Abstract

Predictability is commonly treated as a property of processes, or as the success with which their future states can be represented in advance. This article proposes a different account. Drawing on the theory of comparison developed in Living Worth, it argues that prediction depends upon establishing relevant similarities between anticipated and realized developments. Ordinary-language comparisons, numerical thresholds and exact equivalence impose radically different conditions of success, so that a future may be highly predictable at one level of symbolic specification and apparently unpredictable at another, without any change in the process itself. The article further distinguishes the recursive responsiveness of anticipated counterparts from the practical stabilization of future coordination. Nonrecursive, selfrecursive and interrecursive counterparts differ in how anticipation can enter their unfolding, while preparedness, habits, commitments and infrastructures can make many possible outcomes functionally equivalent for ongoing coordination. Through examples drawn from weather, embodied action, boxing, relationships and institutional arrangements, predictability emerges as a relational and practically constituted achievement. The highest accomplishment of future intelligence is often the creation of conditions in which exact prediction becomes unnecessary.

Introduction: The Second Ball

Someone stands in a field holding three tennis balls. They throw the first as far as they can, and it lands some distance away in the grass. They now undertake to throw the second ball less far than the first. Nothing could be easier. They ease off, shorten the swing of the arm, and make the anticipated relationship overwhelmingly likely. Then they attempt to throw the third ball exactly as far as the first. What seemed an elementary task suddenly becomes almost impossibly demanding. With some luck they might land within a metre of the first ball, and with practice they might narrow the gap further. Reproducing the same continuous physical magnitude, the same distance without remainder, belongs to a different order of difficulty altogether.

The experiment was introduced in Living Worth to show that establishing directional differences is far easier than establishing exact equivalence (Ecks 2022, 50). It was directed there against the habit of treating “greater than”, “less than” and “equal to” as three equally straightforward comparative relations, when in practice the third is enormously more demanding than the other two. The experiment contains a second lesson, which its first use left implicit, and that lesson concerns the future.

Was the second throw predictable? Of course it was. But what, precisely, was predicted? Neither a numerical distance, nor a trajectory, nor a particular landing point. What was predicted was a relationship between two outcomes: the second throw would travel a shorter distance than the first. That prediction could be fulfilled by indefinitely many landing points, anywhere between the thrower’s feet and the first ball. Now consider the third throw. If the task is to reproduce the first distance exactly, the set of acceptable outcomes contracts towards a single value. Nothing about the physics of the ball has changed, and gravity, air resistance and the give of the grass are what they were. Nor has the thrower necessarily become less capable. What has changed is the specification of the anticipated result, and with it the comparison through which success is established.

This introduces the distinction on which the article depends, between the actual variability of a process and the differences that a prediction requires us to discriminate. The ball lands somewhere different every time, whatever the thrower intends. Whether this variability matters depends entirely on what the prediction asked for. Predictability is therefore never independent of what counts as the same, as sufficiently similar, or as relevantly different. We do not simply predict the future. We predict that some specified feature of a future development will resemble, differ from, or stand in a particular relationship to a symbolically specified expectation. The second throw is predictable because a very large range of outcomes counts as fulfilling the anticipation; the third is hard to predict exactly because its criterion of success admits almost no variation.

From this follows the most consequential proposition of the article. Predictability does not require the future to unfold exactly as anticipated; it requires that the differences between what was anticipated and what occurs are sufficiently irrelevant to the coordination at stake. This is more radical than saying that some predictions are approximate and others precise. It means that predictive success depends on which differences have been selected as relevant, how those differences are symbolized, and whether living coordination can be maintained across the outcomes regarded as sufficiently similar. It also opens a second possibility. Living beings do not merely discover that some futures are more predictable than others. Through habits, preparedness, commitments, institutions and infrastructures, they can make a range of possible futures functionally equivalent for the purposes of continuing coordination. This can be called the production of predictive equivalence. The expression needs care: the futures do not become identical, and uncertainty does not disappear, but some of their differences cease to matter for a particular anticipated coordination.

If so dramatic a difference in predictability can be generated merely by changing the required comparison, what becomes of the assumption that predictability is an intrinsic property of the process itself? The question “How predictable is this?” turns out to be incomplete. It must first be asked: predictable as what? As a directional difference, as membership in an ordinary-language category, as remaining within a numerical interval, as the exact repetition of a measured magnitude, or as the continued viability of a particular living coordination? Until that question is answered, assigning a general degree of predictability to anything is conceptually premature.

I. Prediction Is a Comparison

The argument extends the theory of comparison set out in the second chapter of Living Worth, “Relative Value: Culture, Comparison, Commensurability” (Ecks 2022). Its starting point is that comparisons do not simply discover similarities waiting to be recognized. Any two things are similar and different in indefinitely many respects, and comparing them meaningfully requires selecting which respects matter. Nelson Goodman made the point with characteristic economy. Similarity, he argued, cannot be fixed by the features two things happen to share, since anything shares some features with anything else; it becomes determinate only relative to a selection of relevant respects (Goodman 1972). A comparison becomes meaningful within an established practice that supplies that selection.

Goodman’s story of the upholstery sample makes this vivid (Goodman 1978). A customer sends a fabric swatch to a supplier and orders enough material, exactly like the sample, to cover her furniture. The supplier delivers the fabric cut into many small pieces, each with the size and the zigzag edges of the swatch. A similarity has been reproduced exactly, but it is the wrong similarity. The swatch exemplified colour, weave and texture, while its size and the shape of its edges were incidental, and the practice of ordering upholstery fabric made that obvious to everyone except the supplier. Precision has been achieved without practical relevance.

A prediction is a special kind of comparison. One of its terms is present now, as a symbolic anticipation. The other becomes available only as the anticipated process unfolds. A prediction is evaluated by asking whether the realized development satisfies the relevant comparison, and the criterion of that comparison cannot be supplied by the event alone. Suppose someone predicts that tomorrow will be much like today. Similar in temperature? In rainfall? In light? In the possibility of working outdoors? The predicted resemblance must be specified through some criterion. Ordinarily this poses no difficulty, since living coordination supplies the criterion without explicit elaboration: a farmer who says tomorrow will be like today knows which features of the day matter for the work in hand. Science, however, cannot assume that the relevant similarity is self-evident when it sets out to measure predictive success.

Ferdinand de Saussure gave a famous example of sameness of exactly this kind. The 8.25 express from Geneva to Paris is the same train every evening, although its locomotive, carriages and crew may all differ from one day to the next (Saussure 1959). A passenger who predicts that the 8.25 will be waiting at the platform is predicting the recurrence of something whose identity is fixed by its place in a timetable and a network, whatever happens to the steel. The prediction succeeds if a train with that place in the system departs at about that time, and it would be absurd to count it as failed because a different engine pulled it. The criterion of sameness is supplied by the coordination the train serves, and no inspection of the train alone could reveal it.

Living Worth also proposed replacing the familiar opposition between qualitative and quantitative comparison with a distinction between numerical and nonnumerical comparison (Ecks 2022, 44–46). People compare quantities constantly without assigning numbers. One person is taller than another, one journey takes longer than another, one throw travels farther than another. These are comparisons of degree, and they need no numerical specification to be correct. Stephanie Solt’s (2016) analysis of “most” and “more than half” shows how finely ordinary language handles quantity without numbers. The two expressions look logically equivalent, yet they place different demands on the underlying measurement: “more than half” requires a ratio scale, on which halving is defined, whereas “most” can be interpreted on weaker scales that support little more than ranking. Speakers can therefore establish relationships between quantities without committing themselves to explicit numerical thresholds. The implication for anticipation is direct. Everyday prediction can be remarkably reliable without being mathematically exact, because what it predicts is often a nonnumerical relation of degree.

Predictions also have purposes. Someone deciding whether to take a coat needs a different prediction from someone calibrating a precision instrument, and someone catching a train needs a different prediction from someone constructing the timetable. Neither prediction is inherently more scientific than the other. What matters is whether the selected comparison is appropriate to the purpose for which it is made. Here the argument extends the concept of biocommensuration from Living Worth into the analysis of anticipation. The selection of relevant similarities is constrained by what matters for the living coordination concerned, and therefore far from arbitrary. The coat-taker’s comparison is anchored in a body that will be cold or not; the instrument-maker’s in a device whose tolerances will be met or not. Prediction can thus be defined as a temporally extended act of comparison whose successful completion depends upon a relevant criterion of similarity or difference, and whose relevance is set by living stakes.

II. How Exact Must the Future Be? Scottish Weather and Symbolic Resolution

Scottish weather is conventionally regarded as notoriously hard to predict. But is it? Someone who has lived through a few winters on the east coast of Scotland can be extremely confident about the broad character of a November day in Edinburgh. It will be cool. It may well be windy. The light will go by mid-afternoon. Any length of time outdoors will require suitable clothing. These anticipations are highly reliable, reliable enough to organize a life around. Predicting the exact temperature at a specified street corner at a specified minute is much more difficult. The contrast does not arise because the weather has two ontologies, one tame and one wild, but because the two predictions require different degrees and kinds of differentiation. Even the proverbial changeability of Scottish weather is itself predictable at the right resolution. A forecast of “sunshine and showers, some heavy, with brisk winds” will very often be correct for an entire day in which no single hour could have been forecast with confidence. Changeability, specified as a category, is among the most reliable things one can say about the weather of the North Sea coast, and it carries a clear practical implication: take a layer that can go on and come off. What looks like a failure of prediction at one resolution is a success of prediction at another.

The concept of symbolic resolution names this difference. Symbolic resolution is the degree of differentiation demanded by the symbolic specification of an anticipated outcome. “Tomorrow will be cold” ordinarily admits a wide range of physical conditions. A forecast of four to six degrees Celsius admits a narrower range. A forecast of 5.8 degrees at two o’clock on Princes Street is narrower still. None of this implies that ordinary language is necessarily vague and mathematics necessarily precise. Mathematics offers intervals, inequalities, probability distributions and explicit error margins, all of which specify permissible variation rather than eliminating it, and ordinary language can demand exactness, as when a joiner is asked for a shelf that fits an alcove. Different symbolic systems provide different means of specifying and evaluating permissible variation, and every prediction, in whatever system, sets a resolution.

A simple result follows, with considerable consequences. When predictions concern nested sets of possible outcomes, narrowing the acceptable set can never increase the probability that the outcome falls within it, given the same underlying distribution. A forecast that tomorrow’s temperature will fall between four and eight degrees is easier to satisfy than one that it will fall between 5.99 and 6.01 degrees, because every outcome satisfying the second also satisfies the first. Mathematically this is trivial; philosophically it is significant. Increasing the exactness required for a prediction to count as successful makes success less likely without changing the physical process in the slightest. The weather has not become more chaotic because the forecaster demanded more of it.

The science of forecasting already draws the relevant distinctions, and they are instructive. A probabilistic forecast can be sharp, concentrating its probability within a narrow range, or broad. It can be calibrated, in the sense that events it assigns a given probability occur with that frequency, or miscalibrated. The standard aim of forecast verification is to maximize sharpness subject to calibration (Gneiting, Balabdaoui and Raftery 2007). That formulation already concedes the point at issue: precision is a virtue only so long as it does not outrun what the process allows the forecaster to honour. The precision of a forecast, the uncertainty it represents and its actual predictive performance are three different things, and conflating them produces the impression that processes have a single predictability waiting to be measured.

Precision is valuable only relative to purpose. A person deciding whether to walk up Arthur’s Seat gains nothing from knowing whether the temperature will be 5.7 or 5.8 degrees, whereas a council deciding whether to grit the roads near freezing point may care a great deal about differences of that order. Some exact predictions are perfectly legitimate and spectacularly successful. The timing of a solar eclipse can be forecast to the second years in advance, because the celestial mechanics involved are nonrecursive and exceptionally well characterized, and because the purpose of the prediction demands that resolution. Discrete, well-defined outcomes can likewise be predicted exactly when the process allows it. None of this makes measurement subjective or arbitrary. Measuring requires selecting a feature, a scale, a resolution and an acceptable uncertainty for a particular inquiry, and the difficulty never lies in mathematical precision as such, only in failing to establish which precision a given inquiry requires. A process cannot be assigned a single degree of predictability independently of the symbolic resolution at which its future is specified. This does not erase real differences between processes. It identifies a condition that must be satisfied before those differences can be compared.

III. Anticipation Before and Beyond Symbolization

The argument so far could be misread as a theory in which living beings anticipate the future only through language or number. Living coordination is anticipatory through and through. Someone catching a ball does not first calculate its trajectory and then command their arms to move; their embodied coordination is already oriented towards where the ball is going, and their hands arrive there as the ball does. A dog that waits by the door at the hour its owner usually returns anticipates an arrival without constructing any articulated prediction of it. When the owner is late, the dog’s restlessness shows that something anticipated has failed to occur, which is to say that the anticipation was there all along, operating without having been stated. Anticipatory coordination is neither exclusive to humans nor dependent on multisymbolization. This matters for LVT in particular, since the three zoetic mediations, multisensorial embodiment, multispecies being-with and multiversal dwelling, are shared by living beings generally and cannot depend on the specifically human development of symbolic prediction.

It is therefore useful to reserve the term explicit prediction for anticipations that have acquired a communicable or otherwise stabilized specification against which outcomes can be assessed. This need not involve mathematical notation; “it will rain before lunch” is an explicit prediction. Anticipation in the broader sense can remain embodied, relational and spatially organized without ever being stated. The recursivity levels clarify the relation between the two. At L1, much of life proceeds through anticipated continuities that never become explicit problems: the stair is where the foot expects it, and the friend answers as expected. At L2, something anticipated fails to occur, and a snag is felt before anything is said about it. At L3, the difficulty is articulated. At L4, general predictions and predictive models are stabilized, as in a numerical weather model or a proverb about red skies. At L5, the adequacy of the model, or the conditions of prediction themselves, become objects of reflection. These are analytically distinguishable possibilities rather than an obligatory sequence. Most anticipation never leaves L1, and a forecast can be issued at L4 without anyone having felt a snag. Conversely, a snag at L2 can be resolved at L2, by a stumble corrected in mid-step or a glance that finds the friend after all, without the anticipation that failed ever being put into words.

LVT’s account of the three modes of multisymbolization sharpens what happens when anticipation becomes explicit. Symbols can operate as mesocosmic coordination, as reference to absence, and as re-presentation. An explicit prediction typically operates as reference to absence, since it specifies a development not yet available as an accomplished event. Yet the prediction itself is present, and it can immediately enter coordination. A weather warning leads someone to cancel a journey. A medical prognosis changes how someone arranges the rest of their life. A prediction about another person’s behaviour, once that person learns of it, changes how they respond. The symbolic reference to an absent future can be causally active in the present.

A single forecast can therefore meet different types of counterpart at once. An amber warning for snow leaves the snow entirely unaffected, while it changes what thousands of drivers do the next morning. The roads may then be emptier and, in one respect, more predictable than they would otherwise have been, precisely because the forecast entered the coordination of those who heard it. The same symbolic act is idle with respect to the weather and consequential with respect to the traffic. The future is anticipated in living coordination before it is articulated in predictions, but once anticipation becomes symbolically specified, it becomes available to recognition, disagreement, revision and intervention by whatever is able to respond to it. That is why the three types of recursive counterpart differ so fundamentally.

IV. The Gradient of Recursive Predictability

A mountain does not alter its geological development because somebody predicts its persistence, and a storm does not change course because meteorologists forecast its track. These processes can be enormously complex. Their dynamics include nonlinearities and sensitive dependence on initial conditions, which is why the Atlantic depressions that sweep across Scotland defeat exact forecasts beyond a week or so. Yet their unfolding does not depend on any recognition of the content of predictions about them. That is the relevant property of nonrecursive counterparts, and it must not be confused with a claim that such processes are easy to forecast. A nonrecursive counterpart can be chaotic, opaque or poorly understood. What it cannot do is respond to a forecast as a forecast.

Selfrecursive counterparts are different. Consider someone anticipating an examination. They may expect to feel anxious, become anxious about that very possibility, and thereby make the anticipated response more likely. Or they may recognize the risk, prepare differently and reduce it. In either case their anticipation becomes a constituent of the process anticipated. The thrower of the second ball is a simpler case of the same structure: they formulate an anticipated relationship and adjust their action to produce it. Selfrecursivity can therefore produce predictive fulfilment as readily as predictive disruption. What distinguishes it from nonrecursivity is the capacity of the process to change through its own anticipation, whatever the consequences for forecasting accuracy.

Interrecursive counterparts add a further structure, in which anticipations encounter anticipations. In a romantic relationship, each partner’s actions respond to the other’s actions and expectations, including expectations about their own expectations. In a boxing match, opponents attempt to anticipate and exploit each other’s responses, and each knows the other is doing so. In a financial market, participants respond to the anticipated responses of others, and a widely shared forecast can transform the conditions that made it plausible, as when a predicted run on a bank becomes the run. Here a prediction becomes one element among several mutually responsive anticipations. It can reinforce an expected outcome, undermine it, or generate a situation that nobody predicted. This is a causal structure of a different kind from the mountain’s.

Social science has long recognized parts of this structure. Robert Merton’s account of the self-fulfilling prophecy began from precisely the case of a solvent bank brought down by a rumour of its insolvency, as depositors who believed the prediction acted on it and made it true (Merton 1948). John Maynard Keynes compared professional investment to a newspaper beauty contest in which competitors must pick the faces that other competitors will pick, so that each is occupied with anticipating what average opinion expects average opinion to be (Keynes 1936). Both observations are usually treated as curiosities of collective behaviour. The gradient shows why they are structural. Self-fulfilling and self-defeating prophecies are possible only where the counterpart can receive the prediction and respond to it, and higher orders of anticipation are possible only where several such counterparts respond to one another. No forecast of a storm could ever fulfil itself in Merton’s sense, however widely it was believed. A forecast about depositors, voters or partners always can, provided it reaches them.

The three types of counterpart thus form a gradient, defined by increasing scope for prediction-sensitive recursive transformation. Nonrecursive counterparts do not respond to a prediction as such. Selfrecursive counterparts can alter their own response through their anticipation of it. Interrecursive counterparts involve mutually responsive anticipations, each of which can alter the others. The gradient orders possible routes of feedback between anticipation and the anticipated, and it does not rank processes by forecasting error. A tide table, predicting a nonrecursive process, can be astonishingly accurate decades ahead, while a weather forecast, predicting another nonrecursive process, degrades within days; a couple’s anniversary dinner, an interrecursive arrangement, may be more reliable than either. Forecasting difficulty cuts across the gradient, because it depends on dynamics, information and resolution as much as on how anticipation can enter the process. This is the form in which it should be retained as foundational to LVT, because it is grounded in the constitution of processes rather than in observations that some behaviour happens to be hard to forecast.

What looks at first like a counterexample confirms the point. Two partners in a long relationship may anticipate each other remarkably well, often better than either can anticipate next week’s weather. That predictability exists precisely because their expectations have become mutually coordinated over years. Drivers on a motorway reliably anticipate that other drivers will keep to their lanes, and they can do so because everyone anticipates everyone else’s compliance. Such cases show that interrecursive feedback can be stabilizing: a shared expectation helps produce the behaviour expected. The stability remains recursive, however, since it depends on the continued responsiveness of the participants. A partner can decide to stop being predictable, and a driver can swerve. Recursivity increases the routes through which anticipation can transform its referent without necessarily decreasing the reliability of coordination.

Structural and empirical claims must therefore be kept apart. Structurally, the three types of counterpart differ in how predictions can influence the processes anticipated. Empirically, whether one kind of process is harder to forecast than another depends on the selected outcome, the temporal horizon, the symbolic resolution, the established arrangements and the accessibility of relevant information. One could investigate whether comparable predictions become more vulnerable to recursive transformation as one moves from nonrecursive to interrecursive counterparts, and the hypothesis is plausible. Such research would need to hold the comparison itself stable. A broad forecast about whether a marriage will last ten years cannot meaningfully be set against an exact forecast of tomorrow’s temperature in Leith without attending to their different criteria of success. Kept in its structural form, the gradient explains both the instability and the extraordinary stability of recursive living processes.

V. Prediction, Preparedness and Anticipation: What Rocky and Drago Teach

Rocky IV offers a compressed laboratory for these distinctions. A reading of the film in terms of three modes of future intelligence distinguished prediction, which exploits sufficiently stable expectations; preparedness, which builds capacities for managing variation in future events through prior practice; and real-time anticipation, the flexible interrecursive responsiveness required where established expectations and prepared responses prove inadequate (Ecks 2026a). That reading proposed that the modes differ in metabolic cost, with prediction cheapest and most durable and anticipation the most costly, and it interpreted the film as a confrontation between reliance on measured performance and embodied flexibility built through collective training. The theory of comparison developed here allows the reading to be made more exact.

Drago’s mistake does not lie in numerical precision itself. His trainers measure oxygen uptake, recovery time and, as the headline figure, a punch force of 2,150 pounds per square inch. These measurements may well be accurate. The error lies in the comparison they are taken to settle. A reliably measured punch force is no reliably predicted victory. The effect of a blow on a stationary, instrumented target belongs to a largely nonrecursive configuration, while the outcome of a boxing match belongs to an interrecursive one, in which the target moves, reads, and adapts to the puncher’s attempts to read it. The Soviet programme is thus a case of conceptual overcommensuration, in which a possible comparison, between Drago’s numbers and the numbers of other fighters, is mistaken for explanatory relevance to the outcome of the encounter. The programme has not measured nothing. It has assigned excessive explanatory weight to what it measured, at a symbolic resolution suited to the wrong question.

Rocky, by contrast, does not have to anticipate every possible punch in advance. He needs embodied capacities for responding across a wide range of possible encounters, and training increases the number of attacks that can be met through effective coordination. This is the principle of the second ball transposed into a body. The film’s training montage makes the contrast between the two programmes visible in their relation to dwelling. Drago trains on treadmills and instrumented machines, in conditions designed to remove variation so that each repetition can be compared with the last. Rocky trains in snow, on uneven ground, hauling logs and climbing in darkness, among nonrecursive conditions that never present the same footing twice. Whatever the film intends, his training builds capacities for variation by exposing the body to variation, whereas Drago’s builds peak performance under standardized conditions that the fight will not reproduce. Many different future events can satisfy the relevant practical criterion, which for the boxer might be keeping balance, avoiding decisive injury and remaining able to respond. The specific movements required differ from exchange to exchange, while the practical outcome remains sufficiently similar. Preparedness of this kind does not make the opponent’s precise actions more predictable; it makes their differences less consequential for continued coordination. Here the distinction between predictive success and practical resilience becomes decisive. A fighter who slips a punch he never saw coming has not predicted it. Preparedness can make a broad class of outcomes manageable without establishing the accuracy of any narrower forecast, and the two achievements must not be credited to one another.

Preparedness also does not eliminate the need to change responses. Repeated training establishes habits, and Loïc Wacquant’s (2004) apprenticeship in a Chicago boxing gym showed in detail how such habits are manufactured collectively, through drills, sparring partners, a coach’s eye and the gym’s daily rhythm, rather than possessed as an individual cognitive capacity. Effective preparedness also supports the ability to adapt when the familiar response is no longer appropriate. Rocky’s success depends on accumulated preparedness and on the revision of coordination during the fight itself. The boxer moves between established L1 coordination and moments of disruption that require heightened attention, adjustment and occasionally reflective correction between rounds. Successful training enlarges the range of situations in which coordination can be maintained or recovered, rather than eliminating such transitions, and this capacity to move between levels is what LVT calls recursive fluidity.

The question of cost can now be stated more carefully. Where reliable arrangements exist, living beings need not continually inspect every possibility. Where preparedness is effective, they need not invent every response from scratch. Where neither suffices, effortful real-time adjustment is required. The formulation of anticipation as a uniformly costly last resort needs qualification, because much anticipatory responsiveness runs seamlessly at L1: the catcher’s hands, the timing of a conversation, the walker’s adjustment to a slippery kerb. What becomes especially costly is sustained, effortful adaptation under unresolved disruption, the fourteenth round rather than the first step. Preparedness does not predict which future will occur. It expands the range of futures through which living coordination can continue, and that is a form of future intelligence that no amount of exact forecasting can replace.

VI. Making Futures Reliably Similar: From Habits to Institutions

The argument now moves from the symbolic specification of futures to the actual construction of reliable future coordination. Two routes to predictive equivalence must be distinguished. The first is symbolic: a range of different outcomes is specified as sufficiently similar for a particular prediction, so that five degrees and six degrees both count as a cold day. The second is coordinative: practices or arrangements are established that permit similar practical results across otherwise different circumstances. An experienced traveller can secure a bed for the night by any of several routes, and a reliable payment system allows strangers to exchange goods without anticipating each other’s decisions. Symbolic specification changes which differences count as relevant, whereas coordinative stabilization changes what living beings can reliably accomplish across those differences. Together they explain how future coordination can become more reliable without the future becoming more uniform. The second route also blocks a misunderstanding. The production of predictive equivalence is no symbolic trick, a matter of redescribing outcomes until they all look like success, because it rests on real capacities and arrangements under which differences in future developments become less consequential.

Habits are the most elementary form of the second route. A person who has learned to walk over rough ground does not need to identify the position of every stone. Their embodied skill makes a wide range of minor variations manageable, and the result is greater reliability of ongoing coordination without any gain in forecasting accuracy. Habit was already central to the account of value in Living Worth (Ecks 2022). In LVT terms, habits stabilize expectations and responses within a living being’s established relations to its mesocosm, and they make most of future coordination possible without its detailed course ever being specified.

Multimateriality offers the same achievement in material form. Return to the November day in Edinburgh. A good waterproof jacket, a warm layer and boots that keep out the wet make a wide range of possible weathers equivalent for the purpose of walking to work. The forecast becomes less important because the clothing has done in advance what the forecast could only have advised. The saying that there is no bad weather, only the wrong clothes, compresses the principle exactly: the weather remains as variable as ever, while its differences have been made largely irrelevant to a particular coordination. Nothing has been predicted more accurately. The range of futures through which the walk can go ahead has been enlarged.

Institutions extend the same achievement into interrecursive coordination at scale. A reliable currency is particularly revealing. Its usefulness depends on everyone expecting that others will continue to accept it, and these expectations contribute to its continued acceptance. The arrangement is interrecursively constituted through and through. Yet it can become so stable that ordinary transactions proceed without anyone considering whether the money will be accepted. A thoroughly interrecursive process can thus acquire extraordinary practical predictability. Nothing here passes from interrecursivity into nonrecursivity, since the currency remains exactly as dependent on mutual expectation as before; what has happened is the stabilization of interrecursive coordination. The point refines the claim that powerful infrastructures work like terrain, fixing pathways so that outcomes follow before any encounter begins (Ecks 2026a). Infrastructures can come to feel like terrain without ever becoming terrain. Their recursive dependencies are stabilized well enough to make individual variations inconsequential for ordinary outcomes, and their apparent solidity remains a continuing accomplishment.

The activity involved can be called predictability-making, as distinct from predicting. Predicting specifies an anticipated development. Predictability-making establishes conditions in which a desired class of developments becomes more reliable. A railway timetable is a prediction about train movements, while a functioning railway, with its signalling, staff, rolling stock and maintenance regimes, is an arrangement that makes those movements likely. A booking confirmation expresses an anticipated arrangement, and the institutions of reservation, payment and hospitality help make it reliable. A promise anticipates a future action, and the mutual commitment it creates helps produce the action promised. In each case anticipation is accompanied by activity that alters the conditions of its own fulfilment.

Narratives tend to miss this. They dwell on dramatic acts of adaptation and pay little attention to the mundane arrangements that make such acts unnecessary, a pattern that has been called heroic misattribution (Ecks 2026a). The most reliable infrastructures become invisible precisely because they support successful L1 coordination. People notice the reservation system when it fails, the currency when it collapses, the traffic rules when someone breaks them. The coordinative achievements that make futures reliable are most apparent through their disruption, which is one instance of a general feature of living coordination: what works leaves few traces, and what fails is what gets articulated.

The argument must not be overstated. Making futures reliably similar does not make them identical, and it does not guarantee that the arrangements will keep working. Stabilization can introduce dependencies, vulnerabilities and new forms of fragility. A traveller who relies entirely on a single booking platform is superbly coordinated while it works and poorly prepared when it fails, and a currency whose stability rests on widespread confidence is exposed to rapid recursive destabilization once confidence falters. Predictability-making can therefore generate new forms of unpredictability, which follows directly from the dependence of coordinative reliability on the conditions maintaining it. The strongest form of future intelligence is often, then, the organization of living coordination so that many possible developments lead to outcomes sufficiently alike in the respects that matter, rather than a more exact representation of what will happen.

VII. What This Changes for the Science of Predictability and Uncertainty

The implications for psychological science are considerable. Research on intolerance of uncertainty asks whether people can bear not knowing, frequently without establishing which anticipated outcomes are at issue, why they matter, and what would count as coordinating successfully in their presence (Ecks 2026b). A person may be entirely content with very incomplete information about tomorrow’s weather and distressed by an unanswered message from a partner. Neither experience establishes a general quantity of uncertainty that the person tolerates. The relevant differences concern what is anticipated, at what resolution, with what stakes, in relation to what type of recursive counterpart, and with what possible continuations of coordination. The weather is a nonrecursive counterpart for which a coarse resolution usually suffices; the partner is an interrecursive counterpart whose silence may bear on a coordination of the highest stakes. To place both on one scale of tolerance is to compare without having selected the respects of comparison.

The theory also changes how information-seeking should be understood. Someone who books accommodation in advance is not thereby revealed as unable to bear an open future, since they may be engaged in effective predictability-making. Equally, someone who obtains medical information, prepares thoroughly for an examination or draws up contingency plans may be reducing future practical disruption. Questionnaire statements about wanting to know what the future holds, or wanting to organize everything in advance, can register such activity as a symptom. The scientific question concerns what these activities accomplish, rather than whether they indicate a generalized intolerance. Planning that makes many futures similar enough is among the most effective forms of coordination available to living beings, and a science that scores it as a deficit has mistaken an achievement for its absence.

A better research programme would proceed in a definite order. Researchers would begin by identifying the anticipated development and the relevant comparison, whether directional, categorical, interval-based, exact, or tied to the viability of a coordination. They would specify the temporal horizon and the symbolic resolution. They would establish the type of recursive counterpart involved. They would document what preparedness and infrastructural stabilization already exist. Only then would they observe whether subsequent coordination is maintained, disrupted or restored, and in which respects it resembles what was anticipated. Such a programme can incorporate quantitative prediction tests, longitudinal observation, ethnography and experimental manipulation. Its crucial requirement is to avoid treating distinct questions as though they measured one underlying property. A person who forecasts their partner’s replies poorly but repairs misunderstandings quickly, and a person who forecasts them well but cannot recover when surprised, would receive the same score on any single measure of coping with uncertainty, while living very different relationships with their futures.

Three tightly designed comparative studies would test the central claims. The first would compare predictions of relative outcomes with predictions of exact magnitudes within the same task, ball throwing being the obvious candidate, to establish how much apparent unpredictability is produced by symbolic resolution alone. The second would compare how strongly forecasts about an event influence that event across nonrecursive, selfrecursive and interrecursive situations, holding outcome, horizon and resolution as constant as possible. The third would compare the effects of preparedness and infrastructural support on coordinative outcomes when the exact course of events remains unknown, for instance by studying how travellers with different resources recover from the same disruption. These studies concern different mechanisms, and their results should not be collapsed into a single predictability score. A science of predictability must investigate the conditions under which anticipations become practically reliable. It cannot begin by assuming that unpredictability is a single measurable feature of the world, or that coping with it is a single measurable capacity of individuals.

Conclusion: The Future Need Not Repeat the Prediction

Return to the field and the three tennis balls. The second throw was highly predictable although its landing point was unknown. The third throw demanded exact equivalence and was therefore much harder to accomplish. The lesson concerned what successful anticipation requires, rather than ballistics. A future development does not need to reproduce an anticipated image or numerical value in every respect. It must satisfy the comparison that made the anticipation meaningful, whether that comparison concerns a directional difference, membership in a category, a numerical interval, or the continuing viability of a particular living coordination. Selecting such criteria is part of prediction itself, rather than a secondary problem arising after the event.

Living beings also do more than formulate predictions. They build skills, establish habits, enter commitments and create infrastructures that make particular kinds of future coordination increasingly reliable. This is where a theory of comparison meets the three modes of future intelligence. Predictability is neither simply discovered in stable processes nor necessarily destroyed by recursivity. Nonrecursive, selfrecursive and interrecursive counterparts differ in how anticipation can enter their unfolding, and those differences produce stability as well as instability. In selfrecursive and interrecursive processes, anticipation can even help make the anticipated outcome occur. The bank that fails because its failure was predicted and the marriage that lasts because both partners expect it to last are two faces of the same structure, which a theory of predictability confined to forecasting accuracy cannot see.

The account explains several things at once. It explains why increasing numerical exactness can make forecasting success harder without any change in the world forecast. It explains why recursivity can stabilize expectations as well as destabilize them. It explains why training and infrastructure often matter more for living coordination than forecasting does. And it explains why the absence of perfect knowledge so often fails to amount to a practical problem. The theory offers no prospect of eliminating uncertainty. It describes how living beings establish increasingly reliable relationships with futures that remain open in indefinitely many respects. An experienced traveller need not know exactly what will happen. A trained boxer need not know every punch in advance. A person in a dependable relationship need not know every future response. A functioning society need not predict every transaction in order to sustain reliable exchange. Such achievements are no deficient approximations to perfect predictive knowledge; they are among the principal ways in which life coordinates with its future.

The future does not have to happen exactly as predicted for a prediction to succeed. It has to happen similarly enough in the respects that matter. And the greatest achievement of living coordination is often to establish conditions in which ever more possible futures are similar enough.

Selected References

Ecks, Stefan. 2022. Living Worth: Value and Values in Global Pharmaceutical Markets. Durham, NC: Duke University Press.

Ecks, Stefan. 2026a. “Predicting Unpredictability: Three Modes of Future Intelligence in Rocky IV.” Living Value Theory, livingvaluetheory.org.

Ecks, Stefan. 2026b. “When Abstractions Correlate: Intolerance of Uncertainty and the Production of Psychological Objects.” Living Value Theory, livingvaluetheory.org.

Gneiting, Tilmann, Fadoua Balabdaoui, and Adrian E. Raftery. 2007. “Probabilistic Forecasts, Calibration and Sharpness.” Journal of the Royal Statistical Society: Series B (Statistical Methodology) 69 (2): 243–268.

Goodman, Nelson. 1972. “Seven Strictures on Similarity.” In Problems and Projects. Indianapolis: Bobbs-Merrill.

Goodman, Nelson. 1978. Ways of Worldmaking. Indianapolis: Hackett.

Keynes, John Maynard. 1936. The General Theory of Employment, Interest and Money. London: Macmillan.

Merton, Robert K. 1948. “The Self-Fulfilling Prophecy.” Antioch Review 8 (2): 193–210.

Saussure, Ferdinand de. 1959 [1916]. Course in General Linguistics. Translated by Wade Baskin. New York: Philosophical Library.

Solt, Stephanie. 2016. “On Measurement and Quantification: The Case of Most and More than Half.” Language 92 (1): 65–100.

Wacquant, Loïc. 2004. Body and Soul: Notebooks of an Apprentice Boxer. New York: Oxford University Press.