Abstract
A good night's sleep is easy to recognize retrospectively and unusually difficult to produce prospectively. People spend years experimenting with beds, rooms, temperatures, routines, food, exercise, substances, partners, work schedules, and expert recommendations, yet no arrangement reliably guarantees good sleep. This article argues that the difficulty is not merely that sleep is scientifically complicated, but that sleep is a living, recursively mediated process that cannot be converted into a predictable output of fixed inputs. Sleep reveals several general features of Living Value Theory with unusual clarity. It systematically reconfigures the intersections between all five mediations, all five levels of recursivity, and all three kinds of recursive counterpart, rather than merely touching each axis independently. It produces a striking asymmetry between the two poietic mediations, in which multisymbolization becomes radically attenuated while multimateriality becomes disproportionately important, a difference waking life ordinarily conceals. Because sleep cannot readily be produced by direct intervention, enormous effort is displaced into indirect intervention upon the surrounding mesocosm instead, and directness itself turns out to be something recursively discerned rather than simply given. Because outcome discernment is far easier than causal attribution, every sleeper conducts a lifelong, uncontrolled experiment on their own sleep, drawing conclusions that remain permanently underdetermined by the evidence available to draw them from. The article answers its own title in three stages: a good night's sleep is a period from which a person emerges sufficiently restored for subsequent living, produced while the organism remains predominantly at L1 with only selective L2 permeability; it can mostly be had only indirectly, by arranging conditions under which sleep becomes possible rather than by directly causing sleep to occur; and general answers are never enough because the same intervention operates within a constantly changing, multimediated living configuration in which successful outcome discernment is always easier than reliable causal attribution, so that population-level regularities, however real, cannot by themselves determine any particular night's outcome.
I. The Simplest Question That Turns Out Not to Be Simple: “Did You Sleep Well?”
Almost everyone understands the question did you sleep well. It requires no medical theory, no sleep tracker, no polysomnography. A person can ordinarily wake and say, with considerable confidence, that it was a good night, that they slept terribly, or that they slept for eight hours but somehow woke exhausted anyway. That last case is worth pausing on, because it already reveals something the rest of this article will spend a great deal of effort explaining. There is no simple mathematical correspondence between hours slept and how rested a person feels the next day. Duration is a real and useful variable, but it is not the thing itself.
This produces a striking asymmetry, and the whole article follows from taking it seriously. Retrospective discernment is comparatively easy. A person usually knows, on waking, whether the night went well. Prospective production is extremely difficult. The same person very often cannot say, with anything like the same confidence, what would guarantee a repetition of that outcome tomorrow night.
Most adults have, in effect, been running an uncontrolled experiment on their own sleep for decades without ever quite calling it that. Someone exercises one afternoon and sleeps unusually well that night. Someone else has a drink before bed and sleeps unexpectedly badly. Food, screens, bedtime, temperature, bedding, stress, and countless other variables get changed, singly or in combination, and the following morning supplies a verdict. Did that change help? Sometimes the answer seems clear. More often it is maddeningly ambiguous, since the same change that seemed to help last month seems to make no difference, or even to hurt, this month. Expert advice, meanwhile, rarely resolves the ambiguity as cleanly as it promises to. A recommendation that works reliably for a population, and that may even be true on average for the very individual trying it, can still fail on any given night for reasons nothing in the advice itself anticipated.
This licenses three questions worth stating plainly before anything else. Why can a person recognize good sleep so much more easily than they can produce it. Why can an intervention work tonight and apparently fail tomorrow, with nothing else in the person's life obviously having changed. Why does an enormous and still-growing body of scientific knowledge about sleep not translate into anything resembling an algorithm for sleeping well, of the kind available for a great many other bodily processes.
Two tempting answers should be set aside immediately, because both are true enough to sound satisfying while being too weak to explain what actually needs explaining. The first is that the problem is simply insufficient scientific knowledge, and that better instruments, larger studies, and more refined models will eventually close the gap. More knowledge can certainly improve probabilities, and nothing here disputes that. But improving probabilities is a different achievement from turning sleep into an engineered output, and there is no reason to expect the second to fall out of enough of the first. The second tempting answer is that the problem is merely individual variation, that everyone is different, and that a general theory was never going to fit every case. This is true and theoretically weak at the same time. It restates the puzzle without touching its structure. The deeper issue is not that individuals vary around some otherwise stable rule, but that a living outcome like sleep is generated fresh, every night, within a continuously changing, multimediated field of coordination that includes the sleeper's own body, the people and animals nearby, the room, the season, and a symbolic apparatus of expectations and self-monitoring that can itself change what it is trying only to describe.
This gives the article its central claim, stated here so that everything which follows can be read as an explanation of it rather than a series of separate observations. Good sleep is discernible without being fully engineerable. That is not a confession of ignorance waiting to be corrected by future research. It is closer to a structural feature of what sleep, considered as a living process rather than a technical output, actually is.
II. What Exactly Is a Good Night's Sleep? Sleep as Mesocosmic Attunement
Any adequate answer has to refuse a definition of good sleep built exclusively from duration, sleep stages, or medical norms, useful as all three can be for other purposes. Good sleep requires attunement across a great many relations at once, and no single measurable variable, however carefully tracked, captures that attunement on its own.
It helps to distinguish sleep from several states it can superficially resemble. Sleep is not sedation, is not simple loss of consciousness, and is not coma, and the distinctions matter even though this article has no need to adjudicate the more technical questions anaesthesiology and the study of disorders of consciousness would require. The point at stake is narrower and considerably safer to state: a state that reduces consciousness is not thereby equivalent to physiologically normal, restorative sleep, whatever else it may share with sleep from the outside. This matters practically as well as conceptually, because a purely pharmacological intervention can produce unconsciousness, or something resembling it, without necessarily producing the restorative condition anyone actually wanted. And good sleep is not simply long sleep, since duration without restoration is one of the most common and most frustrating experiences reported by people who otherwise sleep for what should, on paper, be more than enough time.
The more basic stake is not symbolic. Nobody needs sleep primarily because sleep means something, in the way a ritual or a qualification might mean something. A living organism needs sleep because it cannot indefinitely sustain waking expenditure without periodic, deep metabolic restoration, and this remains true whether or not the sleeper has any theory of why. Good sleep is accordingly one of the clearest cases in which value precedes explicit valuation. The organism needs restoration whether or not it conceptualizes that need, in exactly the way a plant needs water whether or not anything about the plant represents thirst. Literature recognized this long before physiology had the vocabulary to specify it precisely, from Homer's recurring treatment of sleep as the restorer of a hero's strength to Goethe's Faust waking transformed after a night's rest the text treats as something closer to renewal than mere unconsciousness. The recurrence of this image across such different literatures is itself a kind of evidence, however informal, for how basic and how widely recognized the underlying phenomenon is.
There really is such a thing as having had enough sleep, and this deserves stating plainly against a cultural background that increasingly treats rest as something to be minimized, optimized away, or endlessly deferred. Contemporary discourses of acceleration, in which more work, more output, and more availability are treated as though limitlessly achievable given sufficient willpower or sufficiently clever technology, can obscure the banal biological fact that organisms reach limits, and that satisfactory metabolic recovery is a real, achievable, bounded state rather than an asymptote nobody ever quite reaches. This should not be mistaken for a claim that enough sleep corresponds to a universally fixed number of hours. Enough is a recursively discerned state of restored fit between an organism and what it is about to be asked to do, and that state can shift with age, health, season, workload, and a dozen other factors without ceasing to be a real and recognizable condition on any given morning.
This licenses a preliminary working definition, offered here in full knowledge that later sections will need to complicate it further. A good night's sleep is a period of sufficiently uninterrupted recursive metabolic self-maintenance that restores workable mesocosmic fit without requiring sustained conscious correction. Two phrases in that definition are doing particular work and are worth flagging now. Sufficiently uninterrupted matters because good sleep does not require absolute environmental perfection, only enough continuity for the underlying processes to run to completion; a sleeper who stirs twice in the night without fully waking has not thereby failed to sleep well. Without sustained conscious correction matters because it anticipates the whole of the section that follows, in which the article turns from what good sleep is to what happens, mediation by mediation, level by level, and counterpart by counterpart, while it is actually occurring.
This theoretical definition needs a companion rather than a replacement, since one of this article's own central arguments is that people very often know they slept well long before they have any theory of why, and a definition built entirely from LVT's own vocabulary risks looking as though it were dictating the answer to ordinary discernment rather than being constrained by it. The practical test worth keeping alongside the theoretical one is simpler and more phenomenological: a good night's sleep is one from which the person emerges sufficiently restored for subsequent living. That criterion asks nothing of the sleeper beyond what they already, reliably know how to judge, and it is this criterion, not the more technical formulation above, that the rest of the article will keep returning to whenever the discussion needs to be checked against what an ordinary sleeper actually means by the phrase.
III. The Sleeping Configuration: Five Mediations, Five Levels, Three Counterparts
This section is the conceptual centre of the article, and its argument can be stated in advance. Sleep does not merely reduce consciousness. It produces a systematic reconfiguration of the entire architecture of mediation, recursivity, and counterpart relation that organizes living coordination in general, and it does so in a way that is unusually visible precisely because so much of waking life normally keeps these elements running together, obscuring how separable they actually are.
Begin with the three zoetic mediations, all of which remain fully active throughout sleep, though each in a different register from its waking one. Multisensorial embodiment does not shut down. It becomes differently organized. Breathing, thermoregulation, posture, internal sensation, pain thresholds, and the ordinary work of bodily repair all continue, and the sleeper remains exquisitely responsive to some stimuli, a change in temperature, a full bladder, an unfamiliar sound, while remaining almost entirely unavailable to others that would readily catch a waking person's attention. Sleep is therefore not the absence of embodiment but embodiment with vastly reduced conscious supervision, a distinction worth holding onto carefully, since collapsing it produces most of the confusions this article is trying to dissolve.
Being-with also remains active. Other beings can continue to matter without anything resembling sustained waking interaction. A partner's presence, a parent sleeping near a child, fellow sleepers in a shared room, a dog at the foot of the bed, or a designated watcher can all contribute to a sleeper's safety and sense of security without any of them engaging in continuous reciprocal exchange with the sleeping body. This is worth setting against a background assumption, common enough in contemporary discussions of sleep hygiene, that solitary sleep in a private room is simply the mature, correct, or default arrangement. Historically and comparatively, sleeping alone, in a room built for one person, is closer to an unusual and comparatively recent solution than to a universal baseline, and many animals, humans very much included across most of their history, have slept in groups as a matter of course rather than exception.
Multiversal dwelling, meanwhile, continues regardless of whether it is consciously represented. Day and night cycles, temperature, seasonality, light, and the planet's own rhythms keep operating on the sleeping body whether or not the sleeper has given any of them a moment's thought. Artificial lighting makes this unusually visible, because it is one of the few technologies capable of partially decoupling a living rhythm from the environmental cycle it evolved to track, letting a household extend its waking hours well past sunset while the underlying circadian machinery continues, at least partially, to expect darkness on something closer to the old schedule.
The two poietic mediations behave very differently from these three, and this divergence deserves to be stated as a named proposition, since it is arguably the article's single most original theoretical contribution. Sleep produces a poietic asymmetry, in which multimateriality and multisymbolization, mutually reinforcing poietic mediations throughout ordinary waking life, become radically unequal the moment recursive availability falls. The material world can continue doing things for a person while that person is unavailable. Symbols generally cannot. A mattress does not need the sleeper's cooperation to keep supporting them. A sentence does need a living interlocutor, awake and available, to mean anything at all. This is not merely an observation about the material culture of sleep, useful as that observation is in its own right. It follows from a basic ontological difference between the two poietic mediations that waking life ordinarily conceals, because waking humans constantly coordinate the two so tightly that their difference in kind rarely becomes visible. Sleep, by withdrawing one of them almost entirely while leaving the other undiminished, exposes a distinction that feeds back into Living Value Theory generally, well beyond the special case of sleep: multimateriality and multisymbolization are not two flavours of the same poietic achievement, but two mediations with fundamentally different dependence on the living recursive participation of the beings they serve.
Concretely, multisymbolization becomes radically attenuated. Ordinary sleep cannot sustain the waking forms of symbolic articulation: the sleeper cannot debate a point, plan tomorrow's schedule, interpret a document, answer a direct question, or maintain the kind of institutional reflexivity waking social life constantly demands. Multimateriality, meanwhile, becomes disproportionately important, precisely because material arrangements do not require the sleeper's own recursive participation in order to keep working. The mattress continues supporting the body without needing to be asked. The wall remains exactly where it was left. The blanket retains heat on its own. The lock stays locked without anyone checking it. The curtain keeps out light without being reminded to. The heating system, or the air conditioning unit humming in the next room, continues regulating temperature entirely independently of whether the sleeper is in any position to notice or adjust it. A bedroom, considered this way, is less a room than a small, deliberately engineered zone in which one poietic mediation has been asked to do nearly all of the work that both mediations share during waking hours.
It would be too strong to say that L3, L4, and L5 are simply unavailable during sleep, and the stronger claim invites an obvious objection: dreams contain words, questions, social encounters, and sometimes what look like elaborately structured symbolic scenes, sleep talking produces audible speech, and lucid dreaming and the drowsy transitions either side of sleep proper all involve something that looks, at least partially, like symbolic activity in progress. The more defensible and more interesting claim is narrower. Sleep radically attenuates ordinary, externally testable L3 and higher recursive operation, rather than eliminating symbolic activity outright. Four cases are worth distinguishing rather than collapsing together. Ordinary waking L3 and above involves articulation that can be questioned, corrected, and checked against an external world by other people and by the speaker's own later reflection. Dream-internal symbolic activity can be structurally elaborate, populated with words and encounters, while remaining unavailable to any of that checking from within the dream itself, unfalsifiable and uncorrectable for as long as the dream continues. Transitions and partial arousals, the drowsy minutes either side of sleep proper, supply a genuinely intermediate case, in which fragments of ordinary waking articulation become briefly available before submerging again. And fully developed waking recursive engagement, askable, correctable, socially checked, returns only on waking itself. What sleep removes, in other words, is not symbolic activity as such but askability, the capacity for a symbolic claim to be tested, questioned, and revised against something outside itself. Sleep is therefore, among many other things, a nightly and entirely normal collapse of tested, checkable recursive engagement, repeated with almost no exception across an entire human life, and one that involves no cessation of living coordination whatsoever, only its withdrawal from the levels at which that coordination becomes explicitly and reliably available to itself.
Good sleep approaches prolonged operation at L1, in which metabolic processes continue seamlessly, without requiring articulation, correction, or explicit attention of any kind. L2, felt misalignment, becomes something to be minimized rather than something to be sought, since noise, pain, heat, cold, unwanted movement, or anxious rumination can all produce exactly the kind of disturbance that threatens to escalate a sleeper back towards wakefulness. This needs one immediate refinement, since an unqualified version of the claim would be both false and dangerous: the complete absence of all L2 sensitivity is neither achievable nor desirable, because a sleeper who could not register genuine danger, a smell of smoke, an intruder, a child's distress, would have traded restorative sleep for a state closer to the vulnerability this article has already associated with sleep in general. The relevant goal is not the elimination of L2 responsiveness but its successful non-escalation. Put more precisely, good sleep requires selective L2 permeability rather than any uniform threshold applied indiscriminately across every possible disturbance. Irrelevant differences need to remain safely below the point of escalation: traffic noise a sleeper has grown used to, an ordinary creak in the building, a partner's habitual movement. Relevant dangers need to cross that same threshold reliably: smoke, an unfamiliar sound, a child's distress. A parent who sleeps through steady traffic outside yet wakes instantly to a small, specific change in an infant's breathing has not achieved some generically low level of L2 sensitivity. Their system is performing a remarkably fine-grained act of recursive discernment, continuously sorting the ordinary from the relevant while the parent's own waking attention is almost entirely offline. This is worth holding onto carefully, because it prevents the whole L1 argument from sounding like a case for global sensory shutdown, which it is not and was never meant to be. Good sleep is accordingly best described as prolonged L1 coordination with sufficiently little, and sufficiently well-targeted, L2 escalation, a formulation that leaves room for the ordinary small disturbances every real night of sleep actually contains.
The three kinds of recursive counterpart are redistributed by sleep in a final, related way. Nonrecursive counterparts become maximized in relative importance, precisely because they are predictable and make no active demands on a sleeper's diminished capacity to respond. Most contemporary sleep technology, considered honestly, amounts to an attempt to build the most dependable possible nonrecursive environment around a sleeping body: a mattress that will not shift, a room that will not admit unwanted light, a temperature that will not drift. Interrecursivity is correspondingly attenuated rather than eliminated. A sleeper may remain beside another person all night without sustaining anything like the metabolically expensive reciprocal engagement waking interaction ordinarily requires, so that active interrecursivity gives way to something closer to quiet, low-bandwidth being-with, felt rather than performed. This should not be read as implying that the ideal sleeping environment is simply the most nonrecursive one achievable, with other people present only as an unfortunate necessity to be minimized. The evidence already presented in this article points the other way. Sometimes another person nearby is precisely what makes sleep possible at all, felt security doing work no wall or lock can substitute for. The ideal is accordingly not maximum nonrecursivity but maximum predictability compatible with whatever forms of being-with actually support safety for a given sleeper. A partner can be considerably less predictable, night to night, than a wall, and still help someone sleep far better than a wall alone ever could. What a sleeper seeks from being-with during sleep is low-demand, reliable interrecursivity, not its complete absence. Selfrecursivity, meanwhile, continues zoetically throughout, since a sleeping organism's own living organization keeps regulating itself without needing the symbolic I that waking self-consciousness usually supplies. This last point deserves to be stated with some force, because it is one of the clearest demonstrations available anywhere in ordinary life that selfhood cannot be reduced to articulated self-consciousness. Something continues being a self, coherently and continuously, for the entire duration of a dreamless night in which no articulated self is available to claim the continuity at all.
None of this should be read as though the five mediations, the five levels, and the three counterparts were simply five separate stories, five more separate stories, and three more again, laid side by side. What sleep actually does is reconfigure the intersections between them, and the claim that sleep restructures a genuine five by five by three architecture, rather than merely touching each axis independently, only earns its weight once at least a few of those intersections are traced through explicitly.
Embodiment supplies the clearest case. At L1, embodiment appears as selfrecursive metabolic housekeeping: breathing, thermoregulation, postural adjustment, the ordinary maintenance a body performs on itself without anything resembling deliberate attention. At L2, embodiment appears instead as disturbance: pain, bladder pressure, overheating, the specific felt misalignments capable of pulling a sleeper back towards wakefulness. Embodiment also has an interrecursive face, since another body touching, moving, snoring, or simply breathing nearby becomes, for a sleeper, one more embodied fact to be accommodated, felt rather than reasoned about. It has a nonrecursive face too, in the mattress, the blanket, and the ambient temperature that embodiment is constantly, wordlessly answering to. And it has almost no L3 or higher face at all until waking supplies one, since a sleeping body does not narrate its own thermoregulation, however busy that thermoregulation actually is. One mediation, embodiment, therefore cuts across every level and every counterpart type sleep touches, behaving differently at each intersection rather than occupying a single fixed position marked simply active.
Being-with cuts across the same grid quite differently. Its L1 face is close to embodiment's, a partner's or a child's continued proximity registering as ambient safety without anything resembling explicit attention. Its L2 face is comparatively rare and highly selective, since most of what another sleeping body does goes unnoticed, while specific changes, a child's breathing sounding wrong, a partner's sudden stillness, cross the threshold into felt disturbance precisely because they are relevant rather than because they are loud. Being-with is, by definition, the interrecursive counterpart type in its purest form during sleep, yet it becomes, for the duration, one of the lowest-bandwidth versions of interrecursivity available anywhere in ordinary life, closer to shared proximity than to any waking exchange. It has essentially no nonrecursive face of its own, since a fellow sleeper is definitionally not the kind of thing that stays put the way a wall does. And its L3 face, like embodiment's, waits for morning almost entirely.
Multimateriality supplies a third cross-section, and its profile is the inverse of the first two in an important respect. It has no L1 or L2 face at all, in the sense that the mattress does not itself experience anything, sleeping or otherwise. What it has instead is an overwhelming and disproportionate nonrecursive presence, precisely the presence the rest of this article has been arguing becomes most consequential exactly when a sleeper's own recursive availability is most reduced. Its interrecursive face is essentially absent, since a wall does not respond to a sleeper's response to it. Its relation to L3 through L5 is likewise minimal during sleep itself, though it becomes, the following morning, one of the most heavily symbolized objects a household owns, discussed, purchased, compared, and reviewed in exactly the register it could not participate in overnight.
These three cross-sections do not exhaust the grid, and a fuller treatment could trace multiversal dwelling and multisymbolization through the same levels and counterparts with equally revealing, and equally different, results. But the pattern is already visible from three cases. Sleep does not merely activate or deactivate five mediations, compress five levels, and redistribute three counterparts as though running three separate dials. It produces a genuinely five by five by three transformation, in which each mediation occupies a distinctive, non-interchangeable position across the levels and counterparts, and in which no single mediation's profile can be inferred from any other's. The following schematic condenses the transformation as a whole, contrasting the waking configuration with the sleeping one across its major variables.
The whole configuration can, for present purposes, be condensed into a single working description. Good sleep combines extensive zoetic self-maintenance, sharply reduced symbolic recursion, reliable multimaterial scaffolding, low-demand but reliable interrecursivity, and the predominance of predictable, nonrecursive counterparts over unpredictable ones. Every later section of this article is, in one way or another, an exploration of what follows from taking this description seriously.
IV. Why Trying Directly to Sleep Is Such a Peculiar Intervention
Having established what good sleep is, the article now turns from ontology to intervention, and the turn begins with an observation so familiar that its strangeness is easy to miss.
Sleep cannot simply be commanded, and it is worth contrasting this with a great many other everyday acts to see exactly how unusual that makes it. A person can decide to sit down, walk across a room, drink a glass of water, or send a message, and in each case deciding is very close to sufficient. Nothing comparable holds for sleep. Nobody can straightforwardly decide, at twenty-two minutes past ten in the evening, that they shall now become properly asleep, and the attempt to do exactly that is one of the most reliable ways of remaining awake for a good deal longer than would otherwise have been necessary. This reveals a central and somewhat paradoxical feature of sleep as a target of intervention: the state being sought requires a reduction of precisely the kind of recursive, deliberate activity by which interventions are ordinarily carried out in the first place. To intervene in the usual sense is to bring L3 articulation and L4 planning to bear on a problem. Sleep is a problem that recedes exactly to the extent that L3 and L4 are brought to bear on it.
Direct symbolic intervention can accordingly be self-defeating in a way most other advice is not. Instructions such as relax, stop thinking, go to sleep now, or you need eight hours tonight can easily increase self-monitoring rather than reducing it, since each instruction gives the sleeper something new to check themselves against. The intervention produces exactly the L2 and L3 escalation that successful sleep requires to subside, turning a piece of advice meant to help into a further source of the very disturbance it was meant to resolve. Sleep is unusual, among the states people commonly want to achieve, in that symbolic attempts to control it can recursively amplify the misalignment those attempts were meant to correct, a mechanism with no close equivalent in, say, learning to walk across a room or deciding to have a drink of water.
Direct material intervention encounters its own limits, though of a different kind, and medication is the clearest case. Sleep medicines, and antidepressants used off-label for their sedative effects, can be genuinely efficacious for particular people in particular circumstances, and nothing in this article's argument requires denying that. What deserves emphasis instead is a more defensible and more interesting theoretical point. Even when a pharmacological intervention successfully produces sleep, or something closely resembling it, this does not thereby convert sleep into a simple technological output of a chemical input. The intervention itself alters the wider recursive configuration surrounding sleep, through tolerance that changes how the same dose behaves over time, through expectations that shape how the sleeper interprets a given night, through routines that build up around the medication's use, through effects the following day that were not part of the original bargain, and through anxiety about dependence that can itself become a new and troublesome source of nighttime rumination. A pill that reliably produces unconsciousness has not thereby produced the whole, richly mediated state this article has been calling good sleep, and the gap between the two is where most of the genuine difficulty in pharmacological sleep management actually lives.
Sleep science confronts a version of the same paradox at the level of method rather than treatment, and the sleep laboratory is an unusually revealing case. To observe sleep scientifically at all, researchers must construct an experimental environment, and that environment is itself an unusual and, for most participants, an unfamiliar sleeping environment: wired, monitored, often shared with strangers, and knowingly under observation. This raises a question worth taking seriously rather than dismissing as merely a methodological nuisance to be engineered away with better equipment. Can observation of sleep ever be entirely uncontaminated by the fact of being observed, when placing someone inside a laboratory already changes the mediational conditions under which their sleeping is occurring. The point of raising this is not to dismiss sleep science, which has produced a great deal of genuine and useful knowledge about population-level regularities, but to show that measurement itself belongs to the mediational configuration under study rather than standing safely outside it, which is exactly what should be expected of any attempt to symbolically capture a process whose success partly depends on the absence of exactly that kind of symbolic attention.
These three cases, the failure of direct command, the limits of direct pharmacology, and the self-involving structure of direct observation, are not three separate curiosities but three instances of a single underlying pattern worth naming explicitly before moving on. Sleep is a living process whose successful operation partly consists in the withdrawal of exactly the kind of explicit, symbolically mediated attention that most other interventions rely on to work at all. An intervention aimed at sitting down, drinking water, or sending a message succeeds by adding deliberate, symbolically guided action to a situation that otherwise lacks it. An intervention aimed at sleep succeeds, if it succeeds at all, only by finding some way to remove deliberate, symbolically guided action from a situation that otherwise contains too much of it. This is why the rest of this article turns, almost immediately, away from direct intervention and towards the vast, indirect apparatus through which humans have actually learned to make sleep more likely without ever managing to make it obey.
V. The Explosion of Indirect Intervention: From Pillows to Political Order
The inability to command sleep directly does not leave humans technologically powerless. Quite the opposite. The less controllable sleep itself becomes through direct means, the more human intervention has historically expanded into its surrounding conditions instead, until almost no scale of social and material organization, from a single pillow to an entire legal system, turns out to be entirely irrelevant to it.
Begin with the most immediate material scaffolding, close enough to the body to be almost invisible as intervention at all. Beds, mattresses, pillows, blankets, and nightclothes manage the body's direct physical support and thermal environment. Room temperature, ventilation, light, noise, and sometimes music manage the immediate sensory field around it. Doors, locks, and curtains manage who and what may cross into that field uninvited. None of these objects touches the sleeping process directly, in the sense that none of them causes sleep the way flipping a switch causes a light to come on. Each of them instead removes, or reduces the probability of, some specific disturbance that would otherwise have to be handled, if it arose, by the sleeper's own dwindling waking capacity.
This licenses a useful way of thinking about the successful bedroom, worth stating as its own principle. A successful bedroom is designed so that very little happens inside it. Its purpose is not stimulation, in the way a workspace or a playroom is organized to encourage activity, but the prevention of recursive demand. It might be called, without much exaggeration, an anti-event environment, in which walls, locks, insulation, and controlled temperature all work by the same basic method: preventing differences in the surrounding world from becoming relevant enough to a sleeping body to require a response from it.
A single idea unifies most of what this section is describing, and it is worth making explicit rather than leaving implicit across a dozen separate examples. Sleep is metabolic maintenance, not metabolic innovation, and the overwhelming majority of what gets called sleep technology is, in the same sense, environmental maintenance rather than environmental innovation. A mattress does not need to innovate on sleep. It needs to keep supporting a body reliably, night after night, in exactly the way it did the night before. A heating or cooling system does not improve on sleep. It maintains a narrow band of temperature within which sleep, whatever else it requires, becomes more likely. A building does not enhance sleep by keeping weather out. It simply maintains the boundary between inside and outside that makes sleep survivable at all. A lock does not add anything to the sleeper's night beyond maintaining a boundary that would otherwise have to be maintained by continued waking vigilance. Maintenance, not innovation, is the right frame for almost everything material this section surveys, and it is worth holding onto precisely because it explains why so much of the apparatus surrounding sleep looks, from outside, almost deliberately unremarkable. Unremarkable is what it is trying to be.
The same logic scales outward into architecture proper. Rooms, houses, hospitals, monasteries, barracks, ships, worker housing, prisons, and hotels are all, among their other functions, technologies for making sustained metabolic vulnerability survivable at a larger scale than a single piece of furniture can manage alone. A sleeping organism is exposed in ways a waking one is not, and architecture across an enormous range of otherwise very different institutions exists partly to make that exposure survivable long enough for the underlying metabolic process to complete itself. The history of who is housed where, in what kind of structure, and under what degree of protection is, from this angle, inseparable from a history of whose sleeping vulnerability a given society has judged worth protecting, and how well.
One of the most striking forms of indirect intervention deserves its own treatment, because it works through neither material scaffolding nor institutional architecture but through another living being's continued wakefulness. Sometimes an environment becomes safe for sleep not because anything about the room has changed, but because someone else has agreed, tacitly or explicitly, to remain awake in it. Parents watch over infants. Soldiers stand guard while comrades sleep. Nurses monitor patients through the night. A partner remains sufficiently alert to notice if something is wrong. Religious traditions across an enormous range of cultures imagine a divine watcher keeping guard over sleeping humans who could not, on their own, keep watch over themselves. Watching over another person's sleep may be one of the finest and most easily overlooked forms interrecursivity ever takes, precisely because its success is measured by how little the watcher needs to do, and how completely the sleeper can therefore afford to stop watching altogether. Sleep, put plainly, very often depends on the delegation of recursive vigilance from a body that has, for the time being, given it up to another that has agreed to carry it instead.
From here the scale can be extended further still, into law, work, and infrastructure. Regulations governing driver working hours, limits on doctors' shifts, workplace rules designed to prevent fatigue-related accidents, and safety standards for night-shift industries all indirectly govern the possibilities for sleep available to the people they cover, whether or not sleep is ever mentioned in the regulation's own stated purpose. Electricity grids, domestic heating standards, building codes, policing, and transport infrastructure can likewise function as remote sleep technologies, several steps removed from any bedroom yet still shaping, in aggregate, how easily a given population can arrange the conditions this article has already described as necessary.
Economic security belongs on this same list, and deserves to be stated plainly rather than left as an afterthought. Financial anxiety is an obvious example of a concern capable of remaining recursively active at night, keeping a body oscillating at L2 rather than settling into L1. Savings, insurance, secure employment, and stable housing are not sleep technologies in any narrow commercial sense; nobody buys insurance primarily in order to sleep better. Yet all of them can profoundly alter the conditions under which sleep does or does not occur, by removing, or by failing to remove, one recurrent category of felt misalignment a mind can generate for itself in the dark.
Before extending the argument to its largest scale, it is worth pausing to state a criterion for when this indirect, sleep-relevant framing is actually doing analytical work, since without one the argument risks collapsing into the unhelpful claim that everything, eventually, affects everything else. An arrangement counts as indirectly sleep-relevant only where altering it reliably changes a sleeper's need for vigilance, their exposure to disturbance, their temporal availability for sleep, or the conditions for their metabolic recovery. Investment, policing, labour law, and international stability can each legitimately enter the argument where a specific causal pathway of this kind can be named, a shift pattern that determines when someone is free to sleep, a policing failure that raises perceived risk at night, a housing policy that determines whether a room can be heated. They do not enter merely because everything, in some diffuse sense, touches everything else in a large enough society. The criterion is narrow on purpose, and it is what keeps the argument from becoming unfalsifiable at exactly the scale where it is most tempting to let it.
War supplies the most dramatic illustration available of the same underlying principle, because it exposes what happens when almost every layer of this indirect apparatus is stripped away at once. A sleeping body cannot continuously defend itself, and military organizations have struggled with this hard metabolic limit for as long as organized warfare has existed, managing it through rotating watches, deliberate training in fragmented sleep, rigid schedules, and, in the twentieth and twenty-first centuries, stimulant use on an industrial scale. From this extreme case a more general and more carefully bounded claim can be drawn, and it satisfies the criterion just stated precisely. A stable political order can indirectly support sleep across an entire population by reducing the ordinary demand for continuous vigilance that war, civil breakdown, or pervasive insecurity otherwise impose. This is not a claim that any particular political system guarantees better sleep, but an analytical claim about how far an effective intervention can sit from the apparent site of the problem it is addressing.
That distance deserves a name of its own, since it recurs at every scale this section has covered, from a curtain to a constitution. Intervention distance describes how far from the apparent site of a living outcome an effective intervention may need to occur before it can actually change that outcome. The apparent problem, stated by the person lying awake, is always some version of I cannot sleep. The effective intervention might turn out to be a curtain, a quieter neighbour, a changed work schedule, a different room, or a less precarious financial arrangement, none of which resembles a treatment for insomnia in any conventional sense. Directness and relevance, this section's whole argument suggests, are simply different properties, and confusing them is one of the most common ways sleep advice fails the people who most need it to work.
VI. The Longest-Running Experiment: How Do I Discover What Works for Me?
Turn now from the mesocosm surrounding sleep to the sleeper's own, lifelong attempt to understand it, since every night supplies another trial in an experiment nobody formally designed and nobody can ever fully control.
Sleep gives each person thousands of repeated opportunities to alter some condition and observe an outcome the following morning, which sounds, described that way, like the raw material of a genuinely rigorous experiment. It is nothing of the kind. The variables involved keep changing whether or not the experimenter wants them to: age, health, relationships, weather, season, workload, stress, food, exercise, substances, expectations, and the sleeper's own accumulating history of previous sleep all shift continuously, none of them held constant while some single factor is varied in isolation. The sleeping subject therefore lives through a genuine longitudinal experiment on their own metabolism without anything resembling a stable control condition against which to compare it, which is exactly why two people can follow identical advice and reach entirely different, equally sincere conclusions about whether it worked.
This calls for two concepts to be distinguished carefully, since collapsing them together is responsible for a great deal of confused thinking about sleep, both popular and expert. The first is recursive discernment, the ordinary capacity to notice whether some change has improved or worsened a person's fit with their own night. The basic morning judgement I slept better is already an act of recursive discernment, and it is, on the whole, a reasonably reliable one. People are not generally wrong about whether last night went well, whatever else they might be wrong about.
The second concept, and the harder one, is recursive attribution, the assignment of an experienced change to some particular cause within an enormously multimediated configuration. There is a considerable distance between the modest claim I slept well after doing this and the much stronger claim I slept well because I did this, and everyday sleep-talk moves between the two so easily that the distance is rarely even noticed, let alone respected. This generates one of the article's strongest and most practically important claims. Outcome discernment can be comparatively strong while causal attribution remains weak, so that a person may know, with real confidence, that they slept badly, while having essentially no reliable idea why. Stated at its most general, the underlying proposition reaches well beyond sleep, even though sleep demonstrates it with unusual clarity: living outcomes are often more readily discernible than their causes are attributable. Friendship, creativity, attraction, and a great many other living processes plausibly share this same asymmetry, though establishing that properly is a task for another occasion. Sleep already demonstrates the pattern cleanly enough on its own.
A further complication deserves attention, because it shows that this experiment does not merely fail to be controlled. It actively modifies its own future course as it proceeds. Previous outcomes alter future interventions in a way no properly controlled experiment would ever permit. Someone who sleeps well after exercising is likely to exercise again, treating one good night as confirming evidence rather than as the single, ambiguous data point it actually is. Someone who sleeps badly after drinking alcohol is likely to adjust their drinking, on the strength of an inference at least as fragile as the one that led the first person back to the gym. An intervention that fails repeatedly tends to be abandoned, whether or not it was ever actually responsible for the failures attributed to it. Expert advice that seems wrong for a particular person is trusted less by that person from then on, sometimes correctly and sometimes not. The sleeping subject is accordingly never observing a stable process from a position safely outside it. The recursive history of the experiment becomes part of the material the next night's experiment has to work with, which is one more reason no two people's sleep, however similar their circumstances, ever runs on quite the same track for very long.
The exercise and alcohol cases already introduced are the plainer illustration, and it is worth returning to them here rather than reaching for a more elaborate invented example. Someone exercises one afternoon and sleeps unusually well that night. The natural inference is that the exercise caused the good sleep, and the natural response is to exercise again, expecting the same result. Sometimes the same result follows. Sometimes it does not, for reasons that may include the exercise's timing, the day's accumulated stress, an unrelated change in temperature, or nothing anyone could have named in advance. The same pattern holds for alcohol, tried in the other direction: a drink before bed, followed by unexpectedly poor sleep, generates a causal story that the next attempt may or may not confirm. Neither case is a story about one person's idiosyncrasy. Both are the ordinary shape recursive discernment takes whenever it is applied, patiently and imperfectly, to a genuinely multimediated process, and neither needs an invented protagonist with a suspiciously well-specified set of sensitivities to make the point.
One of the more original claims this article has to offer follows directly from this. People generally begin by assuming, often without noticing they are assuming it, that a given intervention operates directly on sleep: a pill leads to sleep, an early bedtime leads to sleep, a relaxation technique leads to sleep, each relation pictured as a fairly short, fairly reliable causal chain. Experience, accumulated slowly and usually reluctantly, tends to reveal that the actual relation is far more conditional than this, dependent on stress levels, on recent history, on what else happened that day, and on factors the person may never fully identify. Intervention distance, introduced in the previous section as a feature of the mesocosm surrounding sleep, turns out to be something else as well: it is itself an object of recursive discernment. A person spends much of their adult life learning, imperfectly and by increments, roughly where their own sleeping process can actually be influenced and roughly where it cannot, arriving, if they are fortunate, at a rough personal map of intervention distances that no general theory handed to them in advance could have supplied with anything like the same precision, and that no general theory could fully supply to anyone else either.
This points to a deeper refinement of the whole direct and indirect distinction introduced above. Directness and indirectness are not simply objective facts about where an intervention sits, waiting to be correctly identified once and for all. They are themselves recursively attributed, and that attribution can and does change as experience accumulates. A pill looks direct. A fixed bedtime looks direct. A relaxation exercise looks direct. Each presents itself, at first encounter, as a short, reliable causal chain running straight from the intervention to the outcome. Sustained experimentation over months or years routinely reveals a longer and far more conditional chain than the one originally assumed, dependent on factors never named in the original advice at all. Recursive discernment, on this account, does not merely sort interventions into ones that work and ones that do not. It continually revises the sleeper's own working model of where, in the wider mesocosm, an intervention actually sits, sometimes discovering that what felt direct was indirect all along, and occasionally the reverse.
VII. Why Expert Advice Cannot Become a Universal Recipe
This section has to proceed carefully, because its argument could easily be mistaken for an attack on sleep science, which it is not. Population regularities are real, and nothing in the account developed so far requires denying them. Expert advice about light exposure, caffeine timing, routine, substances, exercise, diagnosable sleep disorders, and environmental conditions can identify genuine, well-supported regularities that hold across large numbers of people, and following that advice measurably improves outcomes on average. None of what follows is a case for ignoring it.
The trouble begins when a probabilistic finding is silently converted into an unconditional prediction. A regularity of the form people who reduce caffeine after early afternoon tend, on average, to sleep better becomes, in the retelling, if you do this tonight, you will sleep well, and that second claim exceeds what the underlying evidence can actually warrant for any particular person on any particular night. A population regularity describes a tendency across many cases. An individual prediction claims to know a single, as yet unrealized outcome in advance. Nothing about the first licenses the second, however naturally the two get run together in ordinary conversation.
A further distinction sharpens this considerably, and deserves to be treated as one of the article's key theoretical results rather than a passing caveat. A thermostat can predictably make a room cooler. A blackout blind can predictably reduce incoming light. A lock can predictably prevent casual entry. Each of these is a highly reliable, nonrecursive material intervention, doing exactly what it was built to do on every single occasion it is used. None of them can predict, with anything close to the same reliability, that the person inside the room will actually sleep well as a result. This gives the article one of its cleanest formulations. Highly predictable nonrecursive conditions can scaffold an irreducibly recursive living outcome without thereby making that outcome itself predictable. This is precisely why good sleep cannot be converted into something wholly nonrecursive, however good the surrounding technology becomes at managing everything nonrecursive around it.
Sleep trackers are worth treating as a philosophically interesting case in their own right rather than as a straightforward villain in this story, and the interesting structure is not simply a subjective verdict facing off against an objective score. It is two recursively produced attributions about the same night coming into conflict. The embodied morning state, the sleeper's own felt sense of having rested well or badly, is itself a discernment, arrived at the way every such discernment in this article has been arrived at, through the sleeper's own accumulated, imperfect sensitivity to their own condition. An algorithmically stabilized symbolic representation, the tracker's nightly score, is a different kind of discernment entirely, produced by a fixed measurement procedure applied consistently regardless of context. When the two disagree, and they disagree often enough to matter, the sleeper is left having to recursively discern which of the two attributions should carry more weight, and for what purpose, whether to trust how they feel when deciding whether to push through a demanding day, or to trust the score when deciding whether something in their routine needs to change. Neither attribution is simply correct by default, which is what makes the disagreement a genuine, and genuinely interesting, small-scale instance of Living Value Theory's own concerns, played out on a single wrist every morning.
This is also why sleep hygiene advice, for all its genuine underlying merit, can tip over into a recursive overload of its own, and this failure mode deserves naming clearly rather than treating as an unfortunate but rare exception. Advice becomes counterproductive exactly when the sleeper is required to continuously monitor their own compliance with it: whether caffeine was stopped early enough, whether the room was exactly the right temperature, whether the phone was put away in time, whether the target number of hours was achieved on schedule. An apparatus originally intended to lower recursive disturbance can, applied too rigidly or tracked too anxiously, become a fresh and considerable source of exactly the disturbance it was meant to prevent. Sleep is accordingly a particularly clear case of a more general pattern, in which optimization, pursued past a certain point, becomes an obstacle to the very state it was meant to optimize.
None of this amounts to a case against expertise, measurement, or good advice, and it would be a serious misreading of the argument to take it that way, since the actual case being made is structural rather than adversarial. Population-level sleep science and a sleeper's own first-person recursive discernment are simply answering different questions. Science asks what tends to happen across a specified population under specified conditions. The sleeper asks a much narrower and much more local question: what happened to me last night, and what, if anything, should I change tonight. The translation from the first question to the second is inherently imperfect, not because either question is being asked badly, but because no amount of refinement to the first question ever fully answers the second. That is enough to explain the friction this section has been describing, without needing to cast either science or the sleeper as at fault.
VIII. From “What Works?” to “What Works Here, Now, for This Living Configuration?”
It is time to return directly to the article's title and answer each of its three parts in turn, drawing together everything the preceding sections have established.
What is a good night's sleep? Not an abstract numerical quantity, not merely a certain duration, and not simply unconsciousness, for reasons the second section already gave in detail. A good night is a period from which a person emerges sufficiently restored for subsequent living, produced through sufficient mesocosmic attunement and minimal disruptive recursive escalation, in which embodiment, being-with, and dwelling continue their ordinary work at reduced symbolic supervision while multimateriality quietly does most of the heavy lifting multisymbolization cannot perform until morning.
How can it be had? There is no single direct route, and the article has by now given a full explanation of why not. What remains is a set of general strategies, all of them indirect, all of them improving probabilities rather than guaranteeing outcomes. Recurrent sources of L2 misalignment can be reduced. Dependable nonrecursive material conditions can be created and maintained. Being-with can be organized in ways that provide safety without imposing excessive interrecursive demand on a diminished capacity to respond. Embodied and multiversal rhythms can be respected rather than fought, working with circadian timing and seasonal variation instead of against them. Symbolic knowledge, including expert advice, can be used as guidance rather than command, informing decisions without being expected to determine outcomes on its own. And a person can learn, recursively and over time, from their own accumulating experience, refining a personal map of intervention distances that no general theory could have handed them ready-made. None of this technologically manufactures sleep. All of it works towards making sleep possible, which is a different and more modest achievement, though not a lesser one.
It is worth being concrete about what this looks like in an ordinary life, rather than leaving the strategy at the level of abstraction the preceding paragraph necessarily traded in. Two people with near-identical bedrooms, near-identical bedtimes, and near-identical exposure to the same expert advice can arrive at reliably different nights, because one of them shares a bed with a restless sleeper and the other does not, because one carries a level of financial anxiety the other has never had to manage, because one has spent a decade learning that a warm room defeats them while the other has spent the same decade learning the opposite about themselves. None of this is a failure of the advice they both received. It is what the advice was always going to run into, the moment it met an actual, particular, historically accumulated living configuration rather than the idealized population from which it was originally drawn.
Why are general answers never enough? Because the question joins two very different kinds of thing that do not combine into a single, well-behaved object. On one side sit generalizable conditions, the regularities population-level sleep science can genuinely establish and that this article has never disputed. On the other sits a singular living outcome, produced fresh each night inside one particular body, one particular set of relationships, one particular room, one particular history, and one particular mesocosm, none of them fully substitutable for any other instance a study might have sampled instead. General statements can describe recurrent tendencies with real accuracy. They cannot fully determine how those tendencies will combine tonight, in this configuration, because the combination itself is a recursive, living event rather than the deterministic output of a specifiable set of inputs. The claim is not that general answers do not exist. It is that even a true general answer cannot, by itself, settle what a particular night will bring.
The correct response to this is not relativism, and it is worth being explicit about the alternative, since the two are easily confused from a distance. The wrong conclusion is anything works, or nothing can usefully be said. The right conclusion is more demanding and considerably more useful in practice: there are real constraints, real regularities, and genuinely better or worse interventions available to almost anyone, but their actual effects have to be recursively discerned within a living configuration whose outcomes remain incompletely predictable in advance, however well understood its general tendencies become. That distinction, between denying that anything reliable can be said and denying that everything reliable can be guaranteed, is doing all of the article's remaining work, and it is worth holding onto carefully against the pull, from both popular and some scientific writing about sleep, towards one or the other of its two easier, less accurate neighbours.
Conclusion: The Strange Expertise of Sleeping
Return, to close, to the opening morning question. How did you sleep. The person answering it possesses a strange and easily underestimated kind of expertise. They may not know the neurobiology involved. They may not know their own sleep stages, or care to. They may misunderstand what actually caused last night's outcome, confidently naming a cause recursive attribution cannot really support. Yet they can very often discern something deeply consequential about the state of their own living organization, accurately and immediately, in a way no external measurement has yet managed to fully replace or override.
Sleep therefore demonstrates the limits of any model in which knowledge consists primarily in explicit symbolic representation, whether that representation takes the form of expert theory, a tracker's nightly score, or the sleeper's own confident, and frequently wrong, causal story. Humans spend their whole lives learning, usually badly and always incompletely, how their own sleeping actually works. They rearrange rooms. They try substances, and reject them. They alter habits, change beds, change partners, change schedules. They follow expert advice, and ignore it. They make causal attributions, and quietly revise them a few months later when the pattern breaks down. And then, almost without exception, they try again the following night, because trying again is the only option a living, recursive process of this kind has ever actually offered anyone.
The broader theoretical result is worth stating in its most general form, since it reaches considerably beyond sleep alone even though sleep demonstrates it with unusual clarity. Living well often depends neither on direct technological control nor on possessing the correct symbolic explanation. It depends on recursively discerning how a changing organism fits within a changing mesocosm, and on altering the conditions of that fit as far as they can be altered, without ever being able to convert the living outcome itself into a fully predictable mechanism. Sleep is simply one of the clearest and most nightly available demonstrations of a pattern that runs through a great deal of what makes a life go well or badly.
The answer to how can I have a good night's sleep is, in the end, necessarily a little frustrating, and it is worth admitting this plainly rather than dressing it up as more conclusive than it is. There are many things worth doing. There are many things worth avoiding. There are genuine scientific regularities, established by careful research and worth taking seriously. There are powerful material technologies, from a well-chosen mattress to a well-designed building, capable of doing real and dependable work. But nobody, including the sleeper themselves, can ever issue sleep a final command that it is obliged to obey.
This is not a disappointing place to end, whatever its lack of a tidy final formula might initially suggest. Sleep asks for patient, recursive, never-quite-finished attention, compressed into a single ordinary night and repeated, with whatever accumulated wisdom the sleeper has managed to discern so far, for as long as they live. We can build the conditions for sleep. We cannot make sleep obey.
| Waking configuration | Sleeping configuration | |
|---|---|---|
| Multisymbolization | High, load-bearing | Sharply attenuated |
| Multimateriality | Present, shared with symbols | Disproportionately dominant |
| Recursive levels | L1 through L5 all available | L1 dominant; L2 selectively permeable; L3 and above largely unavailable |
| Interrecursivity | High-bandwidth, reciprocal | Low-bandwidth, felt rather than performed |
| Selfrecursivity | Available to explicit reflection | Continues zoetically, without the symbolic I |