Why does it seem like a good idea to wear this particular pair of socks on a day when a lot is at stake? Superstition is not a failed belief or a response to danger and uncertainty. It is a practical coordination move that appears only when people treat a counterpart as recursive—something or someone that answers back. Living Value Theory calls this threshold coordination: the small, low-cost gestures performed just before an outcome that skill alone cannot guarantee. A nonrecursive domain is governed by technique and measurement. A recursive one cannot be, because any rule becomes part of the interaction itself. Across cultures, these acts therefore cluster around live opponents, audiences, prey, and diffusely attributed forces such as “luck,” while remaining absent from domains treated as purely mechanical. They operate at an immediate, local level and almost never harden into formal doctrine. The familiar shrug that says “I know it does nothing, but I still do it” is not confusion; it is the accurate report of a practice that was never meant to be a scientific claim. This article develops a radically new theory of superstition through the ontological distinction between the recursive and the nonrecursive.
A batter taps the plate three times before he settles into his stance. A trawlerman touches cold iron when someone on deck says the word that summons bad luck. A surgeon forbids the word that names the very calm the operating theatre depends on. Someone crosses themselves as the aircraft begins its roll down the runway, or taps a table and says the formula that wards off the consequence of having spoken too soon. None of these acts changes anything a physicist, a physiologist, or an actuary could measure. All of them are performed anyway, often by people who in the same breath will say they know it changes nothing, and all of them cluster with a regularity too exact to be incidental around a specific kind of moment, the moment before an outcome that competence alone will not secure.
Living Value Theory has a name for what varies systematically between the settings that draw this behaviour and the settings that never do, and the name is neither danger nor uncertainty, though both correlate with the pattern closely enough to have misled a century of research into treating them as the cause. The variable is recursive type. Superstition, or what this article will call threshold coordination, is the small, low-cost, causally opaque coordination move that appears wherever a counterpart is engaged as selfrecursive, or far more often and far more intensely as interrecursive, and it does not appear, in any mesocosm anywhere, addressed to a counterpart engaged as nonrecursive, however violently that nonrecursive counterpart varies and however badly it is predicted. What holds constant across every culture the ethnographic record has ever documented is the coupling itself: a nonrecursive counterpart is governed by technique, and an interrecursive one cannot be, because a counterpart that answers back is not secured by a procedure applied to it from outside as though it would simply hold still. What differs from one mesocosm to the next, sometimes drastically, is not this coupling but its content, which entities a given mesocosm attributes as nonrecursive and which it attributes as responsive, and how much of daily coordination it organises around each attribution. This article develops that argument in full, tests it against the domains where threshold coordination is thickest, and closes by asking what the theory implies about a century of scholarship that studied these acts as failed beliefs rather than as governance moves addressed to the right kind of counterpart in the only vocabulary available for addressing it.
The scholarly literature that already exists on this material, folkloristic, anthropological, and psychological, has produced excellent description and very little architecture. Folklorists have catalogued the acts by the thousand without asking what unites them. Anthropologists since the classical period have connected them to danger and to uncertainty with real insight but no way of explaining why some of the most unpredictable domains a mesocosm faces draw none of this behaviour at all, while some of the most reliably regular ones, an opposing pitcher’s pattern of pitches, a familiar animal’s flight response, draw it heavily. Psychology has mostly treated the acts as a cognitive error to be measured, which produces a great deal of data about the size and shape of the error and almost nothing about why the error, if that is what it is, recurs in exactly this distribution rather than any other. What has been missing in each case is not more data but the one distinction all three traditions lacked the vocabulary to draw cleanly, the distinction between a counterpart that answers back and one that does not. That distinction is recursivity, and it is where this account begins.
Recursive Type as the Hidden Variable
Recursivity is categorical, not a matter of degree. A counterpart is nonrecursive when nothing about how it is engaged alters its course, selfrecursive when it responds to its own ongoing monitoring, and interrecursive when it is one of several beings each responding to the others’ responses in turn. This categorical fact does the explanatory work that danger and uncertainty cannot, because danger and uncertainty both belong to a different set of properties entirely, one that varies independently of recursive type rather than tracking it. Stability concerns how much a counterpart varies on its own account. Predictability concerns whether that variation can be anticipated. Controllability concerns whether it can be deliberately shaped. None of the three is fixed by recursive type. A bone is selfrecursive and highly stable. A heartbeat is equally selfrecursive and rapidly variable. A mountain is nonrecursive and highly stable. A storm is equally nonrecursive and only partly predictable, because its governing laws are fixed while its initial conditions are not fully trackable. The solar cycle is nonrecursive and almost perfectly predictable across the timescales that matter to a human life.
This independence is the whole reason the classical explanation of superstition, that it responds to danger or to unpredictability as such, has never quite closed the case despite a century of trying. A storm is far less predictable than the solar cycle, and it kills people the solar cycle never has, yet no fishing community propitiates the barometric pressure gradient itself, addresses it, or forbids naming it, in the way it propitiates, addresses, and forbids naming the sea. The difference lies not in how hard either is to predict but in whether the counterpart is engaged as one that might respond to being addressed at all. A nonrecursive counterpart, no matter how unstable or poorly forecast, offers nothing for a small act to address, because there is no sense in which its future course could be altered by the manner of its engagement. An interrecursive counterpart offers exactly that, whether or not the offer is real.
Governance logics track recursive type, not the other three properties, and this is the mechanism through which the categorical fact becomes visible in practice. Rules, measurements, and codified procedure achieve excellent fit against nonrecursive counterparts, because the domain holds still while being governed, and technique applied once continues to apply. The same apparatus achieves only moderate fit against a selfrecursive counterpart, because measurement changes what it measures, a body that is constantly monitored behaves differently from one that is not, and a routine has to be tuned to the very state it is meant to regulate. Against an interrecursive counterpart the fit collapses further still, because any rule introduced becomes part of the dynamics it was meant to govern, a pitcher who always throws the same pitch in the same count has not solved the batter, he has handed the batter a rule to exploit. Threshold coordination is what a mesocosm reaches for exactly at the point where this collapse occurs, at the edge where technique has done everything technique can do and something remains that might still respond to how it is approached. This is why the small act is never a competitor to technique and never a substitute for it. It occupies the residue technique cannot reach by its own nature, not the residue technique has merely failed to reach yet.
A minimal pair makes the collapse concrete. A coin toss and a handshake are both brief, both bodily, and both conventionally treated as the opening move of an encounter, yet only one of them regularly attracts a small act. The coin is nonrecursive, an inert object whose landing face depends on physical forces indifferent to how the toss is regarded, and a coin toss is governed, where it is governed at all, by convention about who calls and when, never by propitiation of the coin. A handshake is interrecursive from the first instant of contact, grip answering grip, timing answering timing, each party adjusting in real time to what the other’s hand is already doing, and handshakes across a wide range of documented cultures carry an entire apparatus of small, non-technical correction, a practised phrase, a controlled duration, a specific pressure, addressed not to etiquette in the abstract but to the other party as a responsive counterpart whose reading of the gesture is itself part of what the gesture must manage. Nobody has ever developed a small act to improve a coin’s landing, because there is nothing on the other side of a coin toss for a small act to reach.
Discernment, Attribution, and the Aspect-Specific Situation
No entity carries one recursive type as a whole. Recursive discernment, the ongoing activity through which a living being works out which aspects of an encountered counterpart are relevant and what kind of responsiveness those aspects show, operates aspect by aspect, and a single counterpart routinely presents several aspects under several attributions at once. A fisherman’s radar discerns the sea’s currents as nonrecursive, a physical field governed by depth, tide, and wind, and the discernment is entirely adequate to that aspect. The same fisherman’s silence about the word that must not be spoken on deck discerns a different aspect of the same voyage, not the current but whatever in the day’s fortune might still be listening, as interrecursive, and this discernment operates alongside the first without contradicting it. Nothing forces a mesocosm to attribute a single, uniform recursive type to an entire encounter. Most encounters that generate threshold coordination are exactly this kind of layered case, technically governed at the aspects where technique achieves fit, and addressed at the aspects where it does not.
Recursive attribution is the settled, transmitted answer a mesocosm gives to the question of what kind of responsive being a given counterpart is, and it is not the same fact as the counterpart’s actual recursivity, whatever that might turn out to be under independent inspection. Attribution is a mesocosmic fact, sedimented into posture, craft transmission, and unreflective practice long before it becomes an explicit claim anyone could defend, and it can be sustained for generations without correction if nothing in ordinary coordination forces its revision. This is the opening threshold coordination lives in. Because attribution is not pinned directly to ground truth, a mesocosm can, and regularly does, attribute interrecursive responsiveness to a counterpart whose actual behaviour would not support the attribution under controlled inspection, and it can just as easily withhold interrecursive attribution from a counterpart that would repay it. Both directions of mismatch matter for what follows, and both are ordinary rather than exceptional, because attribution answers to coordinative history rather than to a standing metaphysical audit.
One clarification protects this argument from a confusion its own vocabulary could otherwise invite. Dwelling is the paradigm nonrecursive mediation, the one mode of coordination structured by terrain, climate, and the great astronomical regularities that answer back to no one, and nothing in what follows revises that architecture. What varies mesocosm to mesocosm is not whether dwelling itself is recursive, it is not, but how a given mesocosm distributes attribution across the specific beings and domains that populate the dwelling field. A mesocosm can engage a mountain’s dwelling aspects, its mass, its elevation, its seasonal snowline, as flatly nonrecursive while simultaneously overlaying an interrecursive attribution onto the same mountain considered as a whole, a presence that can be offended by careless conduct or flattered by correct approach. The overlay borrows the responsiveness proper to being-with and extends it onto a counterpart dwelling supplies, and this extension, not a revision of dwelling’s own nonrecursive character, is what produces threshold coordination toward mountains, seas, and forests wherever it occurs. The mediation stays what it is. The attribution laid across specific entities within it is the mesocosmic variable, and it is exactly here, at the boundary between dwelling and being-with, that most of the ethnographic material on propitiated landscapes and addressed weather actually lives.
Why the Coupling Is Universal and the Content Is Not
The claim that superstition tracks recursive type rather than raw unpredictability would be uninteresting if it only held within a single tradition. It holds across every mesocosm the record allows comparison of, and the reason it holds everywhere is structural rather than cultural. Interrecursive coordination poses a governance problem no nonrecursive apparatus can solve, because the apparatus itself becomes an object the interrecursive counterpart can respond to, anticipate, and route around. This follows from what interrecursivity is rather than from any particular society’s beliefs, and it holds with the same force in a Trobriand lagoon, a Norwegian trawler, a Nevada casino, and a Tokyo concert hall. Wherever a mesocosm engages a counterpart as interrecursive, it needs something other than rule and calculation to coordinate with it, and threshold coordination, the small, cheap, repeatable gesture addressed to that counterpart directly, is one of the cheapest and most portable answers available, requiring no institution, no officiant, and no capital to deploy. This is the universal, and it is the reason threshold coordination appears in every documented human society without needing to have been transmitted from any one of them to the others.
What is not universal, and what accounts for the entire cross-cultural variation in where threshold coordination appears, is which specific entities get which attribution. Two mesocosms can face materially identical physical regularities and organise radically different coordinative apparatus around them, because they have attributed different recursive types to the same regularity. The sun rises with a predictability few natural phenomena can match, yet across a large part of documented history it was engaged as an entity that could be greeted, thanked, flattered, and, in some traditions, mourned at its setting and welcomed back at dawn, propitiations that make no sense unless the sun is attributed as a being capable of registering how it is approached. Once a mesocosm reattributes the sun as a nonrecursive stellar body, governed by orbital mechanics rather than mood, the entire apparatus of solar address becomes mesocosmically incoherent, not because anyone has demonstrated the sun will rise regardless, that was never in serious doubt under either attribution, but because there is no longer a responsive counterpart there for a greeting to reach. Nobody knocks on wood to make the sun rise in a mesocosm that has settled on the nonrecursive attribution, no matter how much residual, purely epistemic uncertainty a given observer might harbour about tomorrow. The uncertainty was never what the address was for.
Attribution’s conservatism explains why this reattribution, once it occurs, does not travel evenly through a mesocosm or arrive all at once. Attribution is sedimented into posture and craft and is resistant to casual revision, changing, where it changes at all, through sustained contestation over long stretches of time, while recursive relevance, how much of ongoing coordination actually depends on a given attribution, moves far faster, redistributed by every new technology, institution, and practice that takes over a function an old attribution once served. A fishing crew can adopt satellite forecasting, GPS positioning, and automated stability systems, each of which withdraws relevance from the sea’s attributed responsiveness for the specific tasks those systems now handle, while the attribution itself, the sense that the sea is a she with moods rather than an it with a Beaufort number, persists unrevised in exactly the register the new technology was never built to reach. This is the ordinary, well-documented pattern by which attribution and relevance move at different speeds, not a contradiction and not evidence of an incompletely modernised crew, and it leaves threshold coordination intact around whatever aspect of a domain retains interrecursive attribution even as the domain’s overall unpredictability, measured technically, falls.
The three processes that redistribute relevance, conferral, withdrawal, and restoration, give this drift a further, non-linear shape. Relevance withdrawal is what forecasting technology performs on the sea and what antisepsis performs on the surgical wound, quietly removing a domain from active coordination without touching how it continues to be attributed. Relevance conferral is what happens when a new stake enters a mesocosm fast enough to outrun any settled attribution at all, an emerging occupation, a newly dangerous technology, a game not yet old enough to have accumulated its own lore, and threshold coordination in these settings is often visibly under construction, borrowed piecemeal from adjacent domains before it has had time to sediment into anything locally specific. Relevance restoration is the least discussed of the three and the most relevant to why threshold coordination revives rather than simply persisting where it is left alone. A domain whose interrecursive relevance had truly lapsed, a form of gambling grown fully automated and impersonal, a sport whose outcomes had come to feel statistically foreclosed, can have relevance rebuilt around it by a single change that reopens the felt sense of an unresolved contest, a new format, a rivalry, a longer losing streak than the format usually allows, and threshold coordination reappears at exactly that point, tracking the restoration precisely rather than the raw technical uncertainty, which in some of these cases has not changed at all.
The Threshold and the Stake
Threshold coordination is scaled by stakes, not primarily by anxiety, though anxiety is very often the felt correlate that a stake has become salient. Stakes are mesocosmic facts about what depends on a coordination’s success, not symbolic assessments layered on top of an otherwise neutral situation, and they are distributed unevenly across the parties to any coordination. A batter’s turn at the plate and a fielder’s routine play carry unequal stakes even within the same game, because more of the game’s outcome typically hangs on the former than the latter, and threshold coordination tracks this asymmetry with some precision, concentrating where the stake is highest rather than merely where the task is hardest. This is also why threshold coordination is a phenomenon of the future rather than of the past or the timeless present. It occurs before an outcome has been fixed, at exactly the interval in which the coordination’s success or failure remains open, and it becomes pointless the instant that interval closes, no fisherman touches iron after the catch is already in the hold, no actress avoids the final line after the curtain has come down. The registration that something is now at stake and not yet settled is what LVT calls L2, the level at which threatened vitality first becomes felt before it can be fully named, and threshold coordination is one of the most immediate and least symbolically elaborated responses L2 registration can produce, a direct move to the small act rather than a passage through extended reflection.
Because stake allocation is asymmetric, threshold coordination should be expected to vary by role within a single event, not only by event, and it does. A veteran performer and a debutant sharing the same stage do not carry the same stake in that night’s judgement, since a veteran’s standing survives one poor performance in a way a debutant’s introduction to a discerning audience may not, and the density of small acts tracks this difference closely, heaviest around first appearances, title defences, and any occasion on which the coordination’s failure would cost something not recoverable by the next attempt. A ship’s captain and a deckhand share a voyage but not its stake, since responsibility for the boat, the catch, and the crew concentrates on the one role in a way it does not on the other, and the captain’s own private observances are frequently reported as more extensive and more strictly kept than the crew’s shared ones, a distribution that a purely occupational account of maritime superstition, attentive to the trade rather than to the office within it, tends to miss entirely.
Misattribution: The Wager and the Nonrecursive Exception
The theory faces its sharpest test in gambling, because a die, a shuffled deck, and a drawn lottery ball are, by any external measure, paradigm cases of nonrecursive mechanism, and yet gambling is among the most superstition-dense domains the record contains, dense with lucky numbers, blown dice, and objects carried to the table for no stated reason anyone can defend. If the coupling argued for here were exact, no small act should attach to a mechanism that cannot respond to being addressed, and the persistence of exactly this behaviour around exactly this mechanism looks, at first pass, like a direct counter-example rather than a confirming instance.
The resolution is attribution creep, and it is one of the more useful extensions this article proposes to the existing architecture. Recursive attribution is not constrained by the actual physics of a counterpart, and repeated, personally consequential engagement with any sequence of outcomes reliably generates the phenomenology of a relationship, streaks that feel earned, patches that feel owed, an object or a table that feels warm or cold toward the player, regardless of whether the underlying mechanism supports any of it. A gambler does not, typically, attribute interrecursivity to the die’s crystalline structure. He attributes it to something closer to luck itself, treated as a diffuse, personified, addressable force that can be courted, offended, or currently withheld, and the die becomes the point of contact for an address that is not really directed at the die at all. This is attribution running ahead of ground truth rather than tracking it, and it is exactly what the theory predicts will happen whenever stakes are high, outcomes are repeated, and feedback is emotionally loaded enough to be read as pattern regardless of its actual statistical structure. The honest version of this argument also has to concede that the feedback is occasionally not pure noise. Sequential dependencies in human motor performance can produce real short-run streaks in skilled tasks, a fact that took formal statistics several decades to establish correctly after an influential early analysis undercounted them. But no equivalent mechanism exists inside a fair die or a shuffled deck, and the persistence of streak-attribution there, where it cannot be rescued by any true underlying pattern, is the cleanest evidence available that attribution, not the mechanism’s actual responsiveness, is what threshold coordination answers to.
Two further predictions follow from treating gambling this way rather than as a simple exception. Threshold coordination in gambling should concentrate around the moments and objects that feel most like a counterpart with a memory, the specific table, the specific dice, the specific hand, rather than around the formally identical mechanism considered abstractly, and this is exactly the documented pattern, with rituals attached to particular sessions and particular runs rather than to gambling as a general category. And threshold coordination in gambling should decline wherever the mechanism is engineered to remove any trace of felt continuity between outcomes, a fully anonymised, single-shot digital draw with no visible object and no persisting table should draw less of this behaviour than a live table with a felt history, even where the two are statistically identical, because the felt history is precisely the material out of which misattributed interrecursivity is built.
Commercial gambling design confirms the mechanism from the opposite direction, by manufacturing the felt continuity deliberately where a fully anonymous, single-shot draw would otherwise remove it. Slot machines engineered to produce frequent near misses, a third matching symbol stopping just short of the payline, are not adding informational value, since a near miss carries no more predictive content than any other loss, but they are adding exactly the texture that supports interrecursive misattribution, a sense that the machine came close, that it is warming toward a result, that persistence is being noticed and rewarded incrementally. The same design logic explains why a machine’s superficial theme, its sounds, its visible symbols, its named characters, measurably affects how strongly it draws this behaviour even though none of it touches the underlying probability at all. What is being engineered is the attribution, not the odds.
Poiesis, Institutions, and the Zoetic Residue
Institutions and threshold coordination solve the same governance problem at opposite ends of a single continuum. Interrecursive coordination among strangers at a scale beyond direct responsiveness is the problem institutions exist to solve, and they solve it by depositing recursive coordination into durable multimaterial and multisymbolic form, a court, a contract, a calendar, an office, so that the coordination need not be renegotiated from nothing on every occasion it is required. This is poiesis working at civilisational scale, converting an interrecursive problem into something closer to nonrecursive stability by manufacturing an anchor no single interaction could have produced on its own. Threshold coordination is what remains available at the opposite end of the same continuum, at the individual, momentary, zoetic scale, precisely where poietic manufacture has not reached and in many cases cannot reach, because no institution can engineer a guaranteed hit off a live pitcher, a guaranteed catch in an unpredictable sea, or a guaranteed ovation from an audience that has not yet formed its judgement. Controllability is the operative variable here. Poiesis is the deliberate raising of a counterpart’s controllability by manufacture. Threshold coordination operates precisely where controllability cannot be raised this way, because the counterpart in question is a live opponent, a wild animal, an audience, or a diffusely attributed fortune, none of which durable manufacture can convert into an engineered anchor without destroying the very interrecursive contest the domain depends on. A stadium roof controls the weather’s relevance to a game. Nothing controls the batter’s relationship to a pitcher who is at that moment inventing his response, and this is exactly where threshold coordination concentrates, at the residue poiesis structurally cannot colonise rather than the residue it has simply not gotten around to yet.
This also explains why threshold coordination survives technological advance rather than being displaced by it, a pattern the record shows repeatedly and that a simple ignorance model of superstition cannot easily accommodate. Mesocosmic coordination is rarely secured by a single mechanism for any given stake, and redundancy, several coordinations supporting the same stake simultaneously, is one of the deepest and most protective properties a mesocosm can have. A modern surgical team uses monitors, protocols, checklists, and antisepsis, an entire poietic apparatus of manufactured control, and still, in the same theatre, forbids the word that names the calm the apparatus depends on, because the apparatus addresses the nonrecursive and selfrecursive aspects of the case, the physiology, the equipment, the procedure, while a residual, diffusely interrecursive aspect, whatever in the day’s fortune the forbidden word might disturb, remains outside anything the apparatus was built to reach. Technique and threshold coordination are not rivals competing for the same explanatory territory. They are redundant layers addressed to different aspects of the same encounter, and improving one does nothing to displace the need for the other, because the other was never answering the question the first one solves.
Aviation shows the same redundancy in an occupation built almost entirely from poietic manufacture. A modern cockpit is saturated with checklists, redundant instrumentation, and codified procedure precisely calibrated to strip interrecursive uncertainty out of as much of the flight as engineering allows, converting weather, mechanical failure, and human error into domains rule can reach wherever rule can be made to reach them. Pilots nonetheless report, with the same disavowing shrug documented everywhere else in this material, private observances kept before specific flights, a particular boarding sequence, an object carried but never mentioned, concentrated overwhelmingly around the two intervals the checklist cannot fully colonise, take-off and landing, the moments at which the aircraft’s fate depends most on a live, momentary coupling between machine and pilot that no procedure written in advance can finish specifying. The redundancy reflects the accurate placement of a small act exactly where manufactured control runs out rather than any failure of the checklist.
Levels: Why Superstition Stays Below L4
Threshold coordination operates almost entirely within L1, L2, and L3, and this restricted range is what gives it its characteristic shape. It is performed without demanding reflection, which is L1. It is triggered by a felt registration that something consequential remains unresolved, which is L2. When it is named at all, it is named locally and non-portably, my ritual for this specific match, this specific role, this specific voyage, embedded in the coordinative field that produced it rather than generalised beyond it, which is L3. What threshold coordination almost never does, while remaining threshold coordination, is consolidate into L4, the level of stabilised classification, doctrine, and institution that travels beyond the occasion that generated it. The moment a small act acquires an officiant, an exegetical tradition, a text that explains why it must be performed in exactly this sequence, and a claim to apply beyond the immediate coordinative field that first produced it, it has been deposited into durable multisymbolic form and has become something else, a rite, a liturgical element, a piece of institutionalised religion, built from the same underlying governance move but no longer identical with it. Superstition, in the ordinary sense the word is used to disparage, is precisely the L1 through L3 material that sedimentation has not yet reached, or that persistently resists reaching, remaining perpetually re-performed rather than administered.
This architecture explains the disavowal that puzzled every observer from the earliest folklorists onward, the practitioner who performs the act and denies its efficacy in the same breath. Asked to justify a piece of L4 doctrine, a practitioner can point to text, officiant, and tradition, because that is precisely what L4 consolidation deposits for later use. Asked to justify an L1 through L3 practice by the same standard, there is nothing to point to, not because the practitioner is confused about his own behaviour but because the practice was never built to answer an L4 question in the first place. The shrug that accompanies threshold coordination is the accurate report of a practice operating at a level that has no doctrine to cite rather than evidence of insincerity, delivered by someone who nonetheless knows, at L1, exactly when the moment calls for it.
The boundary is nonetheless crossed often enough that the crossing itself repays tracing, because it is where threshold coordination and institutionalised religion turn out to share a common origin rather than a merely analogous structure. A single sailor’s private habit of leaving a small offering at a shoreline rock before a dangerous crossing is pure L1 through L3 material, unexplained, locally specific, and freely varied from one practitioner to the next. Where enough sailors converge on the same rock over enough generations, the site itself begins to accumulate multimaterial deposit, a cairn, a shrine, eventually a named chapel with a keeper and a feast day, and the practice that once needed no justification acquires one, a founding story, a saint, a liturgical calendar specifying exactly when and how the offering is to be made. Nothing about the underlying governance move has changed, an interrecursively attributed hazard is still being addressed at the threshold before an uncertain crossing, but the act has been deposited into durable symbolic and material form and administered ever after rather than freshly reinvented, which is the entire difference between threshold coordination and the religion that so often grows out of it. Most threshold coordination never crosses this line, remaining exactly as local and undeposited as the sailor’s first, unwitnessed gesture at the rock, and it is this majority, not the rare case that sediments into doctrine, that the word superstition has always most precisely named.
What separates the rare case from the majority is determinacy rather than the mere presence of an interrecursive counterpart. A counterpart crosses into L4 only if it carries enough fixed characterisation, as a single persisting individual or as a recognised type, to give successive occasions of address something stable to elaborate rather than reinvent from nothing each time. The rock and the sea qualify by individuation, each a single, re-encountered being that can be greeted on this voyage and returned to on the next, which is exactly the continuity a shrine or a cult accumulates around. A counterpart can also qualify by a different route, without individuating into any single being with a history of its own: a diffusely attributed force with a fixed, recurrent mechanism, a stable textual counter, and an accumulated exegetical tradition attached to it carries the same kind of stability of type that supports doctrinal elaboration, even where no one being is ever named or encountered twice. What blocks consolidation is neither interrecursivity nor any general vagueness in how the counterpart is conceived, but the absence of any fixed characterisation at all, individual or typed. Knocking on wood, blowing on dice, and the felicitous-utterance taboo are addressed to a counterpart that is freshly generic on every occasion, this table’s luck rather than a luck with a history, and there is no persisting identity, individual or typed, for a liturgy to be about. Determinacy is therefore a categorical gate rather than a matter of degree, and an undetermined counterpart is structurally excluded from L4 no matter how much time or relevance accumulates around the practice addressed to it, exactly the position luck occupies throughout the misattribution given earlier, a diffuse, personified, addressable force with nothing fixed for any tradition to attach itself to. Once the gate is open, relevance and generational convergence, the sedimentation-time story just given for the shoreline rock, do the remaining work of deciding whether a determinate counterpart is actually consolidated into doctrine or simply remains available for consolidation indefinitely, which is the ordinary fate of most shrine-rocks, determinate enough to qualify and never converged upon by enough generations to complete the crossing.
Invisible Value and the Misfit of the Sceptical Gaze
Value is what sustains, repairs, or improves coordination, and the coordinations that most fully meet a stake are also the ones most likely to disappear from view, because they no longer require attention once they are working. Threshold coordination follows this pattern with unusual clarity. A performer whose pre-show sequence has become fully automatic experiences it as nothing, a few seconds of habit rather than an effortful intervention, and a crew whose taboo observance has been unremarkable for generations experiences it the same way, as the ordinary texture of the work rather than as a theory being tested. This muting is success, not concealment, and it is exactly the property that makes the practice so easy for an outside observer to dismiss, since an outside observer arrives at L4, looking for a claim to evaluate, and finds instead a muted L1 habit that was never making the kind of claim L4 evaluation is built to test.
This is where a century of research into superstition committed its most consequential error, and the error is precisely nameable in this vocabulary. Treating a small act as a false causal hypothesis and then measuring the size of the error applies a nonrecursive governance logic, is the stated claim true, tested by controlled trial, to a coordination that was never operating at the level such a test addresses. This is type misfit in its purest form, a nonrecursive evaluative apparatus turned on an interrecursive coordination move, and it produces exactly the failure the theory predicts type misfit will produce, not a wrong answer so much as an answer to a question the practice never asked. A small act addressed to an audience’s judgement, a prey animal’s attention, or a diffusely personified fortune was never a claim about the causal structure of dice or weather, and asking whether it is true is asking the wrong recursivity level’s question of an L1 through L3 phenomenon that was never trying to be an L4 proposition. This article is not exempt from its own diagnosis. Any account that turns threshold coordination into a stabilised, general, exegetical framework, this one included, runs the risk of performing on superstition exactly the extraction it accuses the older literature of performing, taking a live, aspect-specific, locally muted practice and depositing it into a portable theoretical category for purposes that belong to the theorist rather than to the practice. The remedy is not to stop theorising but to keep the theory answerable to the coordination it claims to describe, and to ask, of any case this framework is applied to, whether the practitioner’s own L1 through L3 sense of what the act is for has been preserved or overridden by the analysis.
The type misfit is visible in the instruments the older literature built to study the phenomenon, not only in its conclusions. A common design asks a participant to rate, on a numbered scale, how strongly they believe a described charm will cause a described outcome, then treats any answer above the scale’s zero point as a measurable trace of irrationality to be correlated against anxiety, need for control, or tolerance of ambiguity. The instrument is built to detect an L4 proposition and will detect one whether or not the practice being measured ever generated one, because a forced-choice numbered scale has no setting for I do not hold this as a claim but I would not skip it either, which is the actual, common, entirely coherent report most practitioners give when asked in less constrained terms. The measured belief the instrument returns is frequently an artefact of the instrument’s own insistence on a propositional answer, manufactured at the moment of asking rather than discovered underneath the practice, and a discipline that studies threshold coordination exclusively through instruments of this kind will keep finding confused believers because its method has no way of finding anything else.
Recursive Density Across Six Thresholds
The theory earns its keep by discriminating correctly inside domains that a cruder danger-based account collapses together. Six sites make the discrimination visible.
The stage carries the heaviest documented load of threshold coordination anywhere in the ethnographic and folkloristic record, and the reason is structural rather than accidental. An audience is a live, evaluating, interrecursive counterpart whose judgement forms in real time and responds to the performance as the performance unfolds, and the performer’s own nervous system is a selfrecursive counterpart whose composure is itself part of the instrument being used. Nothing technical remains to be done in the interval before curtain, every line has been rehearsed, every cue has been set, which means the entire remaining unresolved stake sits with two counterparts technique cannot reach further, the audience and the self, and threshold coordination concentrates exactly there because there is nowhere else left for it to go. The prohibition on speaking the final line before an audience has assembled, and the wider prohibition on wishing a colleague good fortune directly rather than through an inverted formula, are addressed to precisely this counterpart, treated as capable of being tempted into an unfavourable response by a presumption voiced too soon, which is the felicitous utterance problem in its cleanest form, a positive wish withheld because the interrecursive counterpart it would reach is attributed as one that can take offence at confidence it has not yet earned.
Fisheries carry a layered case rather than a uniform one. The sea’s currents and depths are increasingly attributed nonrecursive and governed by instrument, while the sea considered as a whole retains, in many maritime mesocosms, an attribution far closer to interrecursive, a she with moods rather than an it with a chart, and the fish themselves are actively interrecursive counterparts, evasive, learning, responsive to the specific gear and behaviour of the crew hunting them. Add the crew’s own interdependency under confinement, an interrecursive social coordination in its own right, and the dense apparatus of word avoidance, counter-gesture, and taboo that surrounds fishing communities becomes exactly what the layered attribution predicts rather than a mystery instrumentation should have dissolved. The specific content of the taboo, which word is forbidden, which animal must not be named, tends to be locally arbitrary in exactly the way L1 through L3 material generally is, varying from harbour to harbour along a coast that shares the same instruments, the same fish stocks, and the same weather, which is itself evidence that what is being tracked is attribution and its local history rather than the technically measurable riskiness of the work.
Hunting sharpens the same point further, because in a large part of the ethnographic record the prey is not merely attributed interrecursive status informally, it is explicitly engaged as a subject, one whose attention, notice, and response the hunter must actively manage. The widespread prohibition on stating one’s expected success before the hunt follows directly: naming the outcome addresses an interrecursive listener, the animal, or a diffusely personified fortune standing behind the animal, and the prohibition is not a superstition about probability, it is a precisely targeted piece of etiquette toward a counterpart the mesocosm has attributed as capable of resenting presumption. Divinatory practices that precede a hunt belong to the same register rather than to a separate category of technical forecasting, since what they typically supply is not a prediction to be checked against outcome so much as a randomised reorientation of the hunt itself, addressed to a domain understood as responsive to the hunter’s own pattern of approach, and their persistence alongside detailed practical knowledge of the terrain and the animal’s habits is further evidence that they are answering a different, interrecursive question rather than competing, and losing, against ordinary technical skill.
Sport supplies the clearest internal contrast, because a single occupation contains both a strictly interrecursive task and a largely selfrecursive one performed by the same practitioners in the same afternoon. Batting against a live pitcher who is inventing his response in real time is interrecursive in the strict sense, and it draws heavy threshold coordination. Fielding a routine ball is a well-drilled motor sequence against a nonrecursive trajectory, closer to selfrecursive competence than to any responsive counterpart, and it draws almost none. The same players, the same stakes in the broadest sense, the same afternoon, and the divergence tracks recursive type with a precision that danger or importance alone could never produce, since both tasks share the same game, the same crowd, and the same outcome riding on them. The same contrast reappears inside individual sports that combine a technical and an adversarial phase within a single discipline, a gymnast’s landing, largely selfrecursive once the routine is set, draws less threshold coordination than a tennis player’s return of an opponent’s serve, an interrecursive exchange in which the returner is reading and responding to a live decision made a fraction of a second earlier by someone actively trying not to be read.
Gambling has already had its resolution given in full above, misattributed interrecursivity built out of repeated, emotionally loaded engagement with a mechanism that is, by any external measure, nonrecursive, and its inclusion here is mainly to note that it remains, correctly understood, a confirming case rather than an exception, since what draws the small act is never the die’s physics but the felt counterpart the gambler has attributed behind it. Card games that mix a strictly interrecursive element, reading and misleading an opponent, with a nonrecursive one, the composition of a shuffled deck, show threshold coordination concentrating on the former and almost never on the latter within the same hand, a player will manage a rival’s attention and posture with real care while treating the deck itself as beneath address, a division a purely occupational label like gambling superstition obscures by naming the setting rather than the counterpart.
The operating theatre closes the set because it shows attribution splitting across levels within a single mesocosm rather than across different mesocosms. Biomedicine’s official L4 doctrine attributes the patient’s body as thoroughly nonrecursive and mechanistic, governed by protocol, and this attribution is entirely sincere and entirely load-bearing for the apparatus that makes modern surgery survivable at all. At L1, in the practitioner’s own unreflective practice, a residue persists that the official doctrine was never built to address, whatever in a given case’s outcome remains still open despite correct technique, and the forbidden word operates exactly there, in the gap between an L4 attribution the whole institution depends on and an L1 registration the institution’s own vocabulary has no room for. The same split explains why the taboo is reported almost exclusively among the staff physically present at the point of care, nurses, junior doctors, the anaesthetist watching the monitors in real time, and only rarely among administrators and researchers whose relationship to the same cases is mediated entirely through the L4 record, since it is proximity to the unresolved moment itself, not the underlying medical uncertainty on paper, that the practice tracks.
Diagnostic Postscript
Superstition, read through this architecture, stops looking like an error and starts looking like exactly what it has always been doing, a governance move correctly addressed to a counterpart engaged as responsive, deployed at the threshold before an outcome is fixed, redundant with whatever technique the domain has also managed to build, and muted into invisibility whenever it is working well enough not to need defending. What a mesocosm attributes as nonrecursive it governs by rule. What it attributes as selfrecursive it governs by tuned self-monitoring. What it attributes as interrecursive, an opponent, an audience, a prey animal, a diffusely personified fortune, it cannot govern by rule at all, and reaches instead for the small, cheap, unexplained gesture this article has called threshold coordination, performed and disavowed in the same breath because it was never built to answer to the level at which disavowal operates. The question this leaves for any specific case is not whether the practice is true, which was never the question it was answering, but which counterpart it is addressed to, whether that attribution tracks anything the counterpart could plausibly do, and whether the coordination it maintains would survive being taken away. Where the answer is that it would not, that removing the small act would leave the performer, the crew, or the practitioner measurably less able to face the threshold in question, the practice has already answered the only question threshold coordination was ever built to answer, and no further audit at L4 is owed to it.