Abstract
In 1929, René Magritte painted a pipe and wrote beneath it the sentence that outlived the image: ceci n'est pas une pipe. This is not a pipe. No amount of looking turns oil and linen into briar and tobacco, and the representation was never the real thing. In 1972, Clifford Geertz's wrote "Deep Play,” effectively arguing that the cockfight itself is a text. Geertz, and the whole tradition of calling culture a text that followed it, makes exactly the confusion Magritte's caption was painted to correct, mistaking a representation for what it represents, and does so at the scale of an entire discipline rather than one canvas. This article argues that culture is not a text on seven distinct counts, each traceable to that single confusion. Culture-as-text gets the recursivity level wrong, mistaking a late symbolic stabilization for where cultural life actually happens. It gets mediation wrong, letting one register annex the other four. It gets the counterpart wrong, treating an interrecursive population as a fixed inscription. It gets authorship wrong, since no living culture ever finishes writing itself and hands over a completed document. It gets temporality wrong, treating an open future as already written. It gets material consequence wrong, treating signs as though they only mean rather than also do. And it confuses a culture's sediment with the living process that keeps producing it. Geertz did not invent this error. He inherited a concept of culture already stripped, over two and a half centuries, of embodiment, dwelling, and the tending of animals and land, and pushed what remained to its logical end: a cockfight rendered so completely as text that a discipline forgot it was looking at a painting. Ceci n'est pas une lutte de coqs. This text is not a cockfight, and this cockfight is not a text.
Ceci n'est pas une lutte de coqs
In 1929 René Magritte painted a pipe and wrote beneath it, in careful schoolroom lettering, the sentence that outlived the image: ceci n'est pas une pipe. The representation is not the real thing. No amount of looking turns oil and linen into briar and tobacco, and nobody has ever lit a painted pipe. Clifford Geertz's "Deep Play," and indeed the entire tradition of calling culture a “text,” makes precisely the mistake that Magritte mocked: to confuse the representation with what it represents. Geertz’s “thick description” of a Balinese cockfight comes with lots of vivid details, but the description is not the fight. Where Magritte calls out ontological confusion, Geertz, and the entire lineage of symbolic anthropology, deepens it.
Clifford Geertz's "Deep Play,” probably the most famous ethnography ever written, somehow gets it exactly backwards. To say culture is a text is like saying that what is represented in the painting has always already taken the form of a painting. Culture, Geertz proposes, is an ensemble of texts, and the anthropologist is reading this text over the shoulders of the natives. This is, I will argue, absolutely wrong in every possible way.
A Balinese cockfight only works because nobody knows what will happen. Birds are matched for size and spirit, spurs are calibrated with real care, bets are placed and raised, opponents are scrutinized, allegiances are exposed that a man might otherwise have preferred to keep vague, and reputations are put at risk, all of it because the outcome remains unresolved until the birds are released. None of this is an unfortunate byproduct of insufficient information that better anthropological method might eventually clear up. It is the whole point, actively engineered by the people staging the fight, who spend real effort making sure the contest stays close enough that its result cannot be predicted in advance.
Ask a deliberately simple question. What would happen if a cockfight really were like a text, in the sense Clifford Geertz proposed when he described Balinese culture as a story people tell themselves about themselves? The answer is that it would collapse. If the fight's result already existed in the way the final chapter of a novel already exists before a reader reaches it, serious betting would be either pointless, since nobody would risk money on a foregone conclusion, or fraudulent, since somebody would already know what only the birds are supposed to determine. Matching the cocks so carefully would be wasted labour, spent balancing a contest whose balance made no difference to an outcome fixed in advance. Watching would stop being watching a contest and become something closer to a reenactment, a ritual repetition of a result already settled.
The culture-as-text metaphor is attractive for a reason worth naming honestly. Anthropologists write. Fieldwork ends in an article, a monograph, a set of publicly readable claims about what a society's practices mean, and it is a short step from producing texts about culture to imagining that culture itself must already have been text-shaped all along. The mistake is to project the properties of that finished product backward onto the living process from which the product was made. An earlier examination of Geertz's own cockfight essay already supplies the decisive sentence, and it deserves to stand near the front of this argument rather than buried near its end: Geertz mistakes the readability of his own finished ethnography for the ontology of the event the ethnography describes.
Geertz does not, of course, claim that the winner of a cockfight has already been determined in the way the ending of a novel has already been written, and the argument here is not that he confuses a cockfight literally with a book. The claim is narrower than that, and harder to dismiss. The textual analogy imports a particular architecture along withe it: stabilized inscription, a reading that follows composition, symbolic articulation given priority over everything that produced it, and a coherence visible only in retrospect. That architecture fits the finished ethnographic account considerably better than it fits the living event the account describes. Reading the analogy this way, as an import of structure rather than a literal claim about scripts, is what makes the objections that follow worth taking seriously rather than dismissible as an attack on a position nobody actually held.
That architecture, taken seriously, indicts the metaphor on seven separate counts, and the rest of this article works through them in turn. Culture-as-text gets the recursivity level wrong, treating symbolic articulation as though it were where cultural life actually happens rather than a late and occasional stabilization of it. It gets mediation wrong, letting one register annex the other four. It gets the kind of counterpart wrong, describing an interrecursive population as though it were a fixed, nonrecursive object. It gets authorship wrong, since no living culture ever finishes writing itself and hands a completed document to its readers. It gets temporality wrong, treating an open future as though it already existed, waiting only to be read. It gets material consequence wrong, treating signs as though they only ever meant things rather than also doing things. And it confuses sediment with process, mistaking a culture's durable residues for the living activity that keeps producing, revising, and discarding them. Seven errors, one metaphor, and, as the next section shows, a historical process spanning two and a half centuries that made the metaphor feel almost inevitable by the time Geertz reached for it.
II. How Culture Became Something That Could Look Like a Text
Geertz did not invent the idea that culture is fundamentally a matter of symbol and meaning. He inherited it, and the inheritance had been two and a half centuries in the making by the time it reached him. Tracing that inheritance briefly, rather than repeating its full history, is enough to show why the textual metaphor felt less like an eccentric literary flourish and more like the obvious next step for a concept that had already been prepared for it.
The word begins a very long way from anything symbolic. Around 160 BCE, Cato the Elder set down instructions for purifying a farm: a pig, a lamb, and a young bull led around the boundary of the fields, wine offered to Janus and Jupiter, a direct address to Mars asking him to protect the household, the flocks, and the harvest together, in the same prayer, on the same afternoon. The word covering this whole undertaking, ploughing and manuring and breeding livestock and performing exactly this rite, was cultura, and cultura never stood alone. It was always the cultura of something, a field, a mind, a soul, a relational, verb-shaped activity rather than a possession. Its root, colere, covered tilling, dwelling, and honouring or worshipping in a single verb, which is why a Roman farmer engaged in his cultura was simultaneously cultivating his land, dwelling in his place, and honouring the gods on whom the place's fertility depended, three faces of one disposition rather than three separate projects. Nothing about any of this is remotely textual. Words and prayer formulas mattered to Cato's rite, but they occurred alongside walked boundaries, animal bodies, soil, and weather, not instead of them.
For roughly eighteen hundred years the word kept this shape. Cicero could extend it to the soul, cultura animi philosophia est, without breaking its grammar: the soul, like a field, still received cultivation rather than possessing Culture outright. Medieval land registries used cultura to mean a specific, measurable cultivated field, a term that stopped applying the moment tending stopped. Then, in the middle of the eighteenth century, the word's grammatical requirement fell away for the first time in its recorded history. Culture stopped needing an object. It became possible to speak of Culture on its own, a general condition a person or a society either possessed or lacked, immediately recruited by Enlightenment historians as a secular measuring stick for ranking societies along a single line from savagery to civilization.
One further step in this sequence matters as much as the break itself, because it supplies the mechanism by which Culture became the kind of object a reader could eventually approach. Johann Gottfried von Herder, writing in the 1780s, objected fiercely to the new noun's use as a single ranking scale and pluralized it: not Culture, singular and gradable, but cultures, the specific, bounded, historically particular ways of life of different peoples, each worth understanding on its own terms rather than measured against one European standard. Herder never asked whether Culture and Nature should have been split apart in the first place. He accepted the partition entirely and multiplied one side of it. But the multiplication did something the singular noun alone could not do: it turned Culture into a countable object, a bounded whole with edges, exactly the kind of thing that could eventually be approached as a unit and, given the right theoretical vocabulary, read. The sequence from Cato to Geertz therefore runs through one more step than the bare grammar suggests. An activity becomes a thing, and a thing, once pluralized into bounded, countable wholes, becomes available to be treated as an inscription.
What matters for the argument at hand is not only that the noun broke free but what, specifically, got left behind when it did. The tending of body, place, and multispecies relation, exactly what had filled most of Cato's manual, was reassigned to a newly invented opposing category, Nature, while durable material practice, the plough, the boundary, the built farmstead, migrated toward a second new term, Civilization, leaving Culture to hold the remainder: belief, custom, meaning, symbol. Of the three pieces into which Cato's single undertaking was sorted, the animals fared worst of all. They did not become Nature in the sense of being handed to a rival discipline that took them seriously in their own right. They became, more quietly, nobody's business at all, absent from the word that had once organized their care alongside everything else a farm required. Edward Tylor's founding definition of the anthropological discipline in 1871 displays the result with unusual clarity. Culture, he wrote, is that complex whole including knowledge, belief, art, morals, law, custom, and whatever other capabilities and habits a person acquires as a member of society. Read the list again. Its enumeration centres acquired social contents and capacities, while a body under tending as such, a place under tending as such, and the management of relations with other species, the very content that fills half of Cato's own manual, are nowhere named. None of this had vanished from the world Tylor was describing. It had simply stopped being culture.
Geertz, writing in 1973, should be read not as an eccentric theorist who suddenly decided culture resembled literature but as the late, logical consequence of two centuries of narrowing he had no particular reason to notice, since the narrowing had become the water anthropology swam in. By the time he sat down to write about the cockfight, culture had already been trained, for two hundred years, to mean symbolic content and little else, and the very animal at the centre of his essay, the cock whose body is injured and sometimes killed in front of everyone present, receives from him almost exactly the treatment the eighteenth century had already prepared for it: read as a symbol of its owner's masculinity rather than encountered as a living creature whose readiness, injury, and death are themselves objects of intense, unsymbolic, embodied attention throughout the fight. Geertz's own statement of it draws directly on Weber. Believing that man is, in Geertz's own words, an animal suspended in webs of significance he himself has spun, he takes culture to be those webs, and their analysis to be an interpretive search for meaning rather than an experimental search for law. Once culture is defined this way, the anthropologist becomes the interpreter of a public, legible web, and in a companion essay published in the same volume Geertz gives the resulting picture its clearest name: culture as an interlocking set of texts that the anthropologist has to work at reading, standing over the shoulder of people for whom the meaning was never in question to begin with. This is not one throwaway line from one essay about one cockfight but Geertz's programmatic statement of what an anthropologist does, and the cockfight essay is simply where he goes to demonstrate it in practice. The historical thesis this article rests on can be put in three short moves. First culture was detached from tending. Then it was detached from materiality. Geertz textualized whatever was left.
III. First Error: People Do Not Live Primarily at L3 and L4
A text, whatever else it is, is already a highly elaborated symbolic stabilization: articulated, durable, available for repeated inspection without further negotiation. Moment-to-moment living proceeds overwhelmingly through the first and second recursivity levels, and the first error in culture-as-text is to assume that cultural life must therefore be happening somewhere else, at the level a text occupies, rather than here.
This is not a claim that the first two recursivity levels sit outside culture, waiting for a text to arrive and make them cultural. Much of what moves through a body absorbedly at the first recursivity level is itself sedimented there by earlier symbolic and institutional activity operating at the third and fourth levels. Table manners can be performed without a flicker of conscious articulation while still carrying an entire history of etiquette manuals, class distinction, and correction behind them. A driver stops at a red light with no more deliberation than a reflex, inside a traffic system that is itself an enormous, deliberately engineered fourth-level stabilization. A ritual gesture can be executed absorbedly, hand and body moving through a sequence learned so long ago it no longer requires thought, while remaining embedded in centuries of accumulated multisymbolic elaboration. Living Value Theory does not divide the world into a non-symbolic zone at the first two levels and a symbolic zone above them. It shows symbolic achievement sedimenting downward into absorbed coordination, so that the same conduct can be, at once, unreflective in the moment and thoroughly historical in its formation.
Geertz's own foundational example for what counts as meaningful, borrowed from the philosopher Gilbert Ryle, is the difference between an involuntary eye twitch and a deliberate wink, identical as movements, opposed as acts, because a wink carries a publicly shared intention a twitch does not. The example earns its keep as a demonstration that culture cannot be studied as a catalogue of physical movements. It also quietly builds in the very assumption this section is questioning, because it sorts every bodily event into only two bins, meaningless motion or articulated sign, with nothing held in reserve for the vast middle ground where most embodied coordination actually occurs. A bettor's hesitation is not a twitch, since it is read, acted on, and consequential. It is also not a wink, since nobody involved could say, in the moment, what exactly it meant, and no public convention certifies it the way a convention certifies a wink as a wink. Geertz's binary has no place to put it, and a great deal of what happens at a cockfight, and in any life, falls into exactly that missing middle.
A Balinese cockfight supplies unusually vivid examples of coordination that has not yet been, and may never be, articulated at all, whatever sediment lies beneath it. A bettor reads another bettor's hesitation and adjusts a wager before either could say precisely what was noticed. A handler reads a cock's readiness through its posture and movement. A villager responds to a policeman's approach and starts moving before any explicit decision to move has been reached. When the police raid the fight and everyone runs, including the anthropologist himself, nobody has paused to interpret the raid's cultural significance first. These are consequential acts of discernment, not commentaries produced about a coordination from outside it, and most of them never rise to the level at which they could be articulated at all.
One qualification has to be made carefully here, because it is easy to overstate the point into something false. Reading a text can itself become absorbed and automatic: a fluent reader takes in a page without labouring over each word, and in that sense reading, too, can occur at the first recursivity level. This does not rescue the text metaphor. It relocates the problem. Fluent reading is a secondary sedimentation, possible only because a symbolic system has already been built and stabilized at higher levels a text requires: an alphabet agreed upon, a language shared, a genre recognized. Ordinary living coordination, a bettor reading hesitation, a body recoiling from danger, requires no such prior textualization. It does not wait for a symbolic system to be built before it can occur, and in most living beings, and for most of any human life, it never waits for one at all.
This yields an asymmetry the rest of the article depends on. Lived coordination can occur perfectly well without text of any kind, as it does in every organism without language and in the greater part of every day even a fluently literate person lives. Text, by contrast, cannot occur without a great deal of prior living coordination already in place, since it requires shared symbolic conventions that themselves had to be built up, sedimented, and transmitted through exactly the kind of face-to-face, embodied coordination that a text, once finished, no longer needs. Geertz's metaphor reverses this dependence. It treats the possibility of textual legibility as though it were the basic structure of the life that eventually, occasionally, produced a legible trace, when the dependence in fact runs the other way entirely.
IV. Second Error: One Mediation Annexes the Other Four
The culture-as-text metaphor is not simply guilty of overvaluing the symbolic but guilty of a specific and more diagnosable failure, which deserves its own name: mediational imperialism, the annexation of embodiment, being-with, dwelling, and material forming by a single mediation, multisymbolism, that then behaves as though it already contained or explained the other four.
Cato's rite is the clean counter-example, because it shows what an undivided practice looks like before any single mediation has been allowed to conquer the rest. The purification of the farm joined a body walking a boundary, a place under tending, a set of relations of dependence and reverence extending to household, flock, and god, material practices of ploughing and offering, and symbolic formulas addressed to named gods, all operating together in a single afternoon's undertaking. None of these five registers explained the others. Each was doing its own indispensable share of the same continuous work.
Redescribe the cockfight the same way and the annexation Geertz performs becomes visible by contrast. Bodies sweat, tense, recoil, and run when the police arrive. Cocks injure and kill one another, a fact no amount of symbolic reading converts into something other than an animal's death. Men watch rivals and allies with an attention that has nothing yet to do with interpretation. Steel spurs, money, the built ring, and the physical arrangement of the crowd organize what is possible before anyone says a word about what any of it means. Village histories, kin relations, and religious framings locate the whole event within an inhabited world that long predates tonight's fight. A man who bets against his own kin group's cock, or who lets himself fall into debt to a rival rather than to his own people, has broken a constraint no one needed to spell out for him to feel its weight. The loser's cock, torn apart in the ring, is carried home and eaten that same evening, a fact about bodies, appetite, and a household's dinner fully as much as it is a fact about anyone's honour. Words, classifications, bets, and reputations articulate parts of all this symbolically, and that articulation is real. What is false is its promotion from one register among five into the ontology of the whole.
The same annexation can be tested against almost any ordinary cultural event and produces the same result every time. A meal is not its symbolism, however much a meal certainly carries symbolism. A wedding is not its vows. A court case is not its judgment. A football match is not its final score. A house is not its address. A ritual is not its doctrine. In each case a fully multimediated living process gets described, by a text-centred method, through only the fraction of itself capable of surviving symbolic transcription, and the survival of that fraction is then mistaken for evidence about the structure of the whole event rather than for what it actually is: an artefact of which parts of a rich, embodied, material, relational occasion happen to convert cleanly into words on a page.
This gives the argument one of its more general formulations. Culture-as-text takes a fully multimediated living process and preserves only the parts capable of surviving symbolic transcription. It then mistakes that survival bias for the structure of culture itself. This is a stronger charge than saying Geertz simply ignores materiality, embodiment, or dwelling, a complaint that could be met, in principle, by writing a more materially attentive ethnography next time. The method itself, treating culture as a text to be read, preferentially preserves whatever can become text and discards the rest before description has even properly begun, and the omission cannot be fixed by writing more carefully within the same framework. It can only be fixed by refusing the framework's founding move, letting one mediation stand in for all five.
There is a further twist worth making explicit, because it shows the annexation running deeper than a simple preference for one mediation over the other four. Even a text is not merely textual. A printed book requires paper, ink, a press, transport, shelving, and light to read by. A digital text requires a screen, electricity, and a device held in an embodied reader's hands, in a room, at a time, alongside other people or in their deliberate absence. What culture-as-text actually borrows from the text, in other words, is not the text in its full, multimediated existence but one property selectively abstracted from it, symbolically stabilized readability, lifted free of the paper, the press, the shelf, and the trained body doing the reading that make any actual text possible at all. Geertz does something stranger, then, than simply reducing culture to texts. He reduces culture to an idealization of texts, an idealization already achieved by stripping a real text of everything about it that is not multisymbolic.
This has a consequence larger than any single misreading of one cockfight. Once culture is identified with readability, the events that will strike a fieldworker as most obviously cultural are exactly the ones already dense with public symbolism: ritual, myth, ceremony, dramatic performance, elaborate classification. Feeding animals, mending a roof, repairing a fence, cooking, cleaning, carrying water, keeping warm, walking a familiar route, and tending a sick body recede into an unmarked, merely practical background, not because they carry no cultural weight but because they resist being rendered as something a reader could sit down and interpret. This is the same amputation the deep history of the word already uncovered, playing out again at the level of method rather than definition: the animals and the tending disappeared from the word two centuries ago, and a theory that treats culture as a text to be read will keep finding it easiest to notice exactly the fraction of any society's life that survived that disappearance. The more readily an activity can be rendered as something meaningful to read, the more readily an interpretive method will recognize it as cultural at all, which means the metaphor does more than misdescribe events anthropologists already study. It helps decide, in advance, which events get studied as cultural in the first place.
V. Third Error: A Text Is the Wrong Kind of Counterpart
A stabilized text is, in the vocabulary this corpus has developed, principally a nonrecursive counterpart. A reader can change while reading it, can weep, reconsider a long-held opinion, read the same page five times with five different moods, but the text itself does not notice any of this and does not rewrite its next paragraph in response. Macbeth does not alter his final soliloquy because a particularly perceptive reader has correctly anticipated how the play will end. The sequence is settled, and settled sequences are exactly what a nonrecursive counterpart is built from.
The claim here concerns the text as counterpart, not everything a living reader may recursively do with it. Reading can itself be highly interrecursive: a reader argues silently with an absent author, revises an opinion in light of a second reading, annotates a margin in a way that materially changes the object for the next reader, and different readers can become interrecursive with one another around a single shared text. None of this touches the point at issue. However alive the reading is, the stabilized inscription being read does not adjust its own next sentence in response, which is all the claim that a text is a nonrecursive counterpart was ever meant to establish.
A cockfight is a different kind of object from the ground up, and the difference is best displayed by separating four properties usually blurred together under the single word text rather than by sorting events into four separate boxes. A fixed inscription has a settled sequence: nothing that happens during a reading changes what the next page says. A scripted performance fixes major outcomes while leaving execution variable: an actor can deliver a line a hundred different ways, but Duncan is going to die in every single performance, because the script has already decided that much and left open only the manner of arriving at it. A stabilized rule-frame settles its rules precisely so that an outcome the rules do not fix can remain open: football's laws are as settled as any script's stage directions, and their entire purpose is to protect an uncertainty the laws themselves do not resolve. A live enactment goes further than any of these three, because its participants continuously alter one another's fields of possible action as the situation unfolds, so that not even the frame within which the outcome will be decided is guaranteed to stay fixed. These four are not different kinds of event sorted into separate categories but properties a single occasion can combine in different proportions: a football match is simultaneously a stabilized rule-frame and a live enactment, and a wedding or a trial typically layers a fixed inscription in its order of service, a scripted performance in its formal moments, and a live enactment wherever cross-examination or an unplanned exchange breaks through the script. Treating any occasion built from several of these properties as though it were simply the first, a fixed inscription and nothing else, collapses distinctions that matter enormously to the people living through it.
This lets a distinction be drawn that culture-as-text cannot draw, and drawing it sharpens rather than dissolves the argument's respect for cultural artefacts. Cultural sedimentations can indeed function as nonrecursive counterparts in exactly the sense a text does: a law once passed, a recipe once written down, a building once completed, a document once filed, a road once laid, a ritual object once consecrated, an inherited phrase once fixed in speech. None of this is being denied, and none of it should be. But cultural life happens when living beings take these sedimentations up, argue about a law's application, adapt a recipe to what the market happened to have that morning, wear a building down through decades of use, invoke a document in a dispute nobody who filed it anticipated, carry a ritual object into a context its makers never imagined, bend an inherited phrase to a new and unauthorized use. The sediment is not the whole of culture but one indispensable ingredient, alongside the living uptake that keeps returning to it, reworking it, and sometimes wearing it away entirely.
The deeper mistake culture-as-text makes, once this distinction is available, can be stated with some precision: nonrecursive cultural sediment gets mistaken for recursive cultural life. A law, a recipe, a building, a document, and a fixed phrase are treated as though they already were the culture rather than one durable trace a living culture continuously produces, revises, and occasionally discards. Geertz's cockfight essay performs exactly this substitution when it treats the cockfight as a text the Balinese have written about themselves, since a text of that kind, properly speaking, is the sedimented residue a living event might eventually leave behind, in someone's memory, in a proverb, in a later retelling, and not the event's own ongoing structure while it is still being lived.
What the preceding paragraphs have been building toward can now be stated directly. Texts are powerful partly because they stop answering back. Cultures remain alive because people do.
VI. Fourth Error: Nobody Can Say Who Wrote the Text or Who Is Reading It
Take Geertz's most famous formulation, that the Balinese tell themselves stories about themselves, absolutely literally for a moment rather than allowing it to slide past as a graceful turn of phrase, and a question opens up that the formulation itself never answers. Who is the Balinese here, and who is telling whom?
The village Geertz describes contains an owner whose cock is fighting and his rival whose cock is fighting against it. It contains a politically powerful man whose bet carries weight beyond its monetary size and a man caught between two kin allegiances who may, as the earlier examination of this material showed, avoid betting on either side rather than expose a loyalty he cannot comfortably declare. It contains a habitual gambler whose relationship to the event differs from a casual spectator's, a child watching from the edge, a woman standing in yet another relationship to the proceedings, the police officer who may raid the fight before it concludes, Geertz himself, and the animals whose next movement everyone present is trying to anticipate before it happens. None of these participants holds the same information, wants the same outcome, or has the same amount riding on tonight's result. Compressing all of them into a single narrating subject called the Balinese converts an interrecursive plurality, many centres of discernment reading one another under conditions of real and lasting opacity, into an imaginary selfrecursive subject, one being responding to its own responses, which does not exist anywhere in the evidence Geertz himself collected.
Turning this into an explicit author-reader problem sharpens it further than treating it as a complaint about oversimplified collective subjects. A text ordinarily implies some working distinction between inscription and reading: somebody, or some process, stabilizes an inscription sufficiently that a separate encounter with it can meaningfully count as reading it, after the writing has finished and before the reading begins. A living culture supplies no such boundary anywhere a careful look is willing to search for one. The bettor at the ring's edge is simultaneously reading his rivals' hesitation and, by the size and visibility of his own wager, changing what those same rivals now have to read in turn. Geertz reads the villagers around him throughout his fieldwork, but their reading of him, of how he behaved during the police raid, of what his continued presence signals, changes his own status and therefore changes what there is for him to read next. A man's bet makes his factional affiliation more publicly legible, and that increased legibility itself becomes new material for everyone else's subsequent discernment, which then feeds back into how the next round of betting proceeds. At no identifiable point does Balinese culture finish writing itself and hand a completed document over to its readers, because every act of reading inside this situation is simultaneously an act of writing new material for somebody else's next reading, in a loop that has no textual analogue at all.
This is also why the metaphor cannot be rescued by retreating to the claim that it was only ever meant metaphorically, a common and usually effective move for softening an overstated comparison. Here the retreat fails for a specific reason. The more precisely author, reader, and inscription are actually specified for a living cultural situation, the less the resulting structure resembles a text of any kind, rather than resembling one more loosely but still recognizably. A metaphor survives softening when its edges are merely exaggerated. This one dissolves under the same scrutiny that would normally rescue a strained comparison, because pressing on where the text metaphor would need to locate an author or a reader reveals no such fixed position anywhere in the situation to find.
This has a further consequence, because it connects a specific criticism of one essay to a much larger pattern in classical anthropological thought. Culture-as-text slides easily, almost automatically, into a collective-subject model of exactly the kind that produced the Balinese as a single narrator: a society says, a culture believes, a people tell themselves stories, formulations so familiar in anthropological prose that their oddness rarely registers. Living Value Theory starts from the opposite premise, plural recursive centres operating under conditions of partial and often durable opacity to one another, and treats that plurality not as noise obscuring an underlying cultural unity waiting to be recovered by a sufficiently skilled reader, but as the very condition that makes cultural coordination necessary at all. If everyone already shared one reading, coordination would have nothing left to do. The disagreement, the incomplete information, the differently positioned stakes are not culture's imperfection but what culture, understood as ongoing coordination rather than as a finished text, actually consists of.
VII. Fifth Error: The Future of a Text Already Exists
Two forms of uncertainty need to be told apart before the temporal error in culture-as-text can be stated, because ordinary language uses the single word uncertain for both and thereby hides the distinction that matters most here. When a reader opens a novel for the first time, page three hundred is unknown to that reader, but page three hundred already exists, printed, bound, sitting in the pages ahead exactly as it will read when reached. The reader's uncertainty is epistemic: a gap in what one particular reader currently knows about a fact that is, independently of that reader, already settled. Before a cockfight is fought, by contrast, the outcome does not merely remain unknown to the spectators. It remains unmade. No participant, no rule, no cultural classification, and no representation available to anyone present already contains the result, a claim about mesocosmic openness rather than about physical determinism one way or the other, and the actions taken between now and the fight's end, how the birds are handled, how the crowd's attention shapes a handler's last-minute adjustments, will actually help produce whichever outcome eventually occurs. This is not a difference of degree between two kinds of not-yet-knowing but a fundamental ontological difference between a textual future, which already exists and merely awaits a reader's arrival, and a living future, which does not exist yet in any form because it has not yet been made.
The cockfight material examined earlier in this corpus supplies unusually strong evidence for this distinction, because the Balinese do not merely tolerate the uncertainty surrounding a fight's outcome. They actively protect it, matching birds by size and spirit, calibrating spurs, selecting handlers for skill, organizing betting in ways that keep a contest close enough to remain uncertain rather than a foregone conclusion dressed up as a contest. This generates one of the harder blows this argument can land against the essay it is examining. Geertz chose, as his exemplary demonstration that culture is fundamentally textual, an institution whose participants devote considerable, visible labour to ensuring that its ending has not yet been written by anyone, least of all by the community supposedly narrating it to itself in advance.
Living Value Theory's account of mixed futurity supplies the more general principle this specific case illustrates. Cultural life always combines sedimented constraint with a future that remains, in whatever measure the situation allows, unresolved. The past narrows what can plausibly happen next, who can credibly bet on which side, which handler a family would trust, without ever containing what will actually happen, in the way a textual future is already fully contained in pages not yet turned. A cultural situation is always more open than its own sediment, and treating the sediment as though it already specified the outcome mistakes the narrowing of possibility for its elimination.
The text metaphor systematically exaggerates this kind of retrospective coherence, and the reason is almost structural rather than a mere failure of attentiveness on any particular ethnographer's part. Any ethnographic account, by the time it is written, already knows how the event it describes ended, and writing about an event whose ending is already known makes that ending very difficult not to treat as though it had been available, in outline, from the very beginning. This yields a methodological warning that holds independently of Geertz's particular essay: narrative hindsight converts ontological openness into apparent inevitability, quietly, almost without the writer noticing it happen, every time a finished account is mistaken for a faithful record of how open the situation actually felt while it was still being lived.
VIII. Sixth Error: Signs Do Things Before They Mean Things
Geertz's reasoning for treating the cockfight as art rather than as ordinary practical life is not an unmotivated flourish borrowed from literary taste. It draws on a specific and respectable line in Kantian aesthetics, the idea that art works by removing an object from immediate practical urgency and raising it to the level of appearance, where its meaning becomes more available to perception precisely because acting on it is no longer at stake. Applied to the cockfight, this licenses Geertz's claim that money, blood, and status all get suspended in something like an aesthetic bracket, freeing the event's significance to be read all the more clearly because nothing practical is actually riding on it. The reasoning is elegant, and it is also exactly backward for an event whose participants spend real effort ensuring that a great deal is riding on it.
Geertz treats an increasing wager as a sign whose growing size reveals, with increasing clarity, that the fight is really about status, honour, and public esteem rather than about money as such. The reading is not wrong, but it stops one step short of the more important claim available in the same material. A wager does not merely signify intensified stakes from a position safely outside the situation it describes. Raising it directly alters attention, exposes an allegiance that could otherwise have stayed comfortably vague, changes which handlers get selected and how carefully birds get matched, and shifts what will be remembered afterward and by whom. The sign is doing work in the world before, and in addition to, whatever it might also be taken to mean.
Geertz states the consequence of his own bracketing explicitly. Because the large, even centre bets tend, on his account, to balance out across many fights, and because a man's formal status is never actually altered by any single result, only momentarily affirmed or insulted, he concludes that the money at stake is, in the end, beside the point, a vehicle for something that leaves the underlying social order untouched. This is precisely the claim the rest of this section exists to unsettle. A bet that leaves formal status exactly where it was can still leave a great deal else moved: who is remembered as having flinched, who is now read as bolder or more cautious than before, which alliance has just been tested and which has quietly strained. Balancing out across many fights, if it happens at all, is a claim about the long run rather than about any single evening, and it says nothing about the fight in front of anyone actually watching it, for whom the outcome is not an average but a single, unrepeatable event.
The same point holds for every material element the cockfight puts to use, and holding it consistently is what separates a fully multimediated account from an interpretive one wearing multimediation's vocabulary without its substance. The police raiding the fight do not merely symbolize state authority reaching into village life. They can make an arrest, and the threat of arrest is precisely what sends everyone running before anyone has time to interpret anything. A steel spur fixed to a cock's leg does not merely signify aggression or masculine competition. It cuts, and a fight can end because a spur has done exactly that, independently of whatever the cut is later said to mean. Money changing hands does not merely represent a settled social relation. It leaves one man with less than he had an hour before and another with more, a redistribution that occurs whether or not either party ever narrates it as significant.
This yields a distinction culture-as-text has no room for, because its method routes every element of a situation through meaning before asking what, if anything, that element also did. A symbol can signify something while simultaneously intervening materially in the very situation that gives it whatever significance it eventually acquires, and the intervention does not wait for the signification to be worked out first. Geertz's interpretive method, built to recover meaning, has no comparable procedure for registering an act's material consequence except by translating that consequence back into another layer of meaning, at which point the spur's cut and the arrest's threat disappear into symbolism a second time, exactly where they were most in need of being taken as what they also, undeniably, are.
IX. Seventh Error: Text Confuses Cultural Sediment With Cultural Process
The culture-history this article opened with supplies the key to its own positive reconstruction. Cultura originally named an ongoing relationship of tending, not a finished property a person or a people simply possessed, and the eighteenth-century transformation that eventually made Geertz's metaphor possible repeated, at the level of grammar, exactly the move culture-as-text repeats at the level of theory: a process becomes a noun, and a noun then becomes an object available for reading. Recognizing this parallel is what lets the present argument move past criticism into reconstruction.
Cultural life continuously leaves behind relatively durable multimaterial and multisymbolic residues: buildings, worn pathways, tools, clothing, money, legal categories, personal names, kinship classifications, recipes, documents, ritual sequences, songs, stories, institutional procedures. Their relative nonrecursivity, the fact that they do not answer back the way a living interrecursive counterpart does, is not a defect to be lamented but indispensable to life being livable at all, since nothing could function if every relation had to be negotiated afresh each morning with no durable trace of yesterday's negotiations to build on. Culture, properly understood, involves recursive closure fully as much as it involves recursive opening, stabilization as much as improvisation, and a theory that only valued the open, live, interrecursive side of cultural life would be exactly as one-sided as a theory that reduced everything to sediment.
The instability of the alternative shows up clearly enough in the discipline's own history. By 1952, culture had already accumulated so many rival scholarly definitions, more than one hundred and sixty of them catalogued in a single review, that the proliferation looks less like healthy theoretical ferment and more like a symptom: a word straining to do, badly and in a great many separate fragments, what a single, unified cultura had once done without anyone needing to define it at all. Much of twentieth-century social theory's long argument between materialist accounts of culture, grounding it in economic and technological conditions, and symbolic or idealist accounts treating culture as meaning irreducible to material cause, looks the same way once set against this history. It reads less like a perennial philosophical dispute reaching back through the whole of Western thought and more like the predictable consequence of having severed multimateriality from multisymbolization in the eighteenth century, after which two centuries of theorists inherited the separated pieces and spent a great deal of effort arguing about which piece deserved to be called primary, without noticing that the argument exists only because the pieces had been split apart to begin with.
The cockfight's own inherited kinship and status rules illustrate the relationship between sediment and process precisely. These arrangements constrain what a participant can plausibly do, which cock a man can credibly back, which allegiance a bet will be read as confirming, without scripting his conduct in the stronger sense a written script would. This yields the positive formula this article has been assembling piece by piece: sedimentation, followed by living uptake, followed by ongoing coordination, followed by disturbance or opening when the sediment no longer fits smoothly, followed by fresh discernment, sometimes articulated explicitly, sometimes stabilized once again institutionally, and then further sedimented for whatever comes next. Culture is neither the sediment alone, which would make it indistinguishable from an archive, nor the live interaction alone, which would make it impossible to explain why the same community keeps producing recognizably similar occasions year after year. It is the historical, recursive relation between sediment and the living beings who keep taking it up, testing it, wearing at it, and eventually revising it.
This is also the point at which a handful of twentieth-century anthropological interventions deserve to be named directly, because each of them, in its own partial way, was smuggling back a dimension the modern, narrowed concept of culture had already discarded before that theorist ever picked up the word. Roy Rappaport's account of a highland New Guinea pig-raising and slaughter cycle showed that ritual timing, taboo, and obligation to ancestors were doing real ecological regulatory work, redistributing protein and pacing the outbreak of hostilities in step with the land's own carrying capacity, work that only looks like two separate systems, religious belief on one side and ecological process on the other, once a modern vocabulary insists on sorting it that way before description has even begun. Pierre Bourdieu's habitus pulled something like embodiment back into serious cultural theory for the first time in two centuries, treating class position as inscribed in posture, gait, and the handling of cutlery rather than merely believed, and supplying, in the same stroke, one of the clearest cases of historically produced arrangement sedimenting all the way down into unreflective bodily disposition. Tim Ingold's insistence that cultural variation is, at bottom, variation in trained embodied skill, learned anew through practice rather than transmitted ready-made as content, restores something close to Cato's own sense of cultura as an ongoing, practical discipline of tending. None of these three theorists supplies the complete alternative culture-as-text needs, and none of them was working, so far as the record shows, with anything like the present vocabulary of mediations and recursivity levels. What they share is more interesting than any one of their individual contributions: each had to fight, without quite naming what he was fighting, to restore a dimension of cultural life that had been conceptually amputated from the word two centuries before any of them was born.
X. The Balinese Cockfight Rebuilt: From Cultural Text to Recursive Stakes Machine
Criticism earns its keep only if it produces a better account of the very case it examines rather than simply clearing the ground and leaving nothing standing. The cockfight, redescribed through the architecture this article has been assembling, becomes something considerably more precise than a text, and considerably closer to what Geertz's own extraordinary fieldwork actually shows happening.
Call it a bounded interrecursive intensifier, or, in plainer terms, a recursive stakes machine, and trace the sequence it runs through on each occasion it is convened. Prior kin relations, personal and factional reputations, and long-standing alliances sit as relatively sedimented background, present but not currently demanding anyone's active attention. A scheduled event selectively reopens exactly this sediment, rather than leaving it undisturbed or dismantling it wholesale. Rules and the careful matching of birds stabilize the frame within which the reopening will occur, so that not everything is up for renegotiation at once. Money then intensifies the consequences riding on how the reopened relations play out, and embodied attention rises correspondingly, eyes on birds, on rivals, on the size of an opponent's wager. Affiliations that had been comfortably vague become, for the duration of the event, more sharply askable than they were an hour before. Several distinct recursive centres, owners, handlers, bettors, spectators, rivals, read one another under conditions of real and uneven information, none of them holding a complete picture of what any of the others intend. The future remains open throughout this entire sequence, right up to the moment a result occurs. Once it does, memories, reputations, and expectations shift, sometimes only slightly, occasionally decisively, and the relations that had been reopened undergo a partial reclosing that returns the community to something resembling its earlier, sedimented state, though never to a state identical with the one the sequence began from. The next cockfight, whenever it comes, therefore begins from a slightly altered sedimented world rather than from the same starting point twice.
This gives deep play a stronger reading than the one Geertz offers. For Geertz, mounting stakes work primarily as an intensifying sign, revealing more and more clearly that the fight is really about status, honour, and public esteem rather than about money as such. That reading can be kept, in a modified form, once the mechanism behind it is supplied. The depth of deep play does not arise because money secretly stands for something deeper in the way a word stands for its meaning. Raising the stakes changes the interrecursive field itself directly: attention sharpens, scrutiny of rivals intensifies, affiliations that could have stayed comfortably unstated become harder to avoid declaring, and what will be remembered afterward, by whom, about whom, changes in scale along with the money at risk. The wager is not a sign pointing elsewhere but an intervention in exactly the process this whole article has been describing.
This section demonstrates something the preceding seven errors could not demonstrate on their own, however carefully they were argued. A framework built from mediations, recursivity levels, and interrecursive counterparts is not merely a more cautious or more qualified way of saying the same things Geertz said. It preserves everything of value in his extraordinarily attentive ethnographic observation, the matching of birds, the structure of the betting, the raid, the shifting allegiances, while dispensing entirely with the textual ontology he mistakenly imposed on top of it. The observations survive the criticism intact. Only the metaphor organizing them does not.
XI. Culture After the Text
Three easy replacements for the text metaphor should be refused explicitly, because each one repeats the same underlying error in a different disguise rather than correcting it. Culture is not simply practice instead of text, since that substitution risks installing a new master reduction, practice, in place of the old one, meaning, without asking why one register should still be allowed to annex the other four. Culture is not simply interaction, since a great deal of cultural sediment, a law on the books, a building nobody is currently occupying, a document filed and forgotten, persists in full force while nobody happens to be interacting with it at all. And culture is not simply materiality, a fashionable corrective in recent decades that merely reverses Geertz's symbolic reduction rather than repairing it, promoting multimateriality to the same annexing role multisymbolism previously occupied.
The formulation this article has been building toward can now be stated directly, and it has to name a relation rather than a thing if it is to avoid smuggling the same error back in under a new vocabulary. Culture is the historically recursive relation between living multimediated coordination and the multimaterial and multisymbolic sedimentations that previous coordination leaves behind. Sediment constrains what happens next without determining it. Living uptake reproduces, ignores, modifies, or reopens that sediment, taking it up at the first recursivity level, rendering it salient at the second, articulating it at the third, stabilizing it institutionally at the fourth, and, on occasion, repositioning it at the fifth. Culture is neither the sediment by itself, which would make it just another archive, nor the living activity by itself, which would make it impossible to explain why the same community keeps producing recognizably similar occasions year after year. It is the relation, kept alive only by being repeatedly taken up, between what earlier coordination has left behind and what living beings continue to do with it.
Reconnecting this formulation to the word's own deep history does not amount to a nostalgic return to Roman cosmology, and it should not be read as one. Nothing here vindicates a Roman farmer's household gods, still less the institution of slavery his estate depended on, and nothing here suggests historical particularism was wrong to insist that societies cannot be ranked along Tylor's evolutionary ladder, or that the interpretive and critical turns in twentieth-century anthropology were wrong to insist that meaning, representation, and power cannot simply be waved aside in favour of a return to soil and livestock. What recovering the word's history actually shows is narrower and, for that reason, more useful. Anthropology took one historically produced fraction of living coordination, symbolic content, and repeatedly mistook that fraction for the whole, not through any individual theorist's carelessness but because the word handed down to the discipline had already been trained, for two centuries before anthropology existed as a profession, to mean symbolic content and increasingly little else.
Geertz's own achievement can now be judged with some precision, and the judgement is not unkind. He did not discover that culture is textual. He demonstrated, with unusual skill, how completely an anthropologist can textualize cultural life once handed a concept of culture already narrowed enough to make the textualization feel natural rather than forced. That demonstration should not be mistaken for the ontology of the life he textualized.
The contrast this article has uncovered is worth leaving as its final word, since it says in one sentence what the preceding eleven sections have taken considerably longer to establish. A text can surprise a reader because the reader does not yet know what has already been written. A culture surprises the people living it because what happens next has not yet been made. The difference between those two forms of surprise is everything the culture-as-text metaphor leaves out.
Cite as: Ecks, Stefan. 2026. "Culture Is Not a Text: Everything Wrong with Anthropology's Most Famous Metaphor." Living Value Theory, livingvaluetheory.org.