one or another sort of fanafody, or of fanafody in general, were, in fact, so common I would even call them a popular form of entertainment-not as popular, perhaps, as arguments about money or complex polyamorous relationships, but popular nonetheless. In other words, my statement was not some kind of high-handed dismissal of conceptions held uniformly by some people called "the Merina, " it was an intervention in an ongoing Malagasy conversation. If it came off as slightly cavalier, it was only because I identified so thoroughly with my informants that I felt I could express myself as one of them might have done. 9. Obviously our information on such matters is quite limited from earlier periods but the early-nineteenth-century Merina king Radama I was a notorious skeptic who is reported to have told foreign visitors he felt all aspects of fanafody, particularly the royal sampy, to have been simple frauds, noting that religion itself was, in his words, "a political institution" (e.g., Copalle 1827). He was also famous for making up tests of his sampy's powers, say, by hiding an object in his palace and challenging their keepers to find them (e.g., Ellis 1838: 408, 411-12, etc.;Callet 1908Callet : 1104. Needless to say, while observers at the time noted that the keepers always failed his tests, later oral traditions-some of which I heard myself-inverted the stories and insist the most powerful ones miraculously confounded his attempts to expose them. What's more, the existence of such arguments was the very starting point of my original analysis. Because this was one of the things that most surprised me when I started doing fieldwork; something I did not anticipate, and that did indeed unsettle my working assumptions. I went to Madagascar expecting to encounter something much like a different ontology, a set of fundamentally different ideas about how the world worked; what I encountered instead were people who admitted they did not really understand what was going on with fanafody, who said wildly different, and often contradictory, things about it, but who were all in agreement that most practitioners were liars, cheats, or frauds. Coming back from the field, I consulted with colleagues who had been in similar situations (in the Andes, Andaman Islands, Papua New Guinea . . . ) and discovered that such sentiments are actually quite commonplace. They also confessed they never knew quite what to do with them. And in fact, this is precisely the aspect of magical practice that is most often dismissed as unimportant, or simply left out of ethnographic accounts.
So I decided to take my informants seriously, and by doing so, to rethink my theoretical assumptions.
As I point out in the passage cited by Viveiros de Castro, the essay on fetishism is an extension of an earlier argument: that of the last chapter of my book Toward an anthropological theory of value (2001). 10 Now, I feel a bit awkward quoting myself, but in this instance it seems that I will have to. Here is how the chain of argument that led to the conclusions Viveiros de Castro cites originally began: Anthropologists usually acknowledge this sort of skepticism-the aura of at least potential disbelief that always seems to surround the sort of phenomena that gets labeled "magic"-but almost always, only to immediately dismiss it as unimportant. Evans-Pritchard, for instance, noted that most of Zande he knew insisted that the majority of witchdoctors were frauds and that there were only a handful of "reliable practitioners. " "Hence in the case of any particular witchdoctor they are never quite certain whether reliance can be placed on his statements or not" (1937: 276). Similar things have been reported about curers almost everywhere. But the conclusion is always the same: since everyone, or most everyone, agrees there are some legitimate practitioners, the skepticism is unimportant. Similarly with the tricks, illusions, and sleights of hand used by magical performers like shamans or mediums (pretending to suck objects out of people's bodies, throwing voices, eating glass). The classic text here being of course Levi-Strauss' "The sorcerer and his magic" (1958), about a young Kwakiutl man who learned shamanic techniques in order to expose their practitioners as frauds, but who ended up becoming a successful curer anyway. The point is always that while curers (for instance) can hardly help but know that much of what they are doing is stage illusion, they also think that since it does cure people, on some level it must be true. So again, the tricks are of no significance. Now there are good historical reasons why anthropologists have tended to take this attitude-the existence of missionaries being only the most obvious-but what if we were to turn things around and consider this skepticism as interesting in itself? Take attitudes toward 10. In fact it was originally written to be part of that chapter, but had to be cut for space. 2015 | Hau: Journal of Ethnographic Theory 5 (2): 1-41 curers. Evans-Pritchard says that at Zande seances, no one in the audience "was quite certain" whether or not the curer they were watching was a charlatan; I found this to be equally true in Madagascar. People tended to change their minds about particular curers all the time. But consider what this means. Curers, genuine or not, are clearly powerful and influential people. It means anyone watching a performance was aware that the person in front of them might be one whose power was based only on their ability to convince others that they had it. And that, it seems to me, opens the way for some possibly profound insights into the nature of social power. (Graeber 2001: 243-44) In other words, far from arrogantly discounting what my informants told me, I was trying to take those informants seriously, even when they were making the sort of statements that other ethnographers ordinarily dismiss as unimportant, or outright ignore.
Neither was the assertion that medicine only operates by convincing others that it is effective just something I extrapolated from doubts about individual healers.
Most people I knew in Madagascar considered it a matter of common sense that if a person really didn't believe in medicine, it wouldn't work on them. Very early on, for instance, I heard a story about an Italian priest sent there to take up a parish who, on his first day in the country, was invited to dinner by a wealthy Malagasy family. In the middle of the meal, everyone suddenly passed out. A few minutes later two burglars strolled in through the front door, and then, realizing someone was still awake, ran out again in fear. It turned out they had planted an ody in the house timed to make everyone in it fall asleep at six P.M. but since the priest was a foreigner who didn't believe in that kind of nonsense, it had no effect on him.
That much was common knowledge. Several people went even further and insisted that even if someone was using medicine to attack you, it wouldn't work unless you knew they were doing it. Now, the first time I heard this it was from fairly well educated people and I strongly suspected they were just telling me what they thought I wanted to hear. After all, it almost precisely describes the attitude of most people in America: that if magic does work, it is purely by power of suggestion. But as time went on, I met a number of astrologers and curers, people who had next to no formal schooling and clearly would have had no idea what Americans were supposed to think (one of them was actually convinced I was African), who told me exactly the same thing. And just about anybody would agree with this if you asked them in the abstract. Usually they would then immediately begin to offer all sorts of qualifications-yes, it was true, unless, of course, it was something they' d put in your food. Or unless it was one of those really powerful love charms. Or unless . . .
The bizarre thing is that this principle was utterly, completely, contradicted by practice. Everyone would agree to it, but no one ever acted as if it were true. If you got sick, you went to a curer. The curer would usually tell you that your illness was caused by someone using medicine of some kind and then, reveal who it was and how they' d done it. Obviously, if medicine can harm you only if you know someone is using it on you, the whole procedure would make no sense. In fact, the theory contradicts practice on almost every level. But if no one ever acts as if it were true, why did the theory even exist? (Graeber 2001: 244-45) As I mentioned, people discussed and argued about such matters all the time-not just about fanafody, but anything having to do with spirits, ancestors, or the general category of things that operated by imperceptible means (zavatra manan-kasina). It was precisely these conversations that led me to develop the notion of the paradox of creativity, and hence of politics. 11 Such conversations knew endlessly subtleties, but speaking very broadly, they tended to take one of two directions. Either one started by asserting that what we might call magical powers did exist, but then immediately began qualifying that most of the concrete examples they actually knew about were probably simply social phenomenon (since the astrologers, or mediums, in question were imposters). Or one began by asserting that magical power was purely social in nature-and then, immediately began qualifying that, by noting certain types of fanafody that actually did seem to work whether you thought they should or not and even, in some cases, became particularly irritated at skeptics and punished them in horrible ways. In either case, I came to realize, "you have the same uncomfortable relation between two premises that are pretty clearly contradictory, yet in practice seem to depend on one another" (2001: 245). For instance, the premise that harmful medicine only affects you if you believe in it can only be true if most people think it isn't true-since, obviously no one actually desires to be harmed by evil medicine. Similarly, the opposite premise, that spirits will punish those who scoff at them, obviously depends on the existence of skeptics.
Most were quite aware of these paradoxes as well, and played around with them in endless ways. A teenage sister and brother, Nivo and Narcisse, whose parents had moved from the city to the countryside, once explained to me that as soon as they arrived in the village, their neighbors started using harmful medicine to try to cause them to fall ill, just so they would be forced to submit themselves to local healers who also happened to be figures of political authority. "Oh course it didn't work on me, " Narcisse assured me, "I don't believe in that sort of nonsense. " His sister looked slightly annoyed. "Well, I thought I didn't believe in it either, " she said, "But I guess I must believe in it, because ever since I got here, I just keep getting sick all the time. " Most ethnographers have simply ignored such conundrums, or at best treated the skeptical discourse was somehow extraneous, foreign, a product of "Western" education, or otherwise as dross pasted over the real stuff (that is, whatever seems to most fly in the face of "Western common sense"). But in this case, the tension of the two contradictory perspectives pulling at one another is precisely what is constitutive of the world of fanafody, and everything associated with it. What's more-and this is an argument that I cannot really develop here, but it's important-political power was treated in eighteenth-and nineteenth-century Merina ritual in much the same way. The powerful ody that protected the kingdom were similarly paradoxical: they were created by rituals which posited that they were both products of collective agreement, and autonomous powers in their own right. 12 But so were kings. The Merina monarch was treated effectively as a kind of ody, and as such, both as a being created (and continually recreated) by the people through conscious acts of agreement, 13 and as something prior to the very existence of the people, alien and incomprehensible-both at the same time.
This, too, lead to endless unresolvable arguments: e.g., myths that claimed the ruling dynasty descended from heaven were always balanced by proverbs such as "kings did not really come down from the sky" (Graeber 2001: 237-38). And here too, the paradoxes were not incidental, but constitutive of the object; even Malagasy myths about the origins of life and death, which are surely seen as conveying important truths about the human condition, tend to end with the tag-line, "it is not I who lie; these lies come from ancient times. " Now, of course, the OTer might still object: perhaps what you say about fanafody is true on a certain level of practice. But is this not all premised on the possible existence of certain forms of power fundamentally different from those allowed for in the ethnographer's commonsense universe, and therefore, a certain tacit ontology alien to our own? I would reply that this all turns on what one actually means by "ontology. " The meaning of the term is in no way self-evident. Many anthropologists have come to use it very loosely, as little more than a synonym for "culture" or "cosmology. " OTers have something much more specific in mind. Before responding, then, it well be necessary to delve a little more deeply into what that actually is.

Ontology, epistemology, and other mooted terms
One thing is abundantly clear: when proponents of the ontological turn in anthropology use the word "ontology, " they mean something very different from what philosophers have traditionally meant by the term. 12. One typical missionary source: "until the consecration service had been held, and the pledge of allegiance given, the charm, although finished in regard to its construction and general characteristics, was just a piece of wood to them" (Edmunds 1897: 62; italics mine). Many other sources confirm it is collective agreement that gives power to the charm. However, other stories equally insist that the spirit of certain charms came to their future owner in dreams and visions and caused him to "discover" them (e.g., Domenichini 1985).
13. For instance, any time subjects came to an official agreement or registered an official contract of any kind, they always had to give a small piece of silver to the king; this was referred to as manasin'Andriana, which in this context is best translated, "giving power to the king, " In the annual Royal Bath ceremony, such "gifts of power" in the form of silver coins were repeated by the entire kingdom, and the king went through a ritual process that precisely replicated the creation of a powerful ody or charm (see Graeber 1996: 15-19;2007a: 35-39).
"Ontology", like "epistemology, " or "semiology, " are words of relatively recent coinage. 14 Still the broad conceptual divisions they represent can be traced back to the very origins of Greek philosophy. As a handy mnemonic, I might here make appeal to the three premises put forward by the Sophist Gorgias of Leontini, a contemporary of Socrates, which together comprised the whole of his philosophy: 1) Nothing exists; 2) If it did exist, it could not be known; 3) If it could be known, it could not be communicated. 15 Now, at first glance, these three premises might seem to deny the very existence of (1) ontology, (2) epistemology, and (3) semiology-or as most now prefer to call it, semiotics. But in fact this is not the case. This is because "ontology" is not a word for "being, " "way of being, " or "mode of existence, " but refers instead to a discourse (logos) about the nature of being (or alternately, about its essence, or about being as such, or in itself, or about the basic building blocks of reality . . . the only really important word at this initial juncture is "about"). Therefore, "nothing exists" is an ontological statement. Similarly, "if it did exist, it could not be known" is an epistemological statement, if obviously a rather minimal one: since epistemology is not knowledge of the world but rather, a discourse concerning the nature and possibility of knowledge about the world. (Note too that such knowledge presupposes, as Gorgias was aware, the existence of a world that knowledge can be about. You can't have knowledge of something that isn't there-other than the knowledge that it isn't. Gorgias is just adding that you can't have knowledge of something that is there either [not that anything is there].) Finally, semiotics is not communication, but the study of communication, or more broadly a discourse about the nature and possibility of communication, 16 and therefore presumes that there's something to be communicated.
In contrast, when OTers deploy these words, they seem to mean something quite different. To propose an initial approximation: ontology corresponds to "way of being" or "manner of being, " epistemology "way or manner of knowing, " and semiotics, if the term is used it at all (it has become unfashionable), as "way or manner of communicating. " Now there's nothing wrong with using words in a new way, but if one does so, and does not make it clear how one's new usage differs from the more traditional one, one is likely to cause confusion.
14. Ontology is usually traced back to a German philosopher named Jacob Lorhard in 1606. Epistemology was introduced much later, by the Scottish philosopher James Frederick Ferrier in 1854. Semiotics in mentioned in Locke but only really comes into common usage as a legacy from C. S. Peirce's works of the 1860s onwards, and semiology even later, from Saussure in the 1920s.

Gorgias did apparently write a book
Concerning what is not, or, on nature. It has not survived. For a good summation of his arguments see Barnes 1979: 136-37. 16. I was about to write "of human communication, " but in fact C. S. Peirce, who invented the term, believed that communication took place on all levels of physical reality, and that terms like "iconism" or "indexicality" could even be applied to the workings of atomic particles.
Many of what are now considered the founding texts of OT do seem to be trying their best to avoid such confusion. "Since these terms-'epistemology' and 'ontology'-are much used and abused in present day discourse, " note Henare, Holbraad, and Wastel in what is generally considered the most important founding text of all, the introduction to Thinking through things (2006: 8), "it is important to be quite explicit about what work we want them to do for our argument. " But it's not clear that they entirely succeed. It might be helpful, I think, to look at this essay in particular, to make clear the kind of slippage these terms undergo.
The authors' central argument is that preceding decades had begun to see a broad-and hitherto unacknowledged-shift (or "turn") in anthropological theory "from questions of knowledge and epistemology towards those of ontology" (2006: 8). 17 Previous anthropology, they note, like most social sciences, saw itself as a form of knowledge, and consequently, saw its mission as a matter of understanding and accounting for the forms of knowledge of those it studied (their cultures, symbolic systems, or worldviews). What this tended to mean in practice was imposing some theoretical model (Structuralism, Hermeneutics, Dialectical Materialism . . . ) as the framework for understanding what Malinowski originally called "the natives' point of view. " Yet it has become increasingly clear this was a trap. It's only by moving away from this "epistemological orientation" towards an "ontological orientation" that we can allow our informants to set the terms, even if it means "unsettling" our own theoretical assumptions of what it is possible to say about them. This is an admirable aim, and certainly the idea that an ontological approach would mean taking one's informants more seriously as interlocutors is the heart of its appeal. For the moment, though, I mainly want to draw attention to what is happening to the philosophical terms. The authors cite, as inspiration, a series of now-famous lectures delivered almost a decade before in Cambridge by Viveiros de Castro himself: Anthropology seems to believe that its paramount task is to explain how it comes to know (to represent) its object-an object also defined as knowledge (or representation). Is it possible to know it? Is it decent to know it? Do we really know it, or do we only see ourselves in a mirror? (Viveiros de Castro 1998: 92) Radical alterity is just another way of saying "reality" by "thought. " The simplification of ontology accordingly led to an enormous complication of epistemology. After objects or things were pacified, retreating to an exterior, silent and uniform world of "Nature, " subjects began to proliferate and to chatter endlessly: transcendental Egos, legislative Understandings, philosophies of language, theories of mind, social representations, logic of the signifier, webs of signification, discursive practices, politics of knowledge-you name it. (ibid.) It seems to me that Viveiros de Castro's assessment here is substantially correct as well. Obviously, the soul/body, mind/matter division was hardly the brainchild of Descartes; it goes back at least to Pythagoras. But Descartes introduced a much more radical version of the dichotomy, largely, I would argue, by eliminating the old Stoic/Neoplatonist category of imagination, which for the Scholastics had served as a quasi-material intermediary between the two. 18 As a result, philosophy did turn away from questions about the nature of the world, which were increasingly relegated to science, and toward questions about the possibility of knowledge. Humean skepticism, and Kant's apriorist response, were obviously crucial turning points in this respect.
Viveiros de Castro goes on to argue that as a result, social sciences have tended to focus on questions of mind over body, intellect over lived reality. This is a somewhat tougher case to make (there's an awful lot of resolutely materialist social science) but surely there are strong currents pulling in this direction. What I want to emphasize here though is that as he makes the argument, one can already observe the term "epistemology" shifting from its classic philosophical meaning ("questions about the nature or possibility of knowledge") to "questions of knowledge, " and then to simply "knowledge. " Structuralism itself, to take one fairly random example, is hardly a form of "epistemology. " It might have involved an epistemology, a theory of the nature of knowledge, but when Claude Lévi-Strauss (1958) proposed a structural analysis of the Oedipus myth as a story about eyes and feet, he was in no sense elaborating on that theory. He was simply applying it, engaging in that sort of social science one would engage in if one assumed that theory was true. 19 Henare , Holbraad, and Wastel (2006: 9) go on to argue: The assumption, then, has always been that anthropology is an epistemeindeed, the episteme of others' epistemes, which we call cultures (cf. Wagner 1981;Strathern 1990). The inveteracy of this assumption, argued Viveiros de Castro, is owed to the fact that it is a direct corollary of "our" ontology-the ontology of modern Euro-Americans, that is.
18. See for instance Graeber 2007b: 66-69. I doubt many OTers share this particular diagnosis, by the way. Most seem to reject imagination as yet another subset of what they call "epistemology, " though I myself would argue that this is only true of what I've elsewhere called "transcendent" as opposed to "immanent" imagination, the latter being an element in all forms of action. Their ultimate solution, that of ethnographic creativity, seems to me to be simply a return of that imaginative project under another name.
19. Thus when Henare, Holbraad, and Wastel do define epistemology, it is as "the various systematic formulations of knowledge" (2006: 9), that is, as any knowledge informed by theory of any kind.
And the problem with that, in turn, is that since this Euro-American ontology assumes that there is one real world, one nature-the one revealed by Western science-it also assumes that difference can only be a matter of different perspectives on, or different ways of perceiving, knowing, or representing that single reality. This leads to a bifurcation within the sciences. "Natural" science is dedicated to unveiling the uniform laws that govern that undifferentiated reality; "social" science is the study of different ways different people think about or represent it. These formulations involve a curious effacement of the domain of action. Surely social scientists do not simply study how people perceive, know, or represent the world; they also study how they interact with it, shape it and are shaped by it-not to mention, how they act on one another. But framing things in this way would make it much more difficult to maintain the conceptual clarity of the argument. 20 Instead, the authors conclude that what's needed is not to examine how human projects of action, or for that matter, non-human projects, problematize these divisions (body/mind, nature/culture, material/ideal, etc.) but rather, to rethink the very idea that one can speak of a single, undifferentiated, natural world at all. Our insistence on the unity of nature (and therefore, as a corollary, our assumption that all difference can only be cultural) is, they say, a product of our own Western, dualist ontology. We should not impose it on others. In fact we should not even impose it on ourselves-at least, when we are thinking about others. In the presence of genuine alterity, we must speak not of people who have radically different beliefs about, or perceptions of, a single shared world, but of people who literally inhabit different worlds. We must accept the existence of "multiple ontologies. " Note here how in the course of this argument, the meaning of "ontology" has also undergone profound changes. After all, if "ontology" simply means a discourse about "the nature of being in itself, " one could hardly assert that Western philosophy is particularly monolithic: most philosophers considered "great" are considered great largely because they came up with a different ontology, and even OTers draw much of their conception of what a non-dualist ontology might be like from the work of Gilles Deleuze, who never claimed to be doing anything more than writing his own creative synthesis of ideas derived from such post-Cartesian philosophers as Leibniz, Spinoza, Nietzsche, Bergson, and Whitehead. So "ontology" drifts from being an explicit form of philosophical discourse to referring to the-largely tacitset of assumptions underlying the practice of natural and social science (which do tend to remain stubbornly fixed, whatever philosophers say about them), and from there, to being the tacit assumptions underlying any set of practices or modes of being of any kind at all.
What happens, then, to the older philosophical conceptions-lets call them Ontology 1 , Epistemology 1 , and Semiotics 1 , so as to distinguish them from the new OT usages, which we can refer to as Ontology 2 and Epistemology 2 -under this new dispensation? Well, if Epistemology 2 really just refers, as Henare, Holbraad, and Wastel (2006: 9) claim, to "systematic formulations of knowledge, " then it follows that all branches of philosophy, including Ontology 1 , Epistemology 1 , and Semiotics 1 , are simply different forms of Epistemology 2 -and therefore, precisely what OT thinkers propose to move away from. In which case, would not Ontology 2 have to refer (just by process of elimination) to tacit assumptions about the nature of being "in itself " and the forms of action and modes of experience these make possible (or, possibly, too, to the anthropologists' explicit theories about such tacit assumptions)?
This would appear to be the case. But that raises another problem: What, in that case, would "in itself " mean? Consider here the following definition, which I must emphasize comes from someone I consider to be an unusually subtle and philosophically sophisticated OT thinker: "Ontology-the investigation and theorization of diverse experiences and understandings of the nature of being itself " (Scott 2013: 859) Let's unpack this. So: ontology 21 begins as a mode of academic theory-making, a form of discourse, but its object is not discourse (since that, presumably would be Epistemology 2 ) but "experiences and understandings of the nature of being itself. " "Understanding" sounds a lot like knowledge, but let's say for the sake of argument that we are speaking of the tacit understandings underlying certain forms of "experience. " Arguably this might escape the charge of Epistemology 2 . But that leads to the question: How exactly is it possible to have an experience of "the nature of being itself "? One can certainly have experience of specific manifestations of being (toothpicks, oceans, bad music coming from a party upstairs . . . ). But normally that's just called "experience. " Perhaps a mystical experience, such as might have been had by Jalal al-din al-Rumi or Meister Eckhart, might qualify as an experience of "the nature of being itself "? But presumably, this is not the sort of thing the author is talking about either. It only really makes sense if "being itself " is simply whatever "understandings" people might be said to have of it. In which case all "itself " is really doing here is pointing to that familiar anthropological object, the tacit assumptions about the nature of time, space, action, personhood, and so on, that underlie what used to be called a particular cultural universe-just, now constructed as an "as if, " the sort of Ontology 1 one imagines the people one is studying would construct, were they the sort of people who spent their time engaging in speculative philosophy.
If so, the meaning has changed little since Irving Hallowell first introduced the word in his essay "Objiway ontology, behavior, and world-view" back in 1960. 22 What's changed is not the quest for underlying assumptions, but the larger significance being ascribed to them. What OTers are arguing, unless I very much misunderstand them, is that when in the presence of assumptions, or as they put it, "con-21. Writing either Ontology 1 or Ontology 2 would seem inappropriate here, since the author seems to be trying to formulate a synthesis between them.
22. Hallowell was not quite the first. There was something of a spate of references to ontology around the time Hallowell was writing. As far as I can make out, the person really responsible for introducing the term "ontology" into Anthropology was Ethel Albert, an analytic philosopher working with the Harvard Values Project directed by Clyde Kluckhohn (e.g., Albert 1956). She proposed dividing the underlying principles of any culture order into Metaphysics (consisting of Ontologies and Cosmologies), Epistemologies, Psychologies, and Values, and many working within or influenced by the Values Project adopted variations on this approach (Albert & Kluckhohn 1959). It seems unlikely Hallowell was unaware of this.
ceptions" that are sufficiently foreign to the ethnographer's own (e.g., that stones are persons, or powder is power), the ethnographer must act as if those conceptions are-for the speakers, and anyone presumed to share their Ontology 2 -constitutive of reality, and therefore of nature, itself. This "as if " is crucial. Saying there are "many natures" might seem like a very radical claim. But no one is actually arguing that there are parts of the world where water runs uphill, there are three-headed flying monkeys, or pi calculates to 3.15. They are not even suggesting there are really places where tapirs live in villages-at least, if "really" means it would make sense to say tapirs live in villages even in a world where there had never been Amerindians who said they did. Each different nature, then, can only exist in relation to a specific group of human beings sharing the same Ontology 2 .
This at least saves the formulation from obvious absurdity. But even here, the language often seems to slip back and forth between the subjunctive "as if " and simple declarative. Here is Henare, Holbraad, and Wastel (2006: 14, italics in the original) defending their claims that, say, Cuban Ifá diviners exist in a different "world" against the obvious objection: If things really are different, as we argue, then why do they seem the same? If "different worlds" reside in things, so to speak, then how could we have missed them for so long? Why, when we look at Cuban diviners' powders, do we see just that-powder? . . . [Because] the very notion of perception simply reiterates the distinction that "different worlds" collapses. The point about different worlds is that they cannot be "seen" in a visualist sense. They are, as it were, a-visible. In other words, collapsing the distinction between concepts and things (appearance and reality) forces us to conceive of a different mode of disclosure altogether.
At first glance, this seems to make no kind of sense. If one dissolves away the distinction between appearance and reality as so much false Cartesian dualism, shouldn't that mean that things are what they appear to be, and therefore, that things that look the same are the same and that's pretty much that? But what the authors are really saying is very different: that we shouldn't pay too much attention to what things look like, but should instead listen to what people say. Moreover, [authoritative] statements must be treated as a window onto "concepts, " and concepts treated-through a form of "radical constructivism"-as if they were themselves realities of the same ontological standing as "things, " or indeed, constitutive of the world itself. 23 23. Just in case the reader thinks I am exaggerating: Though Foucault would say that discourse creates its objects, he still works from the presumption that there is some real-world fodder out there. For example, while a body may not be male or female until a discourse of gender invokes this as an operative distinction, there is still a body to which the discourse refers. By contrast, what is advanced here is, if you like, an entirely different kind of constructivism-a radical constructivism not dissimilar to that envisaged by Deleuze. . . . Discourse can have effects not because it "over-determines reality, " but because no ontological distinction between "discourse" and "reality" pertains in the first place. In other words, The "ontological turn, " then, involves not only abandoning the project of ontology 1 , but adopting a tacit ontology which seems indistinguishable from classical philosophical Idealism. 24 Ideas generate realities. One could go even further. What they seem to be proposing is abandoning the entire project of philosophy (or at least, philosophy in anything like the forms it has historically taken in Europe, India, China, or the Islamic World.) Science, in contrast, would be preserved, but as the special property of "Westerners" or "Euro-Americans" 25 -which if taken seriously, would amount to one of the greatest acts of intellectual theft in human history, since after all, much of what underlies what we now call "Western science" was actually developed in places like Persia, Bengal, and China, and in (dare I say?) the real world. Most scientific research is no longer being conducted by Euro-Americans at all. *** I know this is a bit unfair. Such proposals are not really meant to be taken in this kind of programmatic way. More than anything else, OT is a theoretical framework designed to open space in order to engage in a particular form of ethnographic practice. And this form of practice is not without its merits. Having said much that is critical, let me end, then, on a positive note. I think the real strength of OT lies in the fact that it encourages what might be called a stance of creative respect towards the object of ethnographic inquiry. By this I mean first of all that it starts from the assumption that since the worlds we are studying cannot be entirely known, what we are really in the presence of is-as Viveiros de Castro (2015: 13) puts it, borrowing language from Deleuze, "the possibility, the threat or promise of another world contained in the 'face/gaze of the other, '" a possibility that can only be realized through the ethnographer, even as the ethnographer, in trying to describe-let alone explain-this other world, inevitably betrays that promise, or, as he puts it, "dissipates its structure, " at least to a certain extent. Yet despite the inevitability of betrayal, the task of the ethnographer is nonetheless to try to keep that possibility alive. Radical alterity can never be contained by our descriptions, the concepts can bring about things because concepts and things just are one and the same" (Henare, Holbraad, Wastel 2006: 13).
Apparently there's virtually nothing, no matter how obviously crazy, a contemporary academic can't get away with if they find some way to attribute it to Gilles Deleuze. (And in this case the authors themselves admit the link is fairly tenuous.) 24. OTers will no doubt object that this is unfair, since they are trying to dissolve away the very dualism that makes an opposition between Materialism and Idealism possible: but pretty much everyone claims that nowadays. The question is whether they are trying to dissolve Materialism into Idealism, Idealism into Materialism, or both Materialism and Idealism into something else. All evidence points to the first. For example, in Henare, Holbraad, and Wastel's text (ibid.), words like "material" or "physical" regularly appear in scare-quotes, but words like "concept" (aka "idea") or "conception" never do.
25. I hesitate to state the obvious, but the word "Westerners" is largely used, nowadays, as a euphemism for "white people. " Obviously OTers are no more guilty of this than anyone else but it needs to be remarked on.
argument goes, and we cannot understand it through deductive reasoning; rather, the ethnographer's task is a creative, experimental, even poetic project-an attempt to give life to an alien reality that unsettles our basic assumptions about what could exist. Insofar as there is a war going on here, it is a war the ethnographer should never win.

What if the world did exist but we just couldn't prove it?
If the greatest strength of OT is its willingness to embrace the limits of human knowledge (that is, as a form of Epistemology 1 ); its greatest flaw, to my mind at least, is that it doesn't take this principle nearly far enough. Radical alterity applies only to relations between cultural worlds. There is never any sense that people existing inside other Ontologies 2 have any trouble understanding each other, let alone the world around them; rather, out of respect for their otherness, we are obliged to act as if their command of their environment were so absolute that there were no difference whatever between their ideas about, say, trees, and trees themselves. It strikes me that by doing so, and especially, by framing this attitude as an ethical imperative, OT makes it effectively impossible for us to recognize one of the most important things all humans really do have in common: the fact that we all have to come to grips, to one degree or another, with what we cannot know.
In philosophical terms, what OT is proposing is simply an anthropological variation of the transcendental method, an exercise that sets out to deduce the "conditions of possibility" for human experience: essentially, to ask, what would have to be true in order for experience to be possible? 26 Immanuel Kant most famously used the transcendental method to produce his list of a priori conceptual categories of thought (the opposition of unity and plurality; the notion of cause and effect, etc.), along with such basic frameworks as the notion of time as a relation of past, present, and future. 27 All these, he argued, could not be derived from experience, since they already had to be present, in the mind, for us to experience anything the way we do at all. For Kant, these were not ontological categories. Kant rejected the very possibility of Ontology 1 , as he did not believe we could say anything about the nature of things in themselves. Now, there's always been a strain in anthropology that has sought to apply a similar analysis to particular social or cultural forms of experience, and thus, to seek to find cultural categories using the same approach Kant used for conceptual categories. After all, this is very close to what ethnographers invariably doi.e., ask, "what would people have to think in order for all these statements to be true?" Often this leads anthropologists to conflate the two, which is a problem, because Kantian categories and cultural categories are in no sense the same thing. Or even the same sort of thing. A typical (and I've always felt, slightly embarrassing) early example is Durkheim and Mauss' essay Primitive classification ([1903] 1963cf. Schmaus 2004), which argues that Kantian categories are best viewed not as prior to experience but as the products of social organization, and therefore different in differently organized societies-conflating, in this case, the arrangement of time into a particular sequence (e.g., summer, fall, winter . . . ), with the very notion that it should be possible to arrange anything in a temporal sequence of any kind at all. 28 This is obviously a basic category error, as generations of first year graduate students forced to read the essay have regularly pointed out. Yet the temptation to make similar arguments never seems to go away. OT, from this perspective, might be considered an extreme radicalization of such an approach: one that argues that reality is knowable, since concepts are reality, and then deploys a more elaborate mode of transcendental argumentation: instead of proceeding directly from experience to concepts, it starts from certain sorts of verbal statements ("powder is power") and proposes one must employ the transcendental method to derive from these statements the "concepts" (again, a certain kind of time, a certain mode of causality) that must be treated as if they were constitutive not just of experience but of reality itself. 29 In other words, this is not just Idealism-it is about as extreme a form of Idealism as it is possible to have. 30 *** It is possible, however, to deploy the same sort of transcendental method in the opposite direction: to apply it, that is, to problems of Ontology 1 , to questions concerning the nature of reality itself. This is the approach taken by Roy Bhaskar and others 28. Part of this blindness is no doubt made possible, again, by prioritizing abstract reflection ("experience") over action (which integrates experience). Surely one cannot organize one's affairs in any way, let alone organize a social group into moieties labeled "summer" and "winter" or what-have-you, unless you are already operating with an awareness that all events are not simultaneous, that actions have effects, and so on. It is worth remarking that some philosophers, like Alfred North Whitehead (1929), take the same argument further and accuse Kant himself of abstracting experience from action. When one instinctively jumps out of the way of an oncoming vehicle, Whitehead observes, one does not do so because of a series of conscious calculations that being hit will cause injury, but due to an unconscious level of operation where our actions are indistinguishable from those of fish, or insects, or to some degree even plants. It seems reasonable to conclude, therefore, not that notions of time, or cause, do exist in the minds of ants and shrubs, but rather that all physical entities operate in a real world where time and cause are part of reality, and we know this because "we" are not actually distinguishable from our bodies.
29. So Viveiros de Castro (2015: 10) argues that the rise of the term ontology is due in part "to the exhaustion of the critical nomos that separated the phenomenon from the thing in itself "-apparently, by saying there is no thing in itself and relabeling what Husserl would call phenomenology as "Ontology 2 . " 30. Quite recently, some OTers, including Pedersen (2013) and Viveiros de Castro himself (2015: 6), have proposed an affinity between OT and Speculative Realism. I must admit this one leaves me baffled. How does one square what is essentially a variation on Kantian method with a position resolutely opposed to Kantianism in any form? of the Critical Realist tradition (hence, CR: Bhaskar 1975Bhaskar , 1979Bhaskar , 1986Bhaskar , 1989Bhaskar , 1994Archer, Bhaskar, Collier, Lawson, and Norrie 1998;Hartwig 2007;Sayer 2011). 31 Bhaskar's philosophical position is far too complex to sum up in any detail, but it sets out from the same observation as Viveiros de Castro's: since Descartes, Western philosophy has shifted away from questions of ontology and toward questions of epistemology. He parts ways, though, when he adds: in doing so, it has tended to confuse the two. The result is what he dubs "the epistemic fallacy": the question "does the world exist?" has come to be treated as indistinguishable from "how can I prove the world exists?" or even "is it possible for me to have definitive knowledge of this world?" But this implies a false premise: that if a world did exist, it would therefore be possible to have absolute or comprehensive knowledge of it. There is simply no reason to assume one follows from the other. There's no intrinsic reason there could not be a world configured in such a way that philosophers living in it could not come up with absolute proof of its existence, and when it comes to definitive and comprehensive knowledge, the premise actually seems not just wrong but backwards. It makes much better sense to define "reality" as precisely that which we can never know completely; which will never be entirely encompassed in our theoretical descriptions. The only things about which we can have absolute and comprehensive knowledge are things we have made up. Bhaskar applies the transcendental method to ask not just about the conditions of possibility of everyday action and experience, but above all, of the conditions of possibility of contemporary science. Here he focuses particularly on practice, asking not only why scientific experiments are possible (why is it possible to contrive situations with regularly predictable results?), but also why they are necessary (why is it not possible to have predictive knowledge of events unless one has devoted enormous labor into creating such contrived situations?). To answer those questions, he proposes a "depth ontology" that identifies ultimate reality with "mechanisms" and "tendencies" that operate on a series of emergent levels of complexity. How these mechanisms will interact, outside the context of scientific experiments, is inherently unpredictable. This last, CR holds, is true for two reasons: partly, because it is impossible to know how tendencies ("laws") operative on different emergent levels of reality will affect one another in open-system ("real world") situations; partly too because, on every one of those emergent levels, starting with the subatomic, freedom is to some degree inherent in the nature of the universe itself.
Critical Realists argue-compellingly, in my view-that most contemporary philosophical positions are simply variations on the epistemic fallacy. To take one particularly salient example: both Positivists and Poststructuralists tend to agree that if there were a real world independent of the subject, it should be possible (at least in principle) for the subject to have absolute and comprehensive knowledge of it. Positivists argue that such knowledge is possible; Poststructuralists, in most cases at least, argue that since such knowledge is impossible, one must conclude there is no independent reality at all.
Here I can return to the final element in Viveiros de Castro's critique-namely, that there is a contradiction between my political orientation and my theoretical framework: As we shall see, not all political anarchists accept ontological anarchy, i.e. the idea that the only viable political meaning of ontology in our times depends on accepting alterity and equivocation as "unsubsumable" by any transcendent point of view (the very idea of a transcendent point of view is an oxymoron, which did not prevent it from being posited by some ontologies). (Viveiros de Castro 2015: 10; italics in original) The first reference is clearly to myself, since as Viveiros de Castro later suggests, I can only be appealing to such a "transcendent point of view" when I wrote that Ravololona couldn't really stop the hail. Much better to adopt what (he notes) the anarchist essayist Peter Lamborn Wilson has called "ontological anarchy" and acknowledge that any such privileged knowledge, and therefore, moral perspective, is impossible.
It is interesting to note in this context that "ontological anarchy" is a position that Wilson does not, to my knowledge, develop in the works he publishes under his own name, but only in those written pseudonymously under the persona of an imaginary, mad Ismaili poet named Hakim Bey (1985Bey ( , 1994. Here is Bey's description of "ontological anarchy in a nutshell": Since absolutely nothing can be predicated with any real certainty as to the "true nature of things", all projects (as Nietzsche says) can only be "founded on nothing. " And yet there must be a project-if only because we ourselves resist being categorized as "nothing. " Out of nothing we will make something: the Uprising, the revolt against everything which proclaims: "The Nature of Things is such-&-such. " We disagree, we are unnatural, we are less than nothing in the eyes of the Law-Divine Law, Natural Law, or Social Law-take your pick. Out of nothing we will imagine our values, and by this act of invention we shall live. (Bey 1994: 1) Values then are based on nothing but their own assertion. The obvious moral perils of this position might be gauged by the fact that, within anarchist circles, Wilson has ever since been surrounded by controversy, owing to accusations of having invented this very argument to justify pedophilia. 32 So: does "ontological anarchy" mean anyone can make up whatever values they like? (Or does it mean that only certain people can?) 33 32. The fictional persona, Bey, is an out-and-out pedophile; the degree to which Bey can be treated as a stand-in for the author in such matters is, as one might imagine, much contested.
33. I note that while Bey rejects all authorities and all certainties, the actual position of OT seems to be the opposite-at least when speaking of those who can be classified as duly authorized "natives. " This is why Bhaskar (1986: 41) insists that "without realism, fallibilism collapses into dadaism-into an epistemological displacement of Humean ontological scepticism ('anything may happen'). And scepticism here ('anything goes'), as elsewhere, means in practice tacit acquiescence in the status quo, i.e. more or less, 'everything stays' . Irrespective of the author's intentions, scepticism (as anarchism) is These are exactly the sort of questions that motivated Roy Bhaskar-himself a political activist-to turn to the philosophy of science in the first place. One of his primary interlocutors was Paul Feyerabend, an anarchist philosopher of science (e.g., Feyerabend 1975). Feyerabend did indeed take the position Viveiros de Castro endorses, though since he was using traditional philosophical language, he therefore concluded that ontology 1 is unnecessary, and that any contemporary scientific theory is just one of any number of incommensurable perspectives, 34 all of which to some degree construct their ostensible objects, no one of which has any privileged purchase on truth. He referred to this position as "Epistemological Anarchy. " 35 Feyerabend's was a crucial intervention and the notion of incommensurable perspectives has been enormously influential on social theory. On politics as well. Most radical social movements nowadays have come to accept that democracy necessarily means accommodating a diversity of incommensurable perspectives. I have myself tried to incorporate this spirit in my work, long before I was entirely aware of its history. 36 But I also find Bhaskar's response to Feyerabend compelling. 37 Rather than reject the notion that different theories or perspectives largely construct their objects, and are often in many ways incommensurable, Bhaskar argued that this was true-but it did not mean one needed to reject Ontology 1 .
The mistake here, according to Bhaskar, lay precisely in the assumption that a single reality necessarily means acceptance of a single "transcendental point of view. " This, he pointed out, was a perfect example of the epistemic fallacy. The fact that the object of science is, to some degree, constituted by the theory and practice of science itself, does not mean that reality is entirely so constituted; rather, he argued, it is impossible to account for many aspects of scientific practice (experiments, again) without appeal to what he called an "intransitive dimension" of reality-i.e., aspects of the world that would remain the same even if science, invariably dogmatic (and conservative) in effect. " One can observe the slippage perfectly in the Holbraad quote cited above, where ontological anarchy ends up becoming the buttress for existing forms of authority. And speaking as an activist, I can affirm that the most systematic moral relativists I've ever personally argued with have been police.
34. Feyerabend was in fact largely responsible for introducing the word "incommensurability" into academic discourse.
35. Again, using the term in its traditional sense, Epistemology 1 . While Thomas Kuhn (1962) is ordinarily credited with introducing the notion of incommensurable theoretical approaches, Feyerabend was already using the term in the 1950s.
36. For instance, the first two theoretical essays I published (Graeber 1996(Graeber , 1997 were quite intentionally meant to represent different, and to some degree incommensurable, perspectives on overlapping problems. The same might be said of my work on value (2001, 2013b) and on debt (2011).
37. I'm actually in possession of a marked-up copy of Feyerabend's Against method that Bhaskar once plucked off his office shelf to give to me. "Feyerabend is great, " he assured me, "he was a genuine anarchist, and the book is just wonderful. You should read it! Of course I totally disagree with it myself. " scientists, or for that matter humans of any sort, were to disappear entirely. 38 So it's not the perspective that's transcendental (that would indeed be a contradiction in terms) but the most fundamental aspects of reality-in the classic sense of "transcendental" as something which can have effects on us, but we, on the contrary, cannot affect. However, once again, it is one of the defining qualities of reality that it cannot be completely known, let alone encompassed within any one particular perspective. Accepting this makes it possible to say that scientists can say things that are true, and by the same token, they can say things that are false. (It's quite possible-indeed, I would say, likely-that a significant percentage of what currently passes for scientific knowledge is, in fact, incorrect.) It also makes it possible to say that other, incommensurable perspectives on reality, whether common sense, technical expertise, Maori cosmogonic myth, Vedanta, or stand-up comedy, can be able to say other things that are equally true that science cannot say-or indeed, would never think to. All these perspectives are to a certain degree incommensurable. Nonetheless, without a realist ontology 1 , and without some way to anchor values in it, 39 one would have no solid basis on which to argue either that all contain truth, or that a diversity of incommensurable perspectives is in any way desirable.
After all, the mere assertion of a value means nothing in itself: that is, unless, as some of my Malagasy friends would no doubt hasten to point out, you manage to convince others that the value is based on something other than its mere assertion.
It seems to me that taking one's interlocutors seriously means, not just agreeing with everything they say (or even, picking out their most apparently strange or contradictory statements and trying to imagine a world in which those statements would be literally true) but starting from the recognition that neither party to the conversation will ever completely understand the world around them, or for that matter, each other. That's simply part of what it means to be human. Most of what obviously and immediately unites us across borders of every sort, conceptual included, is the recognition of our common limitations: whether that be the fact 38. Anyway it's not my impression that OTers would actually deny this; they just declare it "illegal" to point it out.
39. Hence the "critical" element in "Critical Realism. " A key part of the CR intellectual project is to find a sold philosophical basis for overcoming the Humean fact/value distinction-since simply waving a magic wand and declaring it abolished, as so many try to do, really does not suffice. Bhaskar's argument, again, cannot be laid out in detail, but he begins by pointing out that "fact" is not a synonym for "reality. " A fact is a statement about reality, which has the quality of being true. Simply accepting that certain things are "facts" and others aren't assumes a value: that true statements are preferable to false ones. Any number of other values can be derived from this one: for instance, Bhaskar argues that any form of social organization (he gives capitalism as an example) which can only reproduce itself by representing itself falsely, is therefore less desirable than one that would not. Others working in the CR tradition have tried to root values in reality itself: for instance, Andrew Collier's Being and worth (1999), which begins with a Spinozist argument that all forms of existence with a tendency to persist in their being (Spinoza's conatus) can be considered values to themselves. As this example shows, CR is hardly a theoretical straightjacket: it includes Marxists, Spinozists, Augustinians, Buddhists, Whiteheadians, and many more besides.
that all of us are mortal, or that none of us can never know with certainty how our projects will pan out.
What's more, if one goes slightly further and argues not just that reality can never be fully encompassed in our imaginative constructs, but that reality is that which can never be fully encompassed in our imaginative constructs, then surely "radical alterity" is just another way of saying "reality. " But "real" is not a synonym for "nature. " We can never completely understand cultural difference because cultural difference is real. But by the same token, no one Iatmul, Nambikwara, or Irish-American will ever be able to completely understand any other because individual difference is real too. The reality of other people is the degree to which you can never be quite sure what they're going to do. 40 But finally, all of us are indeed faced with the stubborn reality-that is, immediate unpredictability, ultimate unknowability-of the physical environment that surrounds us.

Malagasy epistemology, or, graceful figures drawn on an abyss
Here is Germain, the younger brother of my friend Armand from Betafo, talking about Vazimba spirits: Vazimba are a kind of thing that isn't seen. They don't show their bodies like, say, people do, or divine spirits when they possess mediums and cure people. If you carry pork to a place where one is, then that night, as soon as you kill the light you look and there's this hand moving towards you. As soon as you light the candle again, it's gone. Or, say you're washing your face in you don't know what it is . . . and likely as not your face will swell up hugely like this, and it absolutely won't go away until you burn incense over it. You take it to someone to give it hasina, and then you're cured. But that's all you know: you have absolutely no idea what was in the water. (in Graeber 2007a: 221) Perhaps the one expression I heard the most, when people talked about spirits, was simply "I don't know. " Spirits were inherently unknowable. (The spirits that possessed mediums were ultimately unknowable as well.) I ended up concluding this lack of knowledge was not incidental; it was foundational. To put it bluntly, while OT would encourage me to privilege the fact that I will never fully understand Malagasy conceptions as to act as if those conceptions were simply determinant of reality, I decided to privilege the fact that my Malagasy interlocutors insisted they did not understand reality either; that nobody ever will be able to understand the world completely, and that this gives us something to talk about. It also gives us the opportunity to unsettle one another's ideas in a way that might prove genuinely dialogic.
In my own ethnographic reflections on fanafody, I argue, in effect, that one cannot begin to understand Malagasy ideas on the subject without understanding their 40. It also means recognizing their freedom. In the afterword of Lost people (2007a: 379-92), I make the argument that our recognition of others as human is grounded above all in their unpredictability, in the limits to our possible knowledge of them. This was largely inspired by my own engagement with Malagasy Epistemology 1 . Epistemology 1 . Because to a large extent, ideas about spirits were themselves a form of epistemology-that is, they are reflections on the possibility of knowledge. On one level, knowledge was power. On another, power was that which one cannot know. People would almost never say such-and-such a person "had" an ody; they would say she "knew" (mahay), "knew how to use" (mahay mampiasa), or "knew how to construct" (mahay manamboatra) one. 41 Power was above all knowledge; but it was also knowledge of the manipulation of forces that were themselves inherently incomprehensible (tsy hita, tsy azo, tsisy dikany).
The word for knowledge in general (fahalalana) was rarely used; normally one spoke either of fahatsiarovana (memory, knowledge of the past) or fahaizana (know-how, practical knowledge, oriented to the future.) Fahatsiarovana was typically in the context of some sort of ancestral authority, but rarely otherwise. Most knowledge was fahaizana, since it was almost invariably tied to a practice of some sort or another. Certainly, anything to do with medicine, or fanafody-a word with a semantic range spanning from herbal concoctions to cure a sore throat to charms that caused one's opponents to become tongue-tied while arguing a law suit-was a form of fahaizana. But so was academic knowledge, which involved doing research, taking tests, and writing and delivering papers. Fahaizana was not authority but sheer power, and often considered at least a little morally ambiguous as a result.
Ody, or "charms, " were not however simply an extension of human knowledge. They were powered by spirits (lolo, zavatra, fanahy), 42 which-in the case of the most powerful ody anyway-were treated as if they had their own autonomous agency, moods, whims, even personalities. But of these spirits "in themselves, " no knowledge was possible. Even the most famous and respected astrologers and curers would insist that it was impossible to say anything about them; anyone who claimed otherwise was a liar or a fool. This is because power was that which was invisible, formless, generic, hidden-it was sheer potentiality. Human knowledge could direct this generic power into specific forms and capacities-most ody were made of bits of wood and other objects, whose names specified one sort of action the ody could take, and someone adept at fanafody would have to know not only the endless varieties of ingredients but the science of astrology to determine the exact moment at which they should be taken, processed, and combined. But the spirits themselves, knowledge could not touch. Even famous ody that had names and personalities-like Ravololona or Ravatomaina-were named after their physical ingredients, and not the spirits that animated them and gave them agency. One 41. The only real exceptions were very famous ody like Ravololona that had names and histories, which often selected their owners. But even this was only when speaking specifically about the ody; when speaking of the owner, one would still likely say "they knew" hail charms, lightning charms, etc.
42. Fanahy is the generic word for "soul, " including the soul of living humans, but was only occasionally used for spirits behind fanafody. Lolo literally means "ghost, " since everyone denied that the spirits' powering charms had anything to do with spirits of the dead, it seems to be used in this context as a generic term for disembodied spirit. Zavatra just means "thing. " In the nineteenth century they were known generically as Ranakandriana (Callet 1908: 83-85, 134, 179-82), but were similarly nameless, indistinguishable one from the other, and unknowable.
didn't even know, in fact, if these spirits were singular or plural. They were that which could not be known. 43 Hasina was the generic term for any kind of power that operates beyond the possibility of human understanding: it was the power of ancestors, spirits, medicine, but also persuasive language (Délivré 1971: 143-45), and finally, it was also the word for objects given to unknown forces to palliate them, pin them down, to beseech them to give a particular favor, and even, from there, to eventually become constituents of ody in their own right (Ellis 1838: 435;Callet 1908: 56;Chapus and Ratsimba 1953: 91n134;Graeber 2007a: 36-38). As noted earlier, an analogous process in the nineteenth century was used in political ritual to produce royal power. Hasina was real enough. Inexplicable things did happen. But you could know so little about the causes of such events that no one could even be sure whether they were a purely social phenomenon, or rooted in something else.
What I' d really draw attention to is that what Malagasy people seem to be doing in many of these cases is strikingly analogous to what OTers suggest for the practice of the anthropologist: they are engaging in an imaginative, poetic process to come to terms with a reality that they know they can never entirely understand. One of the qualities of this imaginative process is that it always tends to linger on the border between artistry and simple fraud. Recall the Malagasy cosmogonic myths mentioned earlier. They grapple with the most fundamental questions of life, love, death-the deepest mysteries of human existence. They are also obviously jokes; people laugh at them, call them "the lies of our ancestors"-though most also feel, on some level, they are also true. Just not true in any literal sense. In fact, for every great existential question there are usually half a dozen mythic answers that plainly contradict. One could, certainly, ask "what would these people have to believe?" or "what would reality have to be like for them?" in order for all these different stories not to contradict, then treat the resulting "concepts" as determinate of a reality we will never fully understand. But doing so would not be a matter of "taking our interlocutors seriously. " As pretty much any one of those interlocutors would be happy to point out, the real point is the tellers don't really understand such matters 43. It is important to emphasize this here because the line of critique that begins with Goldman (2006;echoed, for example, by Tassi & Espirito Santo 2013: 96) asserts that Afro-Brazilian "fetishes" are seen, by their makers, as activating intrinsic powers that exist within material substances, which is of course fitting with a Latourian point of view. I have to assume this is so, but must emphasize that, in the case of Malagasy fanafody, it was explicitly not the case. My own informants were quite insistent about this (and all earlier sources agree). The power came from the conjunction of social agreement and invisible spirit(s); the ingredients essentially channeled that generic agency into the power to do specific things. Explanations of BaKongo nkisi are quite similar, except that the powers are often described as spirits of the dead; and the entire point about the arbitrariness of Early Modern West African "fetishes" does turn on the fact that the qualities of the object are not what's important. The early sources could always be wrong, but since they do in fact conform with the more recent ethnography in this respect, it would be nice to have at least one case of an anthropologist in Africa, fluent in the practitioners' own language, reporting someone who actually said that the power is inherent in the "materiality" of the fetishes before reversing our views on this.
either, nobody does, the ethnographer doesn't either, and that means ultimately, we're all in the same boat.
To give an illustration of the kind of analysis this perspective opens up to us, let me return to hasina. The word is often employed where an English speaker might refer to "luck, " "chance, " or "fortune" (though in the latter case it overlaps with another word, vintana). It took me quite some time to understand this usage. How did it fit with the notion of "invisible efficacy" or "sacred power"? Was everything, even everyday events, ultimately caused by spirits? Then one day it occurred to me: my Malagasy friends, even those who did speak European languages, were equally puzzled whenever I applied the language of statistical probability to everyday events: e.g., "what's the chance the van will leave in the next ten minutes?" let alone "I say 10-to-1 it'll turn out you left it in your other bag. " Such statements made no sense to them. On reflection, it occurred to me that our own application of statistics to everyday events is really just as peculiar as purportedly mystical concepts like mana, hasina, or sakti. We are, effectively, quantifying the exact degree to which we don't know what's going to happen.
Ultimately, human beings are all in the same existential dilemma. We can almost never predict future events with any accuracy; but at the same time, the more time passes since something does happen, the less sense it makes to speak as if anything else "could" have happened instead. This is equally true of social scientists, who make a specialty of writing about past events as if they could have been predicted, even though when they actually do turn their hands to predicting the future, they almost invariably get it wrong. Whenever we encounter an "apparently irrational" belief, we are likely to be in the presence of an existential quandary, a puzzle which no one, really, will ever be able to completely figure out. 44 A final note on the political ramifications of theoretical ideas I began this essay by observing that I seem to have been presented with a challenge. Essentially, I was being asked why my theoretical position was not simply a form of condescending positivism. I also said I thought it would be useful to clarify the actual theoretical difference between Eduardo Viveiros de Castro and myself, since there seemed to be something genuinely at stake in the matter.
It should now be clear that we do, indeed, take almost diametrically opposed metatheoretical positions. I am an ontological realist and theoretical relativist. 45 I value the development of a rich diversity of (at least partly) incommensurable theoretical perspectives on a reality that, I believe, can never be entirely encompassed by any one of them-for the very reason that it is real. Viveiros de Castro, 44. Just to head off possible misunderstandings, I am not, here, advocating anything like the Wittgensteinian position advanced by philosophers like Peter Winch (1964)-that magic, and similar ideas, are best seen not as assertions of power but ways of reflecting on the limits of human power. Magic is indeed an assertion of power. It is an assertion of power that plays on the limits of our knowledge.
in contrast, takes a very different approach to ontology, and (unless I misread him) finds previous theoretical approaches (e.g., Hermeneutics, Dialectical Materialism, or classical Structuralism), insofar as they are so many forms of Epistemology 2 , to be profoundly wrongheaded in their approach. Now, insofar as OT is just developing a new set of questions for ethnographers to put to their materials, it would be silly to object (indeed, it would be inconsistent for theoretical relativist to do so). Anyway, who would really oppose a call to approach our material in a more creative and experimental fashion? This is surely a good thing. What's more, the attempt to grapple with ontological questions, from Irving Hallowell (1960) to Fred Myers (1986) to Michael Scott (2013) has clearly contributed greatly to the discipline, and Viveiros de Castro's own writings on Perspectivism (1992Perspectivism ( , 1998Perspectivism ( , 2004 in particular make up, in my opinion, the single most startlingly original contribution to anthropological theory in recent years. The problems, it seems to me, arise largely when OT begins making explicitly political claims, and therefore, setting itself up as a metatheory that can legislate what anthropologists should and should not say. 46 Much though I hesitate to say it, considering OT's stated hostility to the tradition of cultural relativism, the problems seem strikingly similar to what happened when relativists stopped seeing themselves just as promulgating a methodology, a way of understanding cultures "in their own terms, " and started to insist that relativism should be seen as having a moral and political status, too. 47 It was understandable that they should do so. Many of the people studied by anthropologists were, at the time, widely dismissed as "savages" or "primitives" whose perspectives, ideals, and aesthetics were treated as intrinsically unworthy, or even pathological. Some basic moral points-that it makes no sense to argue that wearing lip-plugs is somehow objectionable, but wearing earrings is not; that it is objectionable to enact laws forbidding the holding of potlatches-obviously had to be made. And no one else was really making them. But the moment relativism became a political principle, let alone a guide to legislation-the moment some began to say that no one 46. Viveiros de Castro has long spoken of "ontological self-determination" as a political project, but the most explicit declaration of OT as a political movement is Holbraad, Pedersen, and Viveiros de Castro's "The politics of the ontological turn" (2014), in which the authors declare OT to be "revolutionary" in every sense of the term. In an accompanying online discussion hosted by the journal Cultural Anthropology they compile fourteen different reflections, by a variety of authors, on the political implications of OT. Here is a list of words that do not appear in a single essay: serf, slave, caste, race, class, patriarchy, war, army, prison, police, government, poverty, hunger, inequality. (I leave out "gender" because the phrase "tacitly gendered perspective" does appear in one.) Granted, the essays were quite short, but for a discussion of ostensibly revolutionary politics, the absence of any of these terms across all of these essays is still remarkable.
47. So, Marshall Sahlins: "Relativism was not and should not be a vulgar moral relativism. It was always a mode of assessing the conditions of possibility of the cultural practices of others, hence of comparative ontology. In that sense, ontological investigation was built into the discipline: a condition of the possibility of anthropology itself " (personal communication, September 2015).
had the right to stand in judgment over any statement or action carried out within a cultural universe different from their own-problems arose. First of all, who gets to define what counts as a "cultural universe"? Can Nuer not judge Dinka, or are all Nilotic speakers close enough that they can be considered members of the same moral community? In drawing borders, one can't simply follow "native categories" because you need to have those borders to know who the relevant "natives" are. So there needs to be an external authority who decides on borders. But then the same problem crops up again when you have to decide who, inside those borders, gets to define what should be considered "Nuer ideas. " Chances are there's next to nothing that every single individual you have just defined as "Nuer" will agree on. So the relativist must appeal to authoritative views. But how are the local authorities to be identified? One cannot use "Nuer ideas" to identify them because that's just circular again: you need to know who the authorities are, first, in order to know what "Nuer ideas" about authority actually are. 48 So, oddly, if you are a cultural relativist, authority is the one thing about which you can't be relativistic. Finally, the moment one decides one cannot stand in judgment over the views of someone residing in a different cultural universe (someone who is Nuer, Dinka, etc.), one immediately develops the need for a special supercategory-such as "modern" or "Western"-in which to include those views one feels one should be allowed to disagree with or condemn. This category therefore tends to balloon endlessly, until it encompasses everyone from Malaysian scientists to Sinhalese anthropologists, Creole plantation-owners, or Iraqi politicians-i.e., pretty much anyone a relativist might possibly wish to say is in any sense wrong about anything-until it looks nothing like any of the other categories in any way. 49 Does OT, or introducing the Deleuzian notion of radical alterity as a political principle, actually improve this situation? It seems to me it makes it even worse. The only major difference I can myself make out with the relativist position, in regard to these specific problems, is that some advocates (e.g., Holbraad 2011) take the conservative implications of classical relativism even further, and propose that OT protects even authoritative views within "the West. " What's more, not only does it appear to continue to require universal standards for recognizing legitimate authority (even across "worlds"), it proposes that those authorities be granted authority over determining the nature of reality itself, within their designated territory, whether or not the individuals in question actually wish to be granted such authority! This, to my mind, is the ultimate irony. Having been accused of introducing Marxist theories "behind the natives' back" I cannot help but turn the question 48. This is not just playing games with logic; one of the first things one usually learns, on settling into an anthropological field site, is that opinions about who can speak with authority are sharply divided. To return to the Nuer: obviously local elders ("bulls") have a certain authority-almost everyone would agree with that. But what about prophets? And if prophets count, do we include all of them, or just the ones that seem to have the least controversial views? If prophets and local elders disagree, whose views take precedence? And what about leopard-skin priests/chiefs? 49. The first part of the argument about relativism I've developed at greater length in an essay called "Oppression, " and the second, in one called "There never was a West" (both in Graeber 2007b).
back again: do OTers really think that most of the people who anthropologists study would actually agree with the proposition that they live in a fundamentally different "nature" or "ontology" than other humans-let alone that words determine things? 50 *** The problem with cultural relativism is that it places people in boxes not of their own devising. As a mere intellectual problem, it's not a big one. The moment relativism becomes a moral or political position, however, it becomes very big indeed. Ontology 2 just substitutes a deeper box. Some people like deep boxes. There seems every reason to believe that those Viveiros de Castro works with, those with whom he struggles for rights to "ontological self-determination, " count among their number. But by that same token, one must respect the desires of those who wish for their boxes to be shallower, or do not wish to be placed in any sort of box at all.

***
An idea that is not dangerous is unworthy of being called an idea at all.
In my more cynical moments, I sometimes think of social theory as a kind of game, where one of the prizes is to see who can come up with the wildest, most shocking, most dangerous-sounding idea, that still does not offer any meaningful challenge to existing structures of authority. And that we have become so used to playing this game that we no longer recognize what a genuinely dangerous idea would even look like.
Let me illustrate what I mean by this. I remarked earlier that an ontological realism that makes it possible to say some scientific statements are true also makes it possible to say other ones are false. Let me turn this around for a moment-even if it means violating a kind of unspoken taboo in anthropological writing (and I'm aware that saying what I'm about to say could potentially get me into far more trouble than advocacy of any sort of "radical social theory" ever could): being able to say that certain forms of magic don't really work is what makes it possible to say that other forms of magic do.
In the case of Malagasy fanafody, this might not seem so extravagant a statement. As I've mentioned, fanafody represents a continuum, from herbal infusions and poultices, many of which clearly do have pharmacological properties (some known, some as yet unknown to mainstream science), to charms designed to cast lightning, render the bearer invisible, or drive one's ex-boyfriend insane by causing him to be possessed by an evil ghost. But I am not just referring to the notoriously ambiguous borders between "somatic" and "psychological" effects here. What I am saying is perhaps there are at least some cases where the practice of fanafody, or other forms of what anthropologists are used to calling "magic, " involve causative mechanisms we simply don't yet understand. There are, after all, plenty of alternative traditions in science, uniformly treated with violent hostility by the intellectual mainstream, that speculate about such possibilities. (Some involve investigating ideas originally proposed by philosophers like Peirce, Whitehead, or Bergson, but the moment one makes such ideas out of the lecture-halls and uses them as the basis for scientific experiments, one is cast amongst the flakes.) No doubt many of their exponents are every bit the cranks and lunatics they're regularly made out to be. But what if some of them were right?
What I'm effectively asking, then, is "what if Ravololona really could prevent the hail from falling on people's crops?" I must confess it still strikes me as unlikely. When I had to call it, I definitely came down on the side of the skeptics on this one. But maybe, just possibly, I was wrong. Still, of one thing I am certain: we'll never have any chance of finding out if we commit ourselves to treating every statement our informants make that seems to fly in the face of accepted ideas of physical possibility as if it were the gate to some alternative reality we will never comprehend. Engaging in such thought experiments does not really open us to unsettling possibilities. Or, anyway, not the kind of unsettling possibilities that are likely to get anyone fired from their jobs. To the contrary, it ultimately protects us from those possibilities, in just the way Holbraad suggested OT protects Western science and common sense.
I began with the Azande, so let me then end with them. It is true that Evans-Pritchard states that Zande witchcraft beliefs cannot be literally true. But there is one famous passage where he seems to qualify that. After telling how his informants explained to him that witches send out their immaterial substance in the form of bright lights moving along paths at night, to seize and destroy the souls of their victims, he remarks: I have only seen witchcraft once on its path. I had been sitting late in the hut writing notes. After midnight, before retiring, I took my spear and went for my usual nocturnal stroll. I was walking in the garden at the back of my hut, amongst the banana trees, when I noticed a bright light passing at the back of my servants' huts towards the homestead of a man called Tupoi. As this seemed worth investigation I followed its passage until a grass screen obscured the view. I ran quickly through my hut to the other side in order to see where the light was going to, but did not regain sight of it. I knew that only one man, a member of my household, had a lamp that might have given off so bright a light, but the next morning, he told me that he had neither been out late at night nor had he used his lamp. There did not lack ready informants to tell me that what I had seen was witchcraft. Shortly afterwards, on the same morning, an old relative of Tupoi and an inmate of his homestead died. This event fully explained the light I had seen. I never discovered its real origin, which was possibly a handful of grass lit by someone on his way to defecate, but the coincidence of the direction along which the light moved and the subsequent death accorded well with Zande ideas. (Evans-Pritchard 1937: 34) True, Evans-Pritchard provides one potential, "rational" explanation, but he doesn't seem to set much stock by it. And the simple, matter-of-fact quality of the description is clearly designed to open up a sense of unsettling possibility: who knows, maybe there actually is something going on here that we just don't know about? Since after all, if someone that no-nonsense tells you there might be something happening that science can't account for, one has to confront the possibility that he might actually be right. réaliste, associée à un relativisme théorique généreux, paraît être une position politique plus convaincante que l' « anarchie ontologique » et l'intolérance théorique défendue par les partisans du tournant ontologique. David Graeber is a Professor of Anthropology at the London School of Economics and Political Science, and author of a fair number of books, including Lost people: Magic and the legacy of slavery in Madagascar (2007), Towards an anthropological theory of value (2001), Direct action: An ethnography (2009), Debt: The first 5000 years (2010), The democracy project: A history, a crisis, a movement (2013), and The utopia of rules: On technology, stupidity, and the secret joys of bureaucracy (2015). He has also developed a certain reputation as a political activist, most notably with the Global Justice Movement and Occupy Wall Street.

David Graeber London School of Economics and Political Science
Houghton Street London WC2A 2AE UK d.graeber@lse.ac.uk