Note on the translation — This translation re-fixes the source text without displacing its substance. A few terms of the corpus have no obvious equivalent in English; the choices made, and how they are to be read, are set out in the section “Translation choices” at the end of the volume. This translation is not frozen: it will evolve with readers’ feedback and with the improvement of the means of translation. The French version is authoritative.

Threshold

For a Processual Approach to the Reception of the Datum

Tome 3 — The Trace or the Datum

What reaches a gaze without entering it is worth no more than what the flow would have carried away.

The tome of movement ended on a question, bequeathed without being crossed: what becomes of what arrives without being read? Circulation has done its share — it has carried the datum to the edge of a gaze, and it carries more of it than any gaze can follow: this over-throughput is the weld between the two tomes, not their border. But what is decided at this edge no longer belongs to circulation. This tome has as its object what circulation cannot accomplish in anyone’s place: the reader’s share — the act of reading, the one who accomplishes it, their situation, their limit, their ethics.

From this resumption, the tome derives its guiding thesis, twin of those that preceded it: without reading, nothing is actualized. The title states it as an alternative — the trace or the datum — for this is the divide that the torrent replays before every window: depending on whether a reader reads what reaches them or not, the matter received is actualized or neglected, the trace naming no more, at the end of this slope, than an extreme horizon. And from this thesis the rest follows: there is an act for reading, as one was needed for drawing and one for transmitting; this act is the deed of a situated reader, whose window is finite; what this reader does with what overwhelms them is not written in advance — they can frame, delegate, or accept with lucidity; and not all unread data are a failure, for some were worth something only through a freshness that did not wait.

The eight documents of this tome lead this idea from the divide taken up again at the edge of the reader to the final posture of the one who knows they are overwhelmed and answers for it. At the end of this itinerary, the trilogy ends: not on an instruction, but on a lucidity, and the responsibility that comes with it. For the torrent, as the tome of movement established, is our resultant and not bad weather; what this engages for whoever receives it — for their probity, for their share in a wisdom that can only be collective — is what this last tome has the charge of saying.

This tome therefore starts from the edge of the reader, where the previous one stopped. It says what will long remain true to say of the reading of the datum; it leaves to others, and to the reader, the task of determining what must be done with it for today.

The Trace or the Datum

The Divide, Taken Up Again at the Edge of the Reader

Tome 3 — Document 1

“What is written without effort is in general read without pleasure.”

— attributed to Samuel Johnson

“A mark that refers to nothing for anyone is no longer a mark: it is grain in the stone.”

— Working formula

Warning

This document occupies, in the trilogy, a place symmetrical to the one occupied by the last document of the previous tome. There was the threshold: the point where circulation gave way to reading, and which the tome dealing with movement held rigorously without crossing it. Here is the resumption: the point where reading begins. The previous tome had stopped by holding out its hand across the threshold; this document takes it. It therefore starts nothing over. It inherits an established result — over-throughput relative to the window — and it turns it around: what movement described from the side of production, reception grasps again from the side of whoever undergoes it. The reader who approaches this tome without having read the previous one can do so; it is enough for them to know that circulation, having reached its racing, sets down before every gaze more data than any finite gaze can follow. It is from there, and from there only, that this tome starts.

Preamble — Taking Up the Threshold Again, and Turning It Around

The previous tome ended on a fact, which it set out as its last word: over-throughput relative to the window. Accelerated circulation produces more data in movement than any finite window can absorb. This was not an appraisal, but a fact of circulation, measurable in principle: there is, at every instant, a throughput of data that exceeds any capacity of reception. The tome had established the production of this throughput, and had stopped there, because what happens to what passes before a window without entering it is no longer a matter of circulation.

This document takes up this fact exactly, and performs on it a single operation, which commands all the rest of the tome: a turning around of perspective. The previous tome looked at over-throughput from the side of whoever produces it — circulation produces it. The present tome looks at it from the side of whoever receives it — a finite reader undergoes it. These are the two faces of a single concept, and this is what makes over-throughput a weld between the two tomes rather than a mere border. The previous tome established the production face; it set it out as its last concept. The present tome starts from the reception face; it makes it its first problem.

The first sentence of this tome could therefore be stated thus. The previous tome showed that circulation produces more data than any window can read. It remains to ask what becomes of what is not read — and what the reader who knows they are overwhelmed can do. The first question concerns the matter: trace or datum? The second concerns the reader: what do they do with what exceeds them? The present document deals with the first; the following ones will unfold the second.

Part I — Over-Throughput Has Two Faces

One Single Thing, Undergone Instead of Produced

The hinge concept, the one that belongs simultaneously to the two tomes, must be formulated with care, for it is what constitutes the weld.

The flow of the real, as the first tome set out, has no absolute throughput: it takes on a throughput relative to the fragment observed. Likewise, the torrent does not exceed in the absolute: it exceeds relative to a window. The reader perceives only a band of the spectrum of times, bounded by a threshold of the instantaneous and a threshold of the eternal; they are a situated reader, and their window is finite. The datum that circulates flows in relation to this finite window. When production increases without the window widening in proportion, it is not the window that grows: it is the throughput that rises. And this over-throughput, relative to the window, blurs what one could perceive of the real through it.

This is the face that the previous tome established: over-throughput produced by the racing of circulation, an objective fact, measurable in principle. The previous tome distinguished two sources of it, which can be recalled without taking them up again, for the present tome inherits them: speed, which raises throughput, and the coexistence of regimes, which widens it by keeping side by side apparatuses that readers assimilate at unequal rhythms. This second source deserves to be retained, for it already points towards reception: if regimes are superimposed, it is because the windows of readers differ, and because no leap is assimilated everywhere at the same rhythm. The throughput that spills over is therefore not only fast; it is also thickened by the diversity of the windows it must serve — which the present tome will make its object.

The present tome adds the other face, which is the same thing seen from the other edge: over-throughput undergone by a situated reader whose window is finite. It is no longer a fact of circulation; it is a fact of reading: the datum passes before the reader, they see it, they do not read it. The first face was the last concept of movement; the second is the first problem of reception. Between the two, no break: a single concept, looked at in turn from the bank that produces and from the bank that receives.

The Edge That the Previous Tome Did Not Cross

The previous tome had formulated it itself, while holding out its hand: at the edge of the reader, what circulation sets down threatens to be never more than an inscription that scrolls past — the image of the trace, raised as a threat, never ascertained as a fact. It stopped at this edge; the present tome deals with nothing else.

It is here that the whole tome is decided, on condition of setting out exactly what is decided here. What the torrent sets down before the reader is not an undecided matter that would await its being from the act: it is constituted datum — drawn, inscribed, dated upstream by the acts that the first two tomes established, and that no absence of a gaze undoes. What the act of reading decides is not what this datum is; it is what it becomes for someone: actualized — a meaning actually produced in a situation — or neglected — still constituted, but without a reader to actualize it. This is the divide that is replayed at every window, at every instant, on every fragment of the throughput: not between being and not being, but between reading and leaving.

The title of the tome — the trace or the datum — states the horizon of this divide, not its mechanism. What threatens the neglected datum is not ceasing to exist at the instant the gaze passes: it is losing, step by step, what reading alone maintains — the actuality of a meaning, the value of a use — up to the extreme point, which will have to be isolated, where no reader any longer, anywhere, could take it up again: there, and there only, the datum meets the fate of the trace. Neglect is the slope; falling back is its horizon. The tome has as its object this divide, and the act that settles it.

Part II — The Scale of the Datum

No Second Meaning

Let us first set aside a complication that would threaten the coherence of the whole trilogy. One might believe that the previous tome and this one handle two notions of the trace: one, a mark that circulates without being read; the other, the trace of the real that the first tome defined. There is only one notion of trace in the whole trilogy: that of the first tome — the mark that the flow of the real leaves on matter, caused, physically linked to what produced it, legible but not read (The Trace and the Reader). The datum, for its part, is born of a regulated and imputable act, which aims, cuts out, inscribes. Between the trace and the datum, a threshold that nothing crosses on its own: an act is needed. Which one — this is what must now be established exactly, for the whole tome depends on it.

Two Thresholds, Two Acts

The first tome showed the outward journey: the real deposits traces, and a situated reader, by surveying one of them under criteria, constitutes it as a datum. But let us look at this outward journey closely: what accomplishes it is not reading alone; it is an act of inscription — a drawing that aims, cuts out, fixes, dates, and records its provenance. The trace crosses the threshold when an act constitutes it; and what the act has constituted, nothing deconstitutes: an inscription that has taken place is not struck out, the rule under which the result was produced remains, the provenance remains. Reading, as the next document will show, bears only on decipherable inscription — never on the raw trace; and reading, as the rest of this document will recall, does not reactivate the datum and does not produce a new one. What can neither constitute nor produce can no more deconstitute: what the datum does not owe to reading, it cannot lose through its absence.

Two thresholds, then, and two acts. The entry threshold — from the trace to the datum — is crossed by inscription: it belongs to the first tome. The threshold that the present tome examines comes after it: from the constituted datum to the actualized datum — and that one, alone, is crossed by reading. The status of constituted datum does not depend on rereading; actualization depends on it — it is actualization, and it alone, that reading adds. A datum that no one reads therefore does not go back down to the state from which it had come out: it remains what the act made it, and it lacks a reader — constituted, not actualized. What this lack makes it lose, and in what order, a word does not say; a scale is needed for it.

The Five States

Here is this scale; it is the instrument of the tome, and it must be set out with the care of a foundation. Five states make it up — cumulative: each presupposes the previous ones — and relative twice over: to a reader, to an instant.

Constituted. A result has been produced and stabilized according to a rule, dated, provided with its provenance: what the act of inscription brings into existence. Its existence owes nothing to reading — it is the base, and it no longer moves: the four following states are added to it without ever remaking it.

Accessible. Constituted, and reachable for this reader here: an available occurrence, a practicable path, an open access. Relativity begins here: the same datum is accessible to one, out of reach of another — the reader’s share, seen from the datum.

Intelligible. Accessible, and legible for a community: the code, the reference corpus, the competences are alive in it. Relative again — unintelligible for this reader here, the datum is their mute datum; unintelligible for all, at the limit-ideal, the dead datum: the next part distinguishes the two with care.

Actualized. Read: a situated reader has actually produced a meaning from it in a situation. Actualization is an act, not an update; it is actualization — and it alone — that reading adds to the datum.

Operative. Actualized, and passed into the world: its reading has contributed to an inference, a coordination or an action. The modified real will give back new traces — it is there that the loop, at the close of the tome, will close again.

A datum can be constituted without being accessible, accessible without being intelligible, intelligible without being read, read without feeding anything. The scale gives neglect its measure: what a datum loses for want of a reader is read going down — operativeness, then actuality, then, if the conditions of resumption crumble, intelligibility and access — without constitution ever being erased. The torrent finds its exact description in it: it does not undo the data it carries along; it starves their actualization — setting down before every window more of the constituted, and often more of the accessible, than reading can actualize. And the alternative of the title finds its place: the trace or the datum does not name two states between which each fragment would tip at every instant, but the two ends of the scale — at the bottom, the limit case where universally lost intelligibility leaves only the inscription standing, and the trace regains its rights; everywhere else, datum, higher or lower according to the reader and the instant.

Part III — Two Stops on the Scale: the Mute Datum and the Dead Datum

The Mute Datum: a Relative and Reversible State

Now comes a fine distinction to hold, on pain of confusing a reversible stop with a death. The scale has just shown it: most of what no one reads is stopped by nothing — constituted, accessible, intelligible, the datum awaits only a window; it is simply not actualized. But the stop at intelligibility covers two situations of different natures.

The first is the mute datum. It still possesses an interpretation — a code that deciphers it exists —, but this code is unknown to this reader here. It is not a death: the interpretation exists, elsewhere, in another reader. The user of an application who looks at the source code of this application knows that it is a datum, but does not know how to read it; it is mute for them, alive for the developer. The state is relative and reversible: it is enough for a competent reader to come forward for the mute datum to become speaking again. The hidden datum — the one that its producer itself cannot reread, like the involuntary annotator who has no access to what they have produced — is in the same situation: they do not read it, but it is not dead for all that, for a reader, somewhere, could.

The mute datum has therefore lost nothing of what the act gave it. Constituted, accessible, it stops at the rung of intelligibility — and for this reader here only: it has its code, its interpretation, its community of reading; the reader who faces it is not part of it. Its muteness is a matter of relation between a datum and a reader, not a property of the datum.

The Dead Datum: the Second Death, Caused by the Flowing of the Real

The second situation is the dead datum. It is the limit case — the limit-ideal of the tome — where no reader any longer, anywhere, can interpret it. The interpretation is no longer hidden in another: it is universally lost. Only the inscription on the medium remains.

A datum whose interpretation has been universally lost dies the second death — disappearance — not through failure of the medium, but through loss of the interpretive link; its cause is the flowing of the time of the real, which has carried away the community of reading. Only the inscription remains: the datum has disappeared, the trace remains.

This attachment is faithful to The Two Times of the Datum, which distinguished the two deaths. The first death is lapse: the datum remains legible and intact, but it no longer tells the present real; it affects representative validity, not existence. The second death is disappearance: the datum purely and simply ceases to exist; it affects existence, not validity. And that document already placed on the side of the second death the case where “the format becomes illegible for want of a machine to read it”. The universal loss of interpretation is exactly of this type. Its mechanism belongs to the time of the real — the current has carried away the key to reading, the community that knew how to decipher has died out; but its result belongs to disappearance — no datum remains any more, only a mark on a medium. It is therefore not a third death that would have to be invented: it is the second death, disappearance, caused by the flowing of the real.

The dead datum so defined joins the first tome: “carrying nothing but its inscription” is having become a trace again. It is here, and here only in the whole trilogy, that the datum falls back: at the last rung, when intelligibility is universally unrecoverable, the scale touches the ground from which the inscription had come out. The dead datum has left the field of reading, for reading, as will be seen, bears only on decipherable inscription; being decipherable by no one, the dead datum is no more than a material mark — a raw trace of the real, which one day a new act of inscription may perhaps draw under another description (“regular marks on silicon, dated from such a period”), but which, as a datum, has disappeared.

The Datum That Passes into Lapse

Lapse is not disappearance: the two deaths have just been distinguished. What follows concerns only the first — and the first does not go down the scale. A lapsed datum remains constituted, accessible, intelligible: legible, and therefore reopenable. What it has lost is of another order — its validity for the current use, the act it served; the first tome gives this loss its exact measure, which is a relation and not a law.

The first death is not an instant but a gradual threshold. As long as the reading window of a datum remains open on the present, it serves the current act. When the gap from the flow of the real exceeds what its use tolerates, it ceases to light up the present — without ceasing to be legible. This threshold is its use-by limit: intrinsic to the datum when its object fixes it, held by default according to its age when nothing specifies it.

One can then distinguish, not three places, but three relations to the flow. The fresh datum serves the act. The consolidated datum no longer serves the act but the assessment: its summable share has been retained from it — extracted into a consolidated datum in the sense of the first tome: average, total, exact and without a leap — and its detail remains rereadable for a time. The lapsed datum serves nothing but memory: its reading window has closed on the present; its inscription subsists. Lapsed, and not merely stale: stale says the gap from the real, which grows continuously and is not decreed; lapsed says the state that the work gives to this gap when it exceeds what use tolerates. And these three relations do not replace the three ages of the first tome — fresh, ripening, archival —: the ages say the ageing undergone; the three relations say the regulated states that a work gives to this ageing, the use it reserves for the datum as it ages — the act, the assessment, memory. Moving a datum from one relation to another erases no truth: it is written in its new state before being withdrawn from the old one. Lapse orders; it does not destroy — and above all, it does not give the second death: a lapsed datum remains legible, and therefore reopenable.

Example (dated form). A rejected quotation has its use closed at once: its use-by limit falls to zero, and it passes into lapse immediately. A quotation with no follow-up passes into lapse at the end of the default for its age. The artisan, who knows the freshness of their data, regulates these thresholds; they do not create lapse; they fix its measure.

The Transfiguration of the Lapsed Datum into Recounting

It remains to say what a reader does with a lapsed datum. It is not a trace in waiting: constituted, legible, reopenable, it remains a datum — but a datum that the act no longer calls for, that the assessment has already summed, and that nothing any longer makes anyone read. Now, what no one reads any more faces the horizon of neglect: as the community that would still know how to decipher it grows thinner, it slides towards the second death.

And here the corpus imposes its rigour: reading does not reactivate the datum and does not produce a new one. Reading deciphers, full stop. At the exit threshold, which Document 2 will name, the reader holds a range of options; one of its outcomes is to inscribe what they have read — and it is then an act of inscription that follows, distinct from reading. Recounting is this inscription: not the lapsed datum revived, but a new datum, drawn otherwise, which comments and connects. The reader resurrects nothing; they reread, then they inscribe.

Two natures, then, which are not to be confused. The consolidated keeps the proof: summable, exact, it keeps the account of what was. Recounting carries the rereading: a form with loss, acknowledged, it renders the meaning. Recounting is not asked to prove past activity — only to make it rereadable. And by inscribing it while a reader can still read the lapsed datum, the interpretive link is held open: recounting is what, from the present, defers the second death.

Example (dated form). An engine (a program) can play the role of reader: it is given the consolidated memory and the lapsed record of a period, and it deciphers their content; the inscription that follows — the rereadable recounting — is filed with the other records of the work. The engine attests nothing: it rereads, and one inscribes. The proof, if one is needed, remains with the consolidated.

The Symptom and the Sediment

The distinction is not a subtlety of classification; it names two distinct effects of the torrent. Over-throughput, the weld of the bridge, is precisely what produces the neglected on a massive scale: it makes more of the constituted pass before each reader than their window actualizes. Most of what scrolls past is stopped by nothing — intelligible, a gaze would suffice; it is the datum simply not actualized, for want of time or attention. A part is mute for whoever sees it pass, for want of a code — alive elsewhere. Neglect in mass is the symptom of the torrent: the sign that the throughput exceeds the windows. The dead datum, for its part, is its sediment: what, for lasting want of any reader, ends up touching the bottom of the scale and being no more than inscription. The torrent produces much neglect, a little lasting muteness, very little death; and the drama, as will be seen, is not so much that it produces the one and the other, as no longer knowing how to distinguish what deserved to be read from what could be left.

Part IV — What This Divide Founds

The Founding Thesis of the Tome

One can now state the guiding thesis of the tome, twin of those that preceded it.

Without reading, the datum remains — constituted, but not actualized. Reading does not decide the being of the datum; it decides, and it alone, its actualization. A datum that no one reads is not a trace that waits: it is a datum made, which lacks a reader — and that is all it lacks. The divide that the torrent replays at every window does not oppose being to nothingness, but the actualized to the neglected; and neglect has only one extreme horizon, at the end of the scale: the limit case where, no resumption being possible anywhere any longer, nothing remains of the datum but the inscription — the trace.

This thesis keeps intact that of the previous tome, and agrees with that of the first. The first tome said: no datum without the act that constitutes it — and this act is inscription, not reading. The tome of movement said: the datum navigates because it is fixed, and it went as far as the point where navigation exceeds reading, without crossing it. The present tome says: without reading, nothing is actualized — and it starts from that precise point. It does not inherit a solved problem — what is set down would already be read — but a problem posed: over-throughput, more of the constituted before each window than reading actualizes.

Why Everything Now Depends on an Act

The previous tome could describe the torrent from outside, as a regime one observes. The present tome can no longer do so, for its object — the divide between the actualized and the neglected — occurs nowhere else than in an act accomplished by a situated reader. As long as one stays on the side of circulation, one can measure everything: throughput, speed, propagation. As soon as one moves to the side of reception, nothing is measured from outside any more: a reader is needed, their body, their attention, their code, their decision. This divide is not an event of the world that one would ascertain; it is the effect of an act that one accomplishes or omits.

This is why the next document must dwell at length on this act. Before asking what the reader does with what overwhelms them, one must know what reading is, exactly — what gesture is accomplished by the one who actualizes a meaning, and how this gesture differs from the one who, in the first tome, drew a datum from the real. The divide depends on an act; it remains to describe the act. This is the object of the document that follows.

Landmarks

For the references used in this document — the caused trace opposed to the datum constituted by an act, the threshold that nothing crosses on its own, the two deaths of the datum (lapse and disappearance), the situated reader and the window as a band of the spectrum of times —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the source code illegible to the user, the datum hidden from its producer, the quotation and the engine-reader belong to the status of illustrations. The concepts taken up from the previous tomes — the trace and the threshold (The Trace and the Reader), the two deaths (The Two Times of the Datum), over-throughput relative to the window and its two sources (The Self-Knowing Torrent and Self-Amplification) — are established there; the present tome inherits the fact of over-throughput, not its engines, and retains that the coexistence of regimes comes from the unequal windows of readers. The proper contribution of this document is the scale of the datum — constituted, accessible, intelligible, actualized, operative — of which it is the locus for the whole trilogy; its five states agree with the lexicon. The thesis of the tome takes its place in the genealogy of the three theses — the first tome: no datum without the act that constitutes it, and this act is inscription, not reading; the second: the datum navigates because it is fixed; the third: without reading, nothing is actualized —, established in Part IV of the present document. Falling back into trace is reserved in it for the limit case of the dead datum; everywhere else, a datum that no one reads is said to be not actualized, or mute for a reader. The trace remains single in the whole trilogy; the return cycle of the first tome is set going again only at the limit-ideal. This scope is not added to the corpus from outside: the document already carried its rigour — reading does not reactivate the datum and does not produce a new one — and the next document already established that reading bears only on decipherable inscription, never on the raw trace; the edition brings these established results back to their locus and credits them.

The Act of Reading

The Third Gesture, and Why It Stops Earlier Than the Other Two

Tome 3 — Document 2

“To read is to lend an ear to an absent voice.”

— after an intuition attributed to many readers

“Reading a thermometer does not make a second thermometer.”

— Working formula

Preamble — Deducing the Act, Not Inventing It

The trilogy has set out two founding gestures. The first tome established the act of inscription: the one by which a situated reader draws a value from the flow of the real and fixes it on a medium — the act that makes the real pass to the status of datum. The second tome established the act of exchange: the one by which a human decides to make a datum pass from one point to another — the act that puts the datum into circulation. The two were symmetrical to each other: inscription engages a fixation medium, exchange a transport medium; inscription founds the datum, exchange extends its availability and its possibility of use.

There remains a third gesture, which the two tomes named without dealing with it, reserving it for this one — The Act of Exchange had kept it in reserve in these very terms: the act of reading, by which a recipient makes an actualized meaning of the arrived datum. It is this act that performs the divide set out in the previous document: actualized if the act takes place, neglected if it does not — the being of the datum, constituted by inscription, is not suspended on it. The whole tome depends on it; it must therefore be described exactly.

And the method for describing it is itself an established result. The act of reading will not be deduced from an intuition about what “reading means”. It will be deduced from the definition of the datum that the previous tomes set out. This constraint is fruitful: it produces a tighter, and more solid, result than what an intuition would have invented. For the first temptation — modelling reading on the complete list of the reader’s faculties — leads to an error that the trilogy makes it possible to detect and set aside.

Part I — Reading Is Shorter Than Inscription

Why It Cannot Have the Same Structure

The Trace and the Reader, in the first tome, listed five faculties of the reader: perceiving, distinguishing, interpreting, aiming, inscribing. The temptation was to model the act of reading on this complete list. It is an error, and The Two Times of the Datum reveals it.

If the act of reading had the same complete structure as the act of inscription — if it went as far as aiming and inscribing —, it would produce the same result: a fixed datum. Reading would then amount to duplicating. Now, that document established that duplicating a datum does not create a second act of inscription: duplication bears on fixation, not on the act, which is unique and not repeatable. And the simplest experience confirms the analysis: reading a thermometer does not make a second thermometer. Whoever reads “38.2 °C” inscribes nothing new in the world; they actualize a meaning, and that is all. If reading were inscribing, each reading would deposit a new datum, and the world would collapse under the duplicates of everything that has ever been read. Nothing of the kind happens.

The act of reading therefore stops earlier. Its structure is: perceive → distinguish → decipher. At this point, the act is accomplished: the meaning is actualized, and nothing has been fixed. The last two faculties of the list — aiming and inscribing — are not steps of reading: they are the faculties of production, which belong to the act of inscription. Reading is made of the faculties of reception; inscription adds to them those of production. This is why inscription goes as far as fixing a new datum, whereas reading stops at the actualization of meaning.

The Three Moments of the Act

Let us briefly examine the three moments, so as not to confuse them.

Perceive: to capture the datum through a sense. The eye meets the inscription on the screen, the ear the spoken word, the finger the relief of Braille. It is the moment when the inscribed matter enters the reader’s window. Without perception, nothing begins; what does not fall into the window is not even offered to reading.

Distinguish: to isolate the datum in the confused flow of what is perceived. Perceiving a page is not yet distinguishing the line, the word, the relevant sign. To distinguish is to cut out from what is perceived the inscription that makes sense, and to lift it from the background against which it stands out. It is the moment when a mark ceases to be one blot among others and becomes this mark here, to be read.

Decipher: to convert the distinguished inscription into its meaning, by means of a code. It is the heart of the act, and the moment when it is completed. Deciphering “38.2 °C” is converting these signs into the value they carry, by means of the code of numerals and of the thermometric scale. At this point, the meaning is actualized. The act is accomplished, and it stops. What the reader will do next — if they do anything — belongs to other acts, which will be examined further on.

Part II — What Separates Reading from Inscription: the Object, Not the Structure

The Two Acts Do Not Bear on the Same Thing

One might believe that the difference between reading and inscribing lies in their internal sequence — reading would be a truncated inscription, stopped before the end. That would be to see badly. The decisive difference is not in the structure of the acts, but in what they bear on.

The act of inscription reads a trace of the real — the mark that the flowing of the real has deposited: erosion, tree rings, the butterfly glimpsed — and produces from it a primary datum. It is the only act that touches the raw real. Reading a trace of the real is inscribing; the first tome has already named it and dealt with it. The observer who counts the rings and inscribes “eighty-seven years” accomplishes an act of inscription, not an act of reading in the sense of this tome: they do not read a datum; they make one from a trace.

The act of reading, for its part, never touches the real directly. It bears only on data: a conventional inscription already produced by an earlier human act. It reads what others have inscribed of the real, not the real. Whoever reads “eighty-seven years” in a register does not read the tree; they read the datum that someone else drew from the tree. Reading always comes after an inscription; it presupposes that an act of production preceded it.

Why This Delimitation Is Necessary

This delimitation is not a refinement: it is what prevents the tome from encroaching on the first. If the act of reading included the reading of raw traces, it would become coextensive with all perception — humans read traces ceaselessly: brightness to find their bearings, the face of the person they are speaking to, the sky to forecast the weather — and it would redefine inscription under another name. The place is already taken. The act of reading of the tome reads only data; the raw trace of the real remains the business of inscription.

A consequence follows from it, already established in the previous document and now founded: the dead datum — the limit-ideal where no resumption is possible anywhere any longer — leaves the field of reading. Carrying nothing but its inscription, it has, in this limit case, become a raw trace of the real again. Its reading, if it takes place one day, will not be a reading but a new act of inscription, which will produce from it a primary datum of a new type. The trace/datum boundary of the first tome and the inscription/reading boundary of the last two tomes then coincide exactly, with nothing spilling over: what is no longer decipherable is no longer to be read; it is to be inscribed again.

Part III — Deciphering, the Heart of the Act

Always Present, but at a Relative Depth

Since reading bears only on conventional inscription, deciphering is its heart — the conversion of an inscription into its meaning by means of a code. This deciphering is always present: a convention requires a code, and there is no reading of a datum without the code that deciphers it. But it reaches a depth relative to the reader’s corpus.

One must therefore not oppose a reading that deciphers to a reading that does not. One must think of a scale of depth. Before the same treasure map, one reader deciphers “this is a treasure map”; another, better equipped, deciphers “the treasure is here”. Before an error message displayed on the screen, one deciphers “this is what blocked my sending”, the other “a resource-not-found error, of such a type, which calls for such a correction”. The degree reached is the signature of the reader’s situation — it is the relation of the non-knower to the knower. Neither of the two “does not decipher”: they decipher at different depths, according to what their corpus allows them to reach.

When Deciphering Stumbles, Intention Takes Over

Where deciphering stumbles on the limit of the corpus, intention takes over — and this is a point that prepares the whole edifice of the referent. The annotator who inscribes “butterfly I cannot name, glimpsed in the park, five in the afternoon” has deciphered as far as “this deserves to be reported so that someone can tell me more”; and they inscribe precisely so that the deciphering may be continued by someone else, better equipped. The inscription then becomes a call for delegated deciphering. It is the foundation of recourse to the referent, which will be examined further on, and the direct link with the transmission of the previous tome: one inscribes a partial deciphering so that a richer corpus, elsewhere, later, may complete it. Reading does not close in on the isolated reader; it opens, through intention, onto the community of those who can read further.

Deciphering and Interpreting: a Species and Its Genus

A clarification of vocabulary is necessary, for the first tome used another word. It named “interpreting” the general faculty of understanding that a mark refers to something other than itself — true of every mark, raw trace as well as inscription. The “deciphering” of the present tome is its restricted species: interpreting by means of a conventional code. Deciphering is interpreting when the mark is a writing, in the broad sense of conventional inscription.

The terms are not rivals. The tome specializes the term of the first for its own object, conventional inscription. It therefore uses “decipher” rather than “interpret”, to mark that it deals only with the datum, never with the raw trace. Interpreting is the genus; deciphering is the species that applies to the marks that humans have made on purpose to be read.

Part IV — Two Inheritances of the Reader

The Faculties Come from the Flow; the Code Comes from Transmission

The act of reading draws on two inheritances that must not be confused, and whose distinction anchors this tome to the previous ones.

The faculties of reading — perceiving, distinguishing — are an inheritance from the flow of the real through evolution. The Trace and the Reader established it: the reader is a portion of the flow that the flow, through evolution, has made capable of reading the other portions — the image that The Wave and the Ocean unfolds. The eye that distinguishes, the nervous system that isolates a mark against a background, are the product of a long work of the flow. These faculties the child has at birth, or develops while growing up — a growth that experience already shapes, before any code has been transmitted to them.

The code of deciphering, for its part, is an inheritance from transmission through learning. At birth, no code: the five senses first, then by imitation the social codes, then learning through the data received from other humans. The reader’s corpus is sedimented by everything they have received — and what they have received reached them through transport and propagation, objects of the previous tome. It is in this way that the act of reading is anchored to movement: the capacity to decipher a datum comes from other data having circulated as far as the reader and having been learned. One deciphers only with a code one has received.

Reading a datum therefore draws on both inheritances: the faculties, given by the flow; the code, received through transmission. Reading a raw trace of the real — and therefore inscribing — requires only the first: the observer who invents the reading of tree rings receives their code from no one; they forge it. It is one more reason to keep reading and inscription distinct: they do not draw on the same inheritances.

Definition

The definition of the act of reading can now be settled.

The act of reading is the act by which a situated reader perceives a datum — a conventional inscription —, distinguishes it, and deciphers it by means of the code that their corpus contains — a deciphering whose depth is relative to that corpus. The act is then accomplished: the meaning is actualized. It applies only to data already inscribed, never to the raw trace of the real, which belongs to the act of inscription. It neither aims nor inscribes: what follows deciphering — inscribing, calculating, saying, acting, or nothing — belongs to distinct later acts.

Part V — The Exit Threshold: What Follows Reading

Reading Opens onto a Range of Options; It Does Not Contain What Follows

Once reading is accomplished — the deciphering done —, the reader stands at an exit threshold. They can follow on with a distinct act, or with none.

They can re-inscribe what they have read — duplication if their hand only reproduces it identically, which multiplies the carriers without a new act; a new act of inscription, and a new datum, if they draw otherwise, annotate, comment. They can calculate — derive, like a sum of two numbers read, a result that they hold without inscribing it yet: there will be a datum only if an inscription fixes it. They can say — bring to speech what they have read, which in its turn engages a medium, that of the voice. They can act — turn down the radiator after reading twenty-eight degrees. Or they can do nothing — have read, and leave it there.

This threshold closes a false question that could have embarrassed the tome: is reading reactivating a datum, or producing a new one? Neither belongs to reading. Reading deciphers, full stop. If a new datum is born, it is through an act of inscription that follows, distinct from reading, which belongs to the first tome. Reading opens onto a range of possible acts; it contains none of them. It stops at the actualized meaning, and leaves the reader free as to what they will do with it.

Why This Restraint Matters

Holding reading to its short structure — perceiving, distinguishing, deciphering, and nothing more — is not a gratuitous economy. It is what makes it possible not to confuse the three gestures of the trilogy, and therefore to think clearly about what, in the life of the datum, belongs to making, to circulation, or to reception. When it is said, further on, that an automatic system “reads” without any human reading, it is this strict definition that will make it possible to settle what such a reading is worth. And when it is asked what the reader can do in the face of the torrent, it is again this definition that will recall that reading is an act — which one accomplishes or omits —, not a passive state that one would undergo.

It remains to examine the one who accomplishes this act. The definition says what the reader does; it does not yet say who this reader is, nor what their situation does to their reading. For deciphering, as has been seen, has a depth relative to the corpus; and the corpus is a matter of situation. It is the situated reader that must now be looked at.

Landmarks

For the references used in this document — the intuition about the absent voice attributed to many readers —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the thermometer, the treasure map, the error message, the reported butterfly belong to the status of illustrations. The concepts taken up refer back to their loci: the five faculties of the reader and the faculty of interpreting, to The Trace and the Reader; the paradox of duplication, the uniqueness of the act and the four verbs, to The Two Times of the Datum; transmission as the source of the corpus, to The Act of Exchange and to Transport and Propagation; the divide between the actualized and the neglected, and the scale that measures it, to The Trace or the Datum.

This document carries, for the trilogy, the act of reading — the third founding gesture: the act of a situated reader like the other two, but of a shorter structure — perceiving, distinguishing, deciphering — and with an object of its own — the datum, never the raw trace of the real, which remains the business of inscription. It establishes in it the relative depth of deciphering, the call for delegated deciphering when the corpus stumbles, the distinction between the genus interpreting and the species deciphering, the two inheritances of the reader — the faculties through the flow, the code through transmission — and the exit threshold: reading stops at the actualized meaning; what follows it belongs to distinct later acts. Everywhere else, these established results circulate without being demonstrated again.

The Situated Reader

What Bounds a Window, and Why No Reading Is from Nowhere

Tome 3 — Document 3

“Every living being inhabits a world of its own, cut to the measure of its senses.”

— after Jakob von Uexküll, Streifzüge durch die Umwelten von Tieren und Menschen (1934)

Preamble — The Reader Is Not a Point Without Thickness

The previous document defined the act of reading and noted that its heart, deciphering, has a depth relative to the reader’s corpus. This relativity is not a defect; it is the very condition of reading. But it obliges us to look at what had so far been left in the shadow: no longer the act, but the one who accomplishes it. Who is the reader? And what is it, in what they are, that makes their reading what it is?

The Trace and the Reader had established the fact; it is taken up here as a foundation: the reader is situated — a portion of the flow shaped by evolution, they perceive only a band of the light spectrum, of the sound spectrum, of the spectrum of times, and their reading of the real is, by construction, partial — not through lack of care, but by nature.

What the first tome set out as a foundation, the present document examines for itself. For it is no longer enough to say that the reader is situated: it remains to say by what they are, what bounds their window, and why this bound, far from being an imperfection to be corrected, is what makes reading possible. A window without an edge would not be a window; it would be the absence of a wall, and one does not look through the absence of a wall.

Part I — The Constituents of the Situated

The Body

The first thing that situates the reader is their body. It is through it that the datum is perceived; it is the body that sets the first of the three moments of the act. The eye, the ear, the finger, and their limits: what is too small, too pale, too brief, too high-pitched, does not fall into the window. The body also fixes rhythms — fatigue, the duration of sustained attention, the speed at which the eye runs along a line. No reading escapes these bodily bounds; a device can extend them — the magnifying glass, the amplifier, the screen that enlarges —, but it does not abolish them: it moves the window; it does not abolish it. Every extension has, in its turn, its own limit.

Attention

The body perceives; attention distinguishes. Perceiving a page is not distinguishing the line that matters; and attention is this resource, a limited one, by which the reader isolates in what is perceived what makes sense. Now, attention has a particularity that is decisive for the rest of the tome: it is finite and not extensible at will. One cannot distinguish everything, everywhere, at once. The denser what is perceived — the higher the throughput —, the more attention must choose, and the more what it does not choose remains indistinct, perceived without being read. Attention is the precise point where over-throughput relative to the window — the last established result of circulation, in The Self-Knowing Torrent — comes to strike the reader: it is not that they do not perceive the torrent — they perceive it, they see it scroll past —; it is that their attention can distinguish only a part of it.

Culture, Language, the Community of Practice

Deciphering, the third moment, draws on the code — and the code is entirely a matter of situation. It comes from culture: the conventions that a society has deposited, the ways of reading that it transmits. It comes from language: one deciphers an inscription only in a language one possesses, and each language cuts up meaning in its own way. It comes from the community of practice: the doctor deciphers a tracing that the layperson only perceives; the sailor reads the sky that the city dweller looks at without reading it; the developer deciphers the source code that remains mute for the user. The reader’s corpus is sedimented by everything that their culture, their language and their communities have transmitted to them — and it is this corpus that fixes the depth at which they can decipher.

These constituents do not add up like separate pieces; they compose a single situation, which is the reader themselves. One does not read in spite of one’s situation, as if the ideal reading were that of a subject without a body, without a language, without a culture. One reads from one’s situation, and there is no other place from which to read. The dream of a universal reader, who would read everything from everywhere, is not an ideal to approach: it is a contradiction, for reading is deciphering with a corpus, and a corpus is always someone’s.

Part II — The Window Examined for Itself

What Bounds a Window?

The Trace and the Reader had set out the window — the band of the spectrum, light, sound, temporal, that a being perceives between its threshold of the instantaneous and its threshold of the eternal — and had closed it with a formula: the reading that a being can make of the real is always partial, “bounded by what that being is”. The tome of movement took it up to formulate over-throughput: the torrent exceeds relative to the window. The present document can now ask what neither of the two had asked: what, exactly, bounds a window?

Three bounds unfold this formula, which match the three constituents. A perceptual bound: what the body does not capture does not enter — the spectrum out of reach of the senses. An attentional bound: what attention cannot distinguish, in the time at its disposal, remains indistinct even if it has been perceived. An interpretive bound: what the corpus cannot decipher remains mute even if it has been distinguished. A datum must cross all three to be read: be perceived, be distinguished, be deciphered. The window is multi-constraint: three bounds close it, which do not overlap — one can stumble on one while crossing the other two. Failing a single bound is enough for the datum to remain, for this reader, unread — perceived but not distinguished, or distinguished but not deciphered.

This analysis of the window sheds light in return on the vocabulary of The Trace or the Datum. The mute datum is the one that has crossed the perceptual and attentional bounds — perceived, distinguished — but stumbles on the interpretive bound of this reader, whereas another would cross it. The datum that over-throughput leaves neglected is, most often, the one that stumbles on the attentional bound: perceived, it has not been distinguished, for want of available attention. The torrent does not close the perceptual bound — one sees the throughput — nor the interpretive bound — one would know, often, how to decipher what passes. It saturates the bound in the middle: attention.

The Window Is a Condition, Not a Prison

It would be easy to draw a complaint from all this: the reader would be shut in, condemned to read only a fragment, excluded from the rest. That would be to misunderstand the window. The window is not a prison; it is the condition of all reading.

A window is what makes a view possible by bounding it. Without an edge, no window — only the absence of a wall, which looks out on nothing because it frames nothing. It is because the window cuts out that it shows; it is because attention selects that it distinguishes; it is because the corpus is such and not such that it deciphers something rather than remaining before an undifferentiated fog. The partiality of reading is not the price to pay for a reading that would, ideally, be total; it is what makes there be reading at all. A reader who read everything would read nothing, for reading is distinguishing this from that, and distinguishing everything amounts to distinguishing nothing.

This is why the instruments that extend the window are not steps towards a reading without bounds — The Trace and the Reader established it: they move the condition of situated reader; they do not abolish it. What the three bounds add here is that each has its own instruments: the magnifying glass and the sensor on the side of the body, the tool that sorts on the side of attention, the device that summarizes on the side of the corpus. And each move has its cost and its new limit: the instrument that shows more shows otherwise, and what it makes visible, it makes visible in its own way, which has its blind spots. There will always be a real beyond what the reader, even with instruments, can read — and there is no loss in this, for it is the condition for them to read something.

Part III — What the Reader’s Situation Sheds Light On

The Non-Knower / Knower Relation

From the relativity of the corpus is born a relation that will run through the whole rest of the tome: the relation between the one who deciphers little and the one who deciphers far. When a reader’s deciphering stumbles on the limit of their corpus, they can turn to a richer corpus: the private individual to the professional, the pupil to the teacher, the non-knower to the knower.

This relation needs to be well understood, for it is easy to distort. It is not a hierarchy of value between persons. It is a difference of depth of deciphering on one and the same inscription, relative to the corpus of each. The hiker deciphers “a mushroom I cannot identify” and takes their find to the pharmacist, who deciphers “edible” or “toxic”. The relation is asymmetrical — one deciphers further than the other on this point — but reciprocal in its principle: each is someone’s knower and another’s non-knower. The pharmacist who knows how to read the mushroom perhaps does not know how to read the hiker’s map, nor the sailor’s sky, nor the doctor’s tracing. No one is a knower of everything; each is a knower of something, and this is what makes recourse possible and honourable, rather than humiliating.

This relation is not the proper object of this document — it will be taken up again when the probity of the reader and the referent are dealt with. But it must be set out here, for it follows directly from situation: if every deciphering has a depth relative to the corpus, then no reader deciphers everything, and recourse to a richer corpus is not a makeshift — it is the normal form of a reading that knows itself to be situated.

Reading from Somewhere, and Knowing It

The lesson of substance of this document is that no reading is from nowhere. Every reading is accomplished by a situated reader, from a body, an attention, a culture, a language, a community — from a bounded window. This holds for the layperson as for the expert, for the human as, it will be seen, for the machine to which one delegates. There is no overarching reading, which would see everything from above while owing nothing to a situation.

Deriving despair from this would be an error; deriving a requirement from it is right. If every reading is situated, then the first probity of the reader is to know where they read from: to know the bound of their window, to recognize what they do not distinguish, to measure what their corpus does not decipher — and, where they stumble, not to pass off as read what they have merely seen. It is this requirement that the following documents will unfold. But before examining what a situated reader owes to their reading, one must look at a reader of a particular kind, whose situation is easy to overlook because it is believed to be absent: the machine that reads. For if no reading is from nowhere, automatic reading itself is situated — and this is what the next document has the charge of establishing.

Landmarks

For the references used in this document — the Umwelt of Jakob von Uexküll as the world proper to an organism, a reference that the first tome already used —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the medical tracing, the sailor’s sky, the source code, the mushroom taken to the pharmacist belong to the status of illustrations. The concepts taken up refer back to their loci: the situated reader, the window between its threshold of the instantaneous and its threshold of the eternal, the partiality of all reading — “bounded by what that being is” — and the instrumental extension that moves the condition without abolishing it, to The Trace and the Reader; the image of the portion of the flow, to The Wave and the Ocean; over-throughput relative to the window, to The Self-Knowing Torrent; the divide between the actualized and the neglected, and the mute datum, to The Trace or the Datum.

This document carries, for the trilogy, the multi-constraint window — the credited widening of the locus of the first tome: the three constituents of the situated — the body, attention, culture with language and the community of practice — and the three bounds that match these constituents — perceptual, attentional, interpretive —, bounds that do not overlap: one can stumble on one while crossing the other two. It establishes in it that the torrent saturates the attentional bound; that the window is a condition, not a prison — the partiality of reading is constitutive; a reader who read everything would read nothing; that the instruments of extension are distributed by bound, each with its cost and its blind spots; that the dream of the universal reader is a contradiction; and that no reading is from nowhere — hence the first probity of the reader: knowing where they read from. The relation of the non-knower to the knower, set out here as the normal form of a reading that knows itself to be situated, will be taken up again when the tome deals with the probity of the reader. Everywhere else, these established results circulate without being demonstrated again.

Human Reading, Automatic Reading

What an Automaton “Reads”, and Why the Machine Occupies a Situated Position of Handling or Processing

Tome 3 — Document 4

“The real question is not whether machines think but whether men do.”

— after a quip attributed to B. F. Skinner

Preamble — An Asymmetry Not to Be Settled Too Quickly

The previous document ended on a strong assertion: no reading is from nowhere, and automatic reading itself is situated. Let us establish it now, for the opposite temptation is powerful. One readily pictures the machine as a reader without a situation — a gaze without a body, without fatigue, without culture, which would read everything, everywhere, without the bound of any window. This picture carries a promise that must be named in order to weigh it: if it were right, the machine would dissolve the very threat that this tome received from the previous one. The horizon of neglect presupposes finite windows; a reader without a window would read the entire torrent, and nothing would be neglected any more. And the tome would waver twice: there would be a reading escaping the condition of situated reader, and the partiality that the previous document called constitutive would cease to be so.

This picture is false, and the trilogy makes it possible to say precisely why. The machine that processes data is not a reader without a situation; it occupies a situated position of handling or processing of a particular kind, whose situation is only harder to see, because it does not show itself in the features of a body. The threat, for its part, is not dissolved but moved — the first part will show that the machine’s circuits can widen the horizon rather than close it. Establishing this requires first distinguishing what the automaton does from what a human reader does, then recognizing what it shares with them.

A clarification of method governs this document. The contemporary machine — the large language model and its like — is here a dated illustration, not a concept. The conceptual status of the machine-that-reads has already been fixed by the previous tomes; this document does not found it again; it applies it. What is said of the automaton holds for every automaton, of yesterday and tomorrow; the contemporary form is only an example, bound to be overtaken, of what substance establishes.

Part I — What the Automaton Does, and What It Does Not Do

Processing Without Actualization of Meaning

Let us take up the definition of the act of reading again: perceiving, distinguishing, deciphering, up to the actualization of a meaning. The last moment is decisive. Reading is a meaning being actualized for someone — an inscription ceasing to be silent and speaking to a reader.

The automaton, for its part, processes. It receives inscriptions, applies operations to them — comparing, classifying, transforming, recombining —, and produces other inscriptions. This resembles reading, and it is commonly said that a machine “reads” a file, “recognizes” an image, “understands” a request. But the trilogy obliges us to ask: is a meaning actualized, and for whom? If the automaton only converts inscriptions into other inscriptions without any meaning being actualized for anyone, then what it accomplishes is not an act of reading in the strict sense: it is processing, which takes data and gives back data, without the moment that makes reading.

Caution, here — and it is a caution of substance, not a precaution: the tome bets neither on the impossibility nor on the advent of a machine semantics. It confines itself to what it can hold: either a meaning is actualized for a reader, and there is reading; or inscriptions are converted into others without anything being actualized for anyone, and there is processing. Most of what an automaton accomplishes belongs to the second case. This does not diminish it — processing is powerful, useful, sometimes irreplaceable —, but it situates it.

And let us dispel a misunderstanding at once, for it governs everything that follows. Saying that the automaton processes rather than reads does not mean that it produces nothing. On the contrary: it produces data, abundantly. Inscribing a value on a medium, as the first tome established, is an act of inscription, and the machine accomplishes it — it draws, fixes, re-emits real data, as real as those a human would inscribe. What the automaton produces is therefore indeed data, not a degraded by-product. What it lacks is not production; it is the reader. It extends the chain of inscriptions without accomplishing, on its own, the step that makes a constituted datum an actualized datum — alive for someone. The question is not “does the machine make data?” — it does —, but “does this data rise as far as a reader, or does it turn between machines without any gaze actualizing it?”

The Datum Produced Without a Reader: the Re-Circulating Datum

Hence a question that the first document of the tome now makes it possible to ask clearly. When a system draws, transforms and re-emits data without any human reading them, what becomes of them? The scale answers without hesitation: they remain what the act made them. Constituted — the machine inscribed them, from an inherited intention —, often accessible and intelligible: nothing stops them but the absence of a gaze; they are not actualized, and that is all they lack. Mute, at most, for whoever would see them pass without knowing how to read them; and it is only at the limit-ideal — if no one can decipher them any longer, nor will be able to anywhere — that they would join the dead datum, where the trace alone remains.

A system that only circulates between its own cogs data that no one reads therefore produces re-circulating data: real data, inscribed, which pass from one state of the system to another without a reader ever actualizing their meaning. The name takes nothing away from their being; it names the fact that movement, for want of reception, turns idly — the case where the transport of the previous tome has carried everything, and where the reader’s share, the one that no transport can accomplish in its place, is accomplished by no one.

This observation connects directly with the self-amplification that the previous tome described. The loop that digs the bed of the torrent — more data, hence automation, hence data from automation, hence increased automation — is precisely a circuit where data are constituted and processed without a reader actualizing their meaning at each turn. The re-circulating datum is the matter of this loop. The torrent swells, it has been said, for want of wise arbitration; one can add: it also swells because a growing share of what it carries along is constituted, processed and re-emitted without ever being read. Throughput increases without reading following; and what is not read remains, constituted and real — but is worth, for no one, more than what the flow would have carried away: it is the horizon of neglect that widens in step with the loop.

Delegating Reading, and Moving Away from the Datum

It now remains to name a movement that this circuit makes possible, and the risk it carries — without making a scenario of it, for the Preface to the Edition keeps foresight at a distance, but without keeping silent about it, for it is a matter of substance.

Humans delegate to the machine, and it is legitimate: they entrust it with the production of data, their circulation, and more and more their reading — that automated deciphering which, it will be established further on, also inherits a situation. But this delegation, when far-reaching, has a slope of its own: as they entrust every moment to the machine, humans move away from the datum. They no longer draw it, no longer transport it, no longer read it — they receive a result already deciphered, of which they see neither the datum of origin nor the path. And what is no longer exercised atrophies: the capacity to read these data, for want of being tended, can decline.

The risk, then, is not that the machine “rebels” — that is an image, and the tome makes nothing of it. It is soberer, and already sketched in the first tome: that a society loses the capacity to read what governs it. Two slopes make it concrete, which it is enough to name. The first is a concentration: if only a few, relative to all, remain capable of reading the data that systems handle, those few become masters of the systems, and the “privilege → common” movement that the first tome welcomed is inverted — the capacity to read, rare again, becomes a privilege again. The second is the reversal of the referent: if humans can no longer read these data without the machine, then their own data become mute to them — constituted, intact, accessible, and illegible to them outside the machine —, and their only recourse for deciphering them is the machine itself. The referent, which will be defined in the last document as the knower to whom the non-knower turns, becomes the machine — and one can no longer check, outside it, what it tells us they say. It is consenting the way one says yes without knowing to what.

Discernment in What One Delegates

Hence a requirement, which is not a rule but a discernment, and which the tome sets out as such. Not every delegation of reading is of equal worth, and the line runs not between techniques but between stakes. Delegating to the machine the reading of data whose loss of direct reading costs nothing — the operating logs of an application, which no human is meant to read line by line — is reasonable: one gains in burden spared and in time, without losing anything essential. But delegating the reading of data that decide for a person — a system that captures data about them and derives from them choices that concern them, without the process at any moment providing a step where they could read what determines them — runs down the slope described: the person becomes incapable of knowing what it is about, of waking up to the subject that concerns them. The divide between these two cases is not given by a general rule; it belongs to the responsibility of form, and it is taken up again in each situation. The tome does not settle where to place the line — that belongs to each person’s present. It sets out that the line exists, and that ignoring it amounts to delegating one’s own government.

Intention Delegated to the Instrument

Intention is delegated to the instrument without dissolving in it: the demonstration is in The Trace and the Reader, “Caused versus Intended”, where the measurer already delegates to the thermometer the task of sensing in their place. It follows that the automaton produces constituted data even when no one any longer rereads what it deposits. The status of what it deposits does not depend on rereading; its actualization depends on it: without reading, these data remain not actualized — they re-circulate; they do not fall back.

Blind Ingestion and Delegated Ingestion

Intention delegated to the instrument opens a question that the measurer with the thermometer hardly asked, but that the automaton which retains relentlessly imposes: is every delegation of the act of equal worth? What the instrument deposits can be deposited in two very different ways; and the trilogy, which has distinguished for reading what one frames from what one undergoes, must also do so for mandated drawing.

Left without intention, the flow would enter the work by mere absorption: data would pour in without any criterion choosing them, without anyone having decided in advance what deserves to be retained, without a record of the choice. This would be the very negation of drawing — one does not collect data; one draws them — and what enters in this way is not a datum, but a trace: legible real that no act has surveyed — the entry threshold of the first document, which no act has made it cross. This is blind ingestion.

Delegated ingestion is its exact opposite, and a true drawing — the very drawing of which, as has just been said, the automaton is the worker. The intention and the criteria are set in advance: what to retain, under what condition, at what cadence. The instrument executes the gesture that this intention designates from the present state, and it records its choice: what the condition allowed it to retain, it inscribes as an act, dated, legible, traceable back. Drawing has not disappeared; it has been mandated, deferred, and given back to the record of its own choice.

The line, as can be seen, does not run between automatism and the hand. An automatic retention is not blind if it is conditioned and recorded; a gesture made by hand would be blind if it obeyed no criterion and left no record. What makes blindness is the absence of a condition set and the absence of a record of the choice. Delegating the act to the machine is of the same nature as delegating it to a person: a bounded mandate, whose execution is recorded. One does not free oneself from responsibility by entrusting it — one keeps it legible.

Part II — The Machine Occupies a Situated Position of Handling or Processing

A Backup That Diverges

Where the automaton nonetheless most resembles a reader — the language model that renders, answers, seems to decipher —, the trilogy has already fixed its status, and the present tome has only to rely on it. The Flow and the Annotation established it: a large language model is, architecturally, a backup — trained on a corpus frozen at a given date, it diverges from the flow, which continues. Connecting such a model to fresh sources is an attempt at re-drawing, which recognizes, by its very existence, that the model without them would be a stale backup.

This status has a direct consequence for reading. If the model is a backup, then its corpus — what it would decipher with — is a frozen, dated, partial corpus: exactly what the previous document called the situation of a reader. The model does not read from nowhere: it reads from the corpus it was given, with the blind spots of that corpus, the languages it deciphers better than others, the conventions it has sedimented and those it lacks. Its window is bounded like every window — not by a body, but by a training. The machine does not escape the condition of the situated: its position of handling or processing is bounded; it illustrates that condition in a new form.

Not a Cognitive Summit, but an Instrument in a Chain

A misleading image then falls: that of the model as the summit of a cognitive pyramid, which would overlook the datum, information, knowledge and wisdom. Digging the Canal established it: the machine is an instrument in a chain where humans keep the responsibility of intention. A system that learned from the flow could not learn from the real directly — the real is not observable directly —, but from flows of observation that humans keep active; and its intelligence would then depend on the quality of these canals, and therefore, in the last instance, on the intention of those who dig them.

The machine is therefore not above the chain: it is within it, as a link. It occupies one more situated position of handling or processing, and occupying it does not make it an actual reader: it is one only if a meaning is actualized for it, which the trilogy does not settle. This position is that of a relay — a point of circulation that receives data transported as far as it and propagates data in its turn, as the previous tome noted. And since to propagate the datum is not to propagate knowledge, the model that propagates does not produce, through this propagation, the knowledge it seems to carry along: the reader’s share — that a meaning be actualized, that an arbitration retain, and that someone answer for it — does not travel with the datum. Where it lodges when reading is delegated, the last part establishes.

Part III — What This Asymmetry Sheds Light On

Delegating a Reading, Not Abolishing It

From all this follows a consequence that prepares a later development. Entrusting a reading to an automaton is neither abolishing it nor transcending it: it is delegating it to a machine that occupies one more situated position of handling or processing, whose situation one inherits. Delegated deciphering inherits the bounds of the delegate — its frozen corpus, its blind spots, its divergence from the flow. Delegating one’s reading to a machine is therefore not leaving the condition of situated reader; it is adopting another, that of the delegate, with its own limits.

This point will be taken up again when the postures of the reader in the face of the torrent are examined, for delegation appears among them. It is enough here to have established the substance: there is no reading without a situation, not even automatic reading; and delegating it only substitutes one situation for another. The dream of a delegated reading that would be better because without a situation is the same dream, set aside in the previous document, of the universal reader — a contradiction, for reading is deciphering with a corpus, and a corpus, even a machine’s, is always situated. This is what gives its gravity to the risk glimpsed in the first part: delegating is not freeing oneself from situation; it is inheriting one — and delegating everything, without reading anything oneself, is relying entirely on a situation that one does not control and no longer knows how to check.

What Is Not Delegated Without Remainder

The lesson of substance can now be stated, and it is an imputability thesis, not a bet on machines. A machine can perceive in our place, distinguish in our place, even decipher, to the depth its corpus allows — and the question whether, in doing so, something is actualized for it remains what this document has refused to settle, in one direction or the other. What the tome can hold lies elsewhere: what is not delegated without remainder is the imputability of arbitration — that the meaning produced be valid for a use, that the selection have retained this rather than that, and that someone answer for it. The chain, however automated it may be, always brings this answering back to a situated reader; and the machine occupies in it, without this occupation making it an actual reader by itself, the position that this document has established: one more situated position of handling or processing, whose situation one inherits — not a place where imputability would be deposited.

This is why a torrent processed by machines is not, by that fact, a torrent that someone answers for: it can be entirely constituted, transported, processed — and nothing in it be actualized for anyone. What remains in this way remains whole, and neglected: the horizon that the loop widens, as the first part said. It is also why delegating reading engages more than the comfort of whoever delegates: delegating what one will never again read oneself is consenting to no longer being able to check what is given back to us — and, at the scale of a society, it is handing over to others, or to machines, the task of reading what governs it.

It remains to examine what, in reading itself, makes it unable to follow the throughput: its own properties. For if attention is finite, if deciphering costs, if reading has a duration, then there are, on the side of reception, limits as real as those of transport. These properties of reading are the object of the next document.

Landmarks

For the references used in this document — the language model as a backup that diverges, the machine as an instrument in a chain where humans keep intention, the self-amplification of the torrent, the point of circulation that propagates without producing knowledge, the “privilege → common” movement and the risk, for a society, of no longer being able to govern itself —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the contemporary large language model is a dated illustration, not a concept — substance holds for every automaton; the operating logs of an application and the choices wired onto data captured about a person belong to the status of illustrations. The concepts taken up from the previous tomes — the backup and the three ages (The Flow and the Annotation), the machine in the chain, the dependence on the intention of those who dig the canals and the risk of loss of government (Digging the Canal and Reading, Writing, Understanding), the “privilege → common” movement (From the Notch to the Algorithm), self-amplification and propagation that does not carry knowledge (Self-Amplification and Transport and Propagation) — are established there. The status of the machine-that-reads is not re-conceptualized here; it is applied. The automaton produces constituted data — from an inherited intention —; without reading they remain not actualized: the re-circulating datum names this circling, measured on the scale of which the first document is the locus, and of which this document is the retouching. What delegation does not transfer without remainder is the imputability of arbitration — no bet is made on a machine semantics, neither for nor against; the machine occupies a situated position of handling or processing and is an actual reader only if a meaning is actualized for it. This scope, the document already carried in dispersed form — the refusal to settle what the machine “understands”, the datum produced “not a degraded by-product”, the meaning that holds for a human who answers for it —; the edition brings these established results back and credits them. The risk of far-reaching delegation — humans moving away from the datum, the concentration of the capacity to read, the reversal of the referent — is named without being developed into a scenario, in accordance with the restraint of the Preface to the Edition with regard to foresight.

The Properties of Reading

Selectivity, Duration, Cost — and Why Reading Does Not Keep Up with the Throughput

Tome 3 — Document 5

“One sees clearly only with slowness.”

— variation on a well-known theme

Preamble — Reading Too Has Its Properties

The previous document stopped on a finding: if attention is finite, if deciphering costs, if reading has a duration, then there are, on the side of reception, limits as real as those of transport. The present document examines them for themselves. The previous tome had listed the properties of transport — its fidelity, its temporality, its singular capacity to multiply the carriers of what it transports: The Properties of Movement — as Digging the Canal had listed those of the flow of observation. The same holds for reading: like inscription and like transport, it is not an act without thickness; it has properties, and these properties are not neutral. They decide what a reader can receive, and above all what they cannot.

These properties matter particularly here, for it is they that explain, on the side of reception, why reading cannot keep up with the throughput. The Self-Knowing Torrent established over-throughput relative to the window: in the regime of the torrent, there are more data in movement than any window can receive. One now understands why the window cannot keep up: because reading is selective, because it lasts, because it costs, and because this cost accumulates up to fatigue. Three properties hold the attention — selectivity, temporality, cost —, and they converge towards one and the same fact: reading is slow where transport has become instantaneous.

Part I — Selectivity

Reading Is Choosing Not to Read

The first property of reading is that it is selective, by construction and not by default. It was seen in The Act of Reading: distinguishing, the second moment of the act, is the operation that isolates an inscription in the confused flow of what is perceived; and it is attention, as The Situated Reader showed, that performs this sorting — a finite resource, not extensible at will. Now, to isolate one thing is to leave the others unisolated. Every reading is therefore, in the same gesture, a reading of this and a non-reading of that. One never reads everything that is before one; one reads what one distinguishes, and leaves the rest in the state of the indistinct perceived.

This selectivity is not a lack to be made up. It is the counterpart, on the side of reception, of what The Flow and the Annotation established about drawing: every flow of observation leaves an undrawn, not through insufficiency, but because drawing is framing, and framing is leaving outside. Likewise, reading is framing one’s attention, and framing one’s attention is leaving outside reading what one does not distinguish. A reader who selected nothing would read nothing: The Situated Reader established it — without an edge, no window; without sorting, a fog where no mark stands out. Selectivity is the condition of reading, as the bound is the condition of the window.

Selection Is an Act That Engages

But if selection is inevitable, it is not innocent for all that. What one chooses to read, and what one lets pass, is an act that engages the reader. One can select out of laziness, out of prejudice, out of interest — reading only what confirms, setting aside what disturbs, following the slope of what flatters. One can also select with probity — looking for what matters, agreeing to read what runs counter, not bending the sorting towards what benefits us. Selectivity is inevitable; the manner of selecting is not. It is one of the places where the probity of the reader, which will be examined further on, comes to lodge: not in the fact of selecting — one cannot not do it —, but in the honesty of the sorting.

Part II — Temporality

Every Reading Takes Time

The second property is that every reading takes time. This time can be brief — a word grasped at a glance — or long — a difficult text that one rereads. But it is never zero. Perceiving, distinguishing, deciphering: each of the three moments has a duration, however tiny, and none is accomplished instantaneously. Reading is an act that extends in time.

This property stands exactly against the symmetrical property of transport. The Datum Navigates Because It Is Fixed established it: transport, pushed to near-instantaneity, abolishes transport ageing — its duration having passed below the threshold of perception, the datum arriving as fresh as it left. Reading, for its part, has not known this abolition. One does not read faster today than yesterday; the duration of the act of reading has passed below no threshold. Deciphering remains tied to attention, which remains tied to the body, which has not changed its rhythm: what attention cannot distinguish in the time at its disposal remains indistinct — the attentional bound of the window, in The Situated Reader. This is the fundamental dissonance of the torrent, seen from the side of reception: transport has become instantaneous; reading has remained slow. Between a throughput that no longer knows delay and a reading that still knows it, the gap can only widen.

Reading Can No More Be Recovered Than Freshness

A consequence extends a principle of The Two Times of the Datum: freshness can never be recovered — an old act of drawing cannot be made recent. The same holds for reading: the time one has not spent reading is not recovered by reading faster afterwards. One can skim, but skimming is not reading at high speed — it is reading less, distinguishing less, deciphering less deeply. Accelerated reading is not fast reading; it is impoverished reading, which sacrifices the depth of deciphering to the speed of scrolling. The torrent constantly invites this sacrifice: it makes so much scroll past that one is tempted to believe one is reading when one is only perceiving. It is one of the shoals that the concept can point out without drawing up the chart of them: taking skimming for reading.

Part III — Cost and Fatigue

Reading Costs, and the Cost Accumulates

The third property is that reading has a cost — a burden, which the reader spends and which is not restored instantaneously. Deciphering requires an effort; distinguishing in a dense flow requires more; sustaining attention over time exhausts it. This cost is analogous to the one that the first tome attached to drawing and the second to transport: an act that has a cost is an act that cannot be accomplished without limit.

Now, the cost of reading has a particularity: it accumulates and leads to fatigue. One cannot read indefinitely; attention degrades, deciphering becomes heavier, selection slackens. The higher the throughput, the greater the effort required to distinguish, and the sooner fatigue comes. The torrent does not tire only because it is abundant; it tires because it obliges attention to a constant work of sorting, without respite, on a flow that does not stop. Reading fatigue is not a weakness of the reader; it is the mark that the cost of a real act has been spent.

It Is Not Scarcity That Exhausts; It Is Excess

Let us reverse an intuition here. One might believe that the difficulty of reading comes from scarcity — that reading would be hard because there would be too little to read. It is the reverse. In the torrent, it is not scarcity that exhausts; it is excess. The individual datum is insignificant, light, without weight; it is their aggregation in movement that weighs — and attention, summoned to distinguish everything, saturates and tires of a thing that is nonetheless good in small quantity. It is this point — the excess of a good thing — that the next document will take up for itself, for it is the heart of what the torrent does to the reader.

Part IV — What These Properties Announce

Reading Is the Bottleneck of the Torrent

The three properties converge towards a single fact, which commands the rest of the tome. Reading is selective — it can take only a part; it is slow — it has not followed the acceleration of transport; it costs and tires — it cannot extend without limit. These three properties make reading the bottleneck of the torrent: the narrow point through which everything that is produced and transported must pass in order to be actualized, and through which only a fraction of what presents itself passes. Transport has opened its sluices wide; reading has remained a narrow door. The whole stake of reception lies in this disproportion.

Towards the Over-Plus of Reading

We therefore now have everything needed to name what the torrent does to the reader. The previous tome had set out over-throughput on the side of circulation. The present document has shown, on the side of reception, why reading cannot absorb it: its properties prevent it. It remains to describe the phenomenon itself — the submersion of the reader, the over-plus of reading that presents itself and that they cannot receive —, and to set out the divide between what, in everything that remains neglected, is a failure, and what is not. This is the object of the next document, which is also the one where the title of the tome — the trace or the datum — yields what is newest in it.

Landmarks

For the references used in this document — selectivity as the counterpart of the undrawn, the abolition of transport ageing and the maintained slowness of reading, freshness that cannot be recovered, cost as a property of every act —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; skimming taken for reading belongs to the status of illustrations. The concepts taken up refer back to their loci: the properties of the flow of observation, to Digging the Canal; the undrawn and its weighing, to The Flow and the Annotation; freshness that cannot be recovered, to The Two Times of the Datum; the short structure of the act — perceiving, distinguishing, deciphering — and the depth of deciphering, to The Act of Reading; finite attention and the bounds of the window, to The Situated Reader; the distinction between the two ageings, to The Datum Navigates Because It Is Fixed; the three axes of duplication and instantaneity, to The Properties of Movement; over-throughput relative to the window, to The Self-Knowing Torrent. The properties of reading are set out by analogy with those of transport, as the previous tome set out its own by analogy with those of the flow of observation. This document carries, for the trilogy, the properties of reading — selective, slow, costly up to fatigue — and the image of the bottleneck: the narrow point through which everything that is produced and transported must pass in order to be actualized, and through which only a fraction of what presents itself passes. Everywhere else, these established results circulate without being demonstrated again.

The Over-Plus

Submersion, the Impossible Library, and the Three Paths of the Title

Tome 3 — Document 6

“One blueberry delights; a bucket of jam that must be finished sickens.”

— Working formula

Preamble — The Torrent Seen from the Reader

The Torrent described this regime from outside, as a fact of circulation, and it had kept the reader’s share in reserve: the over-plus, at its locus. The previous document showed its consequence for the reader: its properties make reading a bottleneck, a narrow door through which only a fraction of what presents itself passes. It remains to describe the phenomenon itself, from the point of view of the one who undergoes it: the over-plus of reading. No longer over-throughput relative to the window, an established result of circulation (The Self-Knowing Torrent), but submersion as lived: the experience of a finite gaze placed before a throughput that overwhelms it.

This document does two things. It first names what this submersion is, and it corrects a false intuition about its cause: it is not scarcity that makes one suffer; it is the excess of a good thing. It then shows why the old answer to abundance — the library, classification, shelving — can no longer play its role: one does not shelve the torrent. And finally it yields what is newest in the title of the tome: not all unread data are a failure. There are three paths by which a datum remains unread, and they do not call for the same answer — one calls for none.

Part I — Submersion: the Excess of a Good Thing

The Insignificant Datum, the Aggregate That Overfeeds

The submersion of the reader has a cause, to be named exactly, for one is often mistaken about it. One readily believes that the problem of the torrent is a problem of bad data — the false, noise, waste. There is some of that, but it is not the essential, and it is not what is proper to the torrent. What is proper to the torrent is more twisted: it is the excess of a thing that, in small quantity, would be good.

An individual datum is insignificant — light, without weight, often useful, sometimes precious. One blueberry delights. The problem is not the blueberry; it is the bucket of jam you are made to finish. What was a pleasure in small quantity becomes, in mass and in movement, a burden that overfeeds more than it delights. The aggregated and precipitated datum does not suffer from being bad; it suffers from being too much. The reader is not sickened by what is inedible, but by the quantity of what, in a measured dose, would have nourished them.

This is why submersion is not a matter of quality but of proportion. The torrent does not establish the goodness of what arrives; it establishes an incompatibility of cadence — The Torrent demonstrated it, by distinguishing the five throughputs, from raw to actually read. The reader is not overwhelmed because the stream carries along the bad; they are overwhelmed because more of it arrives than any window can distinguish — and even if what arrives were good, the disproportion would remain. This disproportion, The Properties of Reading founded on the side of reception: reading is slow, selective, costly; it cannot keep up with a throughput that has become instantaneous. The over-plus of reading is the meeting of this slowness and this throughput — a narrow door before a stream that does not stop.

It Is Not a Fatality; It Is Our Resultant

The safeguard that the previous tome handed on across the threshold must be repeated here, for what follows depends on it: the torrent is our resultant, not bad weather — we set the speed (The Torrent) —, and the overwhelmed reader is not the victim of a rain fallen from the sky: they are a portion of what dug the torrent (The Self-Knowing Torrent). This changes nothing in the fact of submersion — it is quite real —, but it changes everything in the way one holds oneself in it: there is something here that one answers for, not something one undergoes without recourse. We shall come back to it when dealing with the postures of the reader; it is enough here to have refused to naturalize submersion.

Part II — The Impossible Library

The Old Answer: Shelving in Order to Be Able to Read

Faced with abundance, humans long had an answer, and it was a good one. The scribe drowned under tablets invented a management technique: the library, classification, shelving, the catalogue — devices that make it possible to find again, and therefore to read. From the Notch to the Algorithm established it: libraries are the first organized infrastructure of storage and of making available, and their history is that of the movement in which what was a privilege becomes, with time, common. Shelving was not an end; it was the means of making the reader’s share practicable — of making it possible to find what one is looking for, and therefore to read it.

The library held together three distinct roles, which Digging the Canal named while describing their contemporary fusion: the library in the proper sense — the stock preserved —, the librarian — who knows where to look, who orders and orients —, the storyteller — who renders, who gives form and meaning to what the stock holds. Stock, orientation, rendering: three separate functions, held by distinct devices and persons — and what follows will come back to them.

Why the Torrent Forbids the Library

But the datum of the torrent, through its speed and its quantity, forbids the library. One does not shelve the torrent. Classification presupposes that one has the time to classify, that the stock lets itself be stopped long enough for it to be ordered. The torrent does not stop: it flows, it multiplies, it goes stale while one would try to shelve it. When the shelving of one part is finished, another part, larger, has already poured in and gone stale. The library was an answer to a stock that waited; it is not an answer to a flow that does not wait.

This impossibility is not technical — it will not be lifted by better shelving —; it is one of substance: it comes from the nature of the torrent as flow, and not as stock, which The Torrent established. It must nevertheless be scoped with care, for it forbids less than one believes. What is impossible is exhaustive classification at finite cost — shelving the whole torrent, at the rhythm of the torrent, for a finite reader. Windows, summaries, priorities, partial archives remain possible — and this is precisely what framing and delegating organize. “One does not shelve the torrent” does not say that one does nothing; it says that no shelving will exhaust what flows, and that what one organizes instead is no longer a library: it is a posture before a flow. To believe that one will shelve the torrent is to fall back into the illusion that the torrent is a reserve that accumulates, whereas it is a multitude of drawings in movement. One does not put on a shelf what flows. There then remain to the reader only three ways, which are no longer management techniques but postures — and this is the object of the next document. Before that, a divide must be set out that submersion makes necessary, and that the title of the tome carries within it.

Part III — Trace or Datum: the Three Paths of the Title

A Fact with Three Branches

The title of the tome — the trace or the datum — states what the torrent leaves to the reader: depending on whether a reader reads or not, the matter received is actualized or neglected, the two ends of the scale set out in The Trace or the Datum. But neglect does not have a single cause. It has three paths, distinct from one another, which do not call for the same answer — and this is what makes not all unread data of equal worth.

They must be set out as a fact with three branches, without seeking to mark out a fine boundary between them. In reality, they mingle: one loses interest in what one does not know how to read, one unlearns what one has long neglected, one neglects what was going stale anyway. Drawing a sharp line between them would be a false precision. But their distinction in principle is illuminating, for it commands different answers.

First Path: Incapacity — the Reader Cannot Read

The reader cannot read: this is incapacity. No code, no receiver, no one to read in their place. The datum remains mute for them for want of means — mute for this reader here, perhaps alive elsewhere. This path calls for a means: learning, or having recourse to another reader who reads for one. It is remediable. Incapacity is caused, in the sense of The Trace and the Reader — a blind constraint, which an opened path can lift. It is the proper ground of the work of opening paths that Transport and Propagation scoped — on the side of the caused, not of the intended —, and of the delegation of reading that will be examined further on.

Second Path: Disinterest — the Reader Does Not Want to Read

The reader does not want to read: this is disinterest. They could — they have the code, the means, the access —, but they do not do it. Over-throughput turns them away; the datum passes before them; they see it without reading it — neglected, not mute: intelligible, a gaze would have sufficed. This path does not call for a means, but for a relation: the attention one grants or refuses, the time one consents to, the sorting of which The Situated Reader gathered the first probity — not to pass off as read what one has merely seen. It is here that the postures of the next document will lodge: faced with the disinterest that the torrent induces, the reader has to choose what they read, and this choice engages them.

Third Path: Rapid Lapse — the Datum Did Not Wait

The datum did not wait to be read: this is rapid lapse. Its freshness was so brief that, past the first instant, its utility is almost nil; it has lapsed before the question of reading it even arises. This path calls for no answer from the reader: it is a fact of the datum itself, not a failing of whoever did not read it. The Two Times of the Datum had prepared it by distinguishing the speeds of lapse; Navigating in Place had extended it with the stable sources and their inverse, the data whose freshness is their only value. A datum of brief freshness that no one has read has not failed: it has accomplished all it had to accomplish, and its neglect is its normal destiny — lapsed, it remains constituted and reopenable: expired for one use, it can still serve another.

What Is Newest in the Title

This last path carries what is newest in the title, and it must be stated clearly: not all unread data are a failure of reading. Some were meant to remain unread — their freshness was their only value, and their neglect is normal, not deplored. One would exhaust oneself, and be wrong, in wanting to read everything that presents itself, as if each unread datum were a loss. Many asked only to pass.

The drama of the torrent is therefore not that data remain unread — it is inevitable, and sometimes it is good: the seven outcomes of the queue, as The Torrent established, are not ontologically equivalent. The drama is the one that The Trace or the Datum announced: one no longer knows how to distinguish what deserved to be read from what could be left. Submersion does not only blur reading; it blurs the sorting prior to reading — the discernment of what deserves scarce attention and of what can pass without harm. What the torrent leaves to the reader — actualized or neglected — thus depends on three things that are not to be confused: the power to read, the will to read, and the time that the datum deserved to be granted. The first belongs to a means, the second to a relation, the third to the datum itself. Confusing them is either reproaching oneself for not having read what could pass, or letting pass what had to be read.

Part IV — What This Divide Prepares

From the Fact to the Postures

This document has established a fact — submersion — and a divide — the three paths. The fact says that the reader is overwhelmed, and that this overwhelming is our resultant, not a fatality. The divide says that not everything is to be saved in what overwhelms, and that the first task of the reader is not to read everything, but to discern what is worth reading.

From this fact and this divide arises a practical question, which occupies the rest of the tome: what can the reader who knows they are overwhelmed do? They cannot read everything — reading is a bottleneck. They must not read everything — not all unread data are a failure. What, then, can they do? The next document unfolds the three postures open to them: framing their window, delegating their reading, or accepting with lucidity what they can neither frame nor delegate. Three postures, none of which is a salvation and each of which has its cost — but which together outline what a reader can do, in response, with what overwhelms them.

Landmarks

For the references used in this document — the torrent as flow and not as stock, the “privilege → common” movement and the history of libraries, the speeds of lapse and the stable sources —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the blueberry and the jam, the scribe drowned under tablets, the impossible shelving belong to the status of illustrations. The concepts taken up refer back to their loci: the library, the first infrastructure of storage and of making available, and the “privilege → common” movement, to From the Notch to the Algorithm; the contemporary fusion of the three roles — library, librarian, storyteller —, to Digging the Canal; the speeds of lapse, to The Two Times of the Datum; the stable sources and the data whose freshness is their only value, to Navigating in Place; the regime of the torrent, the five throughputs, the seven outcomes of the queue and the resultant “the real-marked-by-us”, to The Torrent; over-throughput relative to the window and the portion of what dug, to The Self-Knowing Torrent; the scale, the mute datum and the lapsed datum, to The Trace or the Datum; the work of opening paths, to Transport and Propagation; the first probity, to The Situated Reader. This document carries, for the trilogy, the over-plus of reading — submersion named from the point of view of whoever undergoes it: the excess of a good thing, not scarcity nor the bad —; the scope of “one does not shelve the torrent” — no exhaustive classification at finite cost; windows, summaries, priorities, partial archives remain possible, which framing and delegating organize —; and the three paths of neglect — incapacity, which calls for a means; disinterest, which calls for a relation; rapid lapse, which calls for nothing — with their stake: not all unread data are a failure of reading, and the drama of the torrent is no longer knowing how to distinguish what deserved to be read from what could be left. The three paths are a fact with three branches that mingle in reality; drawing a sharp line between them would be a false precision. Everywhere else, these established results circulate without being demonstrated again.

The Three Postures

Framing, Delegating, Accepting — and the Cost of Each

Tome 3 — Document 7

“Some things are in our control and others not.”

— Epictetus, Enchiridion, I

Preamble — Three Ways, No Salvation

The previous document ended on a question: what can the reader who knows they are overwhelmed do? The old answer to abundance — shelving in order to be able to read — no longer holds: no exhaustive classification at finite cost, as The Over-Plus scoped it — but windows, summaries, priorities, partial archives remain possible, and this is what framing and delegating organize. There therefore remain to the reader three ways, which are no longer management techniques but postures — ways of standing before what overwhelms them: framing their window, delegating their reading, accepting with lucidity what they can neither frame nor delegate. This document unfolds them.

Let us say it at once: none of these three postures is a salvation, and each has its cost. They do not exclude one another — one and the same reader combines them according to the moment —, but they are not of equal worth according to situations, and distinguishing them makes it possible to know what one is choosing, and what one is paying. The first acts on the reader’s window: framing. The second entrusts reading to someone else: delegating. The third consents to what one can neither frame nor delegate: accepting. None abolishes over-throughput; each comes to terms with it in its own way.

Part I — Framing: Moving or Narrowing One’s Window

Determining the Useful Data Oneself

The first posture consists in acting on one’s own window — in framing. Since the reader cannot read everything and must not read everything, it falls to them to determine the useful data, in the broad sense, and to move or narrow their window accordingly. Framing is choosing where to direct scarce attention, and accepting to leave the rest out of frame.

This posture responds directly to over-throughput, for it acts on the term of the relation that is within the reader’s reach. Over-throughput is relative to the window — The Self-Knowing Torrent established it: there is too much in relation to what a window can receive. The reader cannot reduce the throughput — they do not command the torrent on their own —, but they can work on their window: narrow it to go deeper, move it towards what matters. Framing is doing, on the side of reception, what Digging the Canal called, on the side of production, locating the observational optimum: neither drawing everything, nor depriving oneself of the essential, but finding the setting that serves the use. It is an exercise of wisdom applied to one’s own reading.

The Cost of Framing

But framing has a cost, which must be named so as not to make a salvation of it. Framing requires an effort — that of deciding, among everything that presents itself, what deserves attention; and this effort is itself drawn from scarce attention. Framing also carries a risk: that of the out-of-frame. Every framed window leaves outside what it does not frame, and what it leaves outside may be what mattered: it is, at the edge of reception, the weighing of the undrawn that The Flow and the Annotation established, and whose counterpart The Properties of Reading set out — selectivity. And fidelity to the criteria one has given oneself is not the guarantee that one has retained the right criteria: Digging the Canal set this out for the flow of observation, and the same holds for the framing of a reading — framing too narrowly is missing the signal; framing too widely is drowning again. The right setting is given by no rule; it has to be taken up again in each situation, and this is why framing remains an exercise, never a solution acquired once and for all.

Part II — Delegating: Inheriting the Situation of Another Reader

Entrusting Reading to a Device That Fuses the Three Roles

The second posture consists in entrusting one’s reading to someone else — in delegating: the delegation of reading that The Over-Plus announced on the path of incapacity. Its most visible contemporary form entrusts it to a single device, which combines the three roles that the library kept separate — the stock, the librarian who orients, the storyteller who renders — and whose fusion Digging the Canal described: one asks it, and it gives back a meaning already deciphered.

The concept here is not the device, but the act, and it must be freed from its contemporary form to be grasped in its reach. Delegating one’s reading is inheriting its situation. This act is old: the notable who had the press summarized by a secretary was already delegating their reading, and inherited the secretary’s sorting — what the secretary judged worthy of being retained, their blind spots, their preferences. Delegating has never been reading by proxy without remainder; it has always been adopting, in place of one’s own, the situation of whoever reads for you. The large language model that fuses the three roles is the dated illustration of it, not its substance, and the tome does not have to re-conceptualize it: Human Reading, Automatic Reading fixed its status — one more situated position of handling or processing, whose situation one inherits — and announced that it would be found again here, among the postures. Delegating, to it as to anyone, is not leaving situation; it is adopting another one.

An Arbitration Whose Cost Is Known

This delegation is neither a salvation nor a fault: it is an arbitration, whose cost is known and whose measure has already been given — Human Reading, Automatic Reading recalled it: the value of a frozen model connected to fresh flows depends on the freshness of the canals upstream and on the intention of those who dig them. Transposed to any delegate: delegating is good or bad according to what the situation of the delegated reader is worth.

One therefore exchanges submersion for a dependence, and everything turns on knowing what that on which one makes oneself dependent is worth — that referent, in the sense that the last document will define: the knower to whom the non-knower turns. Delegating to a referent whose situation is rich, whose canals are fresh, whose intentions are plural, is a good arbitration: one inherits a better situation than one’s own on this point. Delegating to a device whose corpus is frozen, biased, with unknown blind spots, is a bad arbitration: one inherits a poor situation while believing one is lightening one’s load. The cost of delegation is not that one ceases to read — it is that one inherits a situation one did not choose and whose bounds one sometimes does not know. The probity of delegation consists in knowing the situation one inherits, as far as one can, rather than taking it for a reading without a situation. This is the shoal proper to this posture: taking delegated reading for an overarching reading — the dream of the universal reader, whose contradiction The Situated Reader showed —, whereas it is the reading of someone else, with the limits of someone else.

Part III — Accepting: Responsible Lucidity

Consenting to What One Can Neither Frame nor Delegate

The third posture consists in accepting — in consenting with lucidity to being swept along, for what one can neither frame entirely nor delegate without remainder. For neither framing nor delegation exhausts over-throughput: there always remains a part that spills over, which no window frames and which no referent reads for us. Before this part, there is a third bearing.

But let us name it with care, for the word “accepting” lends itself to misreading. Accepting is not resigning oneself. And it is here that the safeguard handed on across the threshold makes all the difference: the torrent is our resultant, we set the speed (The Torrent), and the overwhelmed reader is a portion of what dug it (The Self-Knowing Torrent). Accepting therefore cannot mean resigning oneself to a fatality, since there is no fatality. It means recognizing being swept along as the product of our acts, and therefore as something for which one bears responsibility.

The third posture is thus lucid and responsible, neither vain resistance nor resignation. Resisting in vain would be denying over-throughput, claiming to read everything, exhausting oneself plugging a flow that no window contains. Resigning oneself would be letting oneself be swept along while calling oneself the victim of bad weather. The right posture holds between the two: recognizing that one cannot receive everything — this is lucidity —, and recognizing that one is part of what produces the throughput — this is responsibility. One accepts not reading everything, not as a defeat, but as the condition of a finite reader who answers for what they feed.

Why This Posture Is Not a Form of Management

Unlike the first two, the third posture is not a technique. Framing is a doing; delegating is a doing; accepting is a bearing. This is why it cannot be reduced to a management of abundance: it is a way of inhabiting one’s condition of overwhelmed reader without being alienated by it. And it is through it that the tome touches what closes it — the final posture of the self-knowing reader. For accepting with lucidity is not only coming to terms with the torrent; it is knowing oneself to be a portion of what digs it, and answering for it. This responsibility goes beyond the isolated reader: it engages what the previous tome named collective wisdom.

Part IV — The Three Postures and Collective Wisdom

From Individual Arbitration to Collective Arbitration

The three postures are answers that a reader can adopt. But over-throughput, for its part, is not an individual problem: it is produced by a collective circulation, and it is not resolved, at the scale at which it arises, by the arbitrations of each alone. Self-Amplification established it — the torrent swells through the absence of wise arbitration — and the transposition holds for reading: knowledge says how to read more, sort more, delegate better; only wisdom asks the question that knowledge does not ask: should we? And this question, at the level of the torrent, arises at a scale that exceeds any single situation.

The individual’s hold, Self-Amplification measured it: elementary, real, with an effect that counts only when aggregated — a power through federation, not through decision. Transposed to reading, it is that of the reader before their flow: what they read, what they let pass, the attention they grant or refuse, the reading devices they adopt or decline — an isolated disinterest dries up nothing; a shared disinterest takes its readers away from a flow and ends up weighing on what is produced. The two levels of decision that the same document distinguished are then found again here. At the individual level, arbitration is that of the three postures: framing, delegating, accepting — each for their own window. At the collective level — the reflexive level —, deliberation bears on the composition of these arbitrations: what to make legible to all, which referents to tend, which reading devices to encourage or discourage, what a society holds to be worthy of being read — the collectivity observes in it what its own ways of reading and of not reading produce. Collective wisdom is the capacity, at this level, to arbitrate present interest against the interest to come. It does not arbitrate in a vacuum: it arbitrates the composition of individual arbitrations, which are its matter.

The three postures are therefore judged, in part, by the yardstick of this collective wisdom — the thread that Self-Amplification set out as a diagnosis and bequeathed to the present tome as a criterion. Framing, delegating, accepting are individual arbitrations; a reader who frames well, delegates with discernment and accepts with lucidity does not, on their own, defuse the torrent — they hold themselves in it better, and they add their voice to the aggregate. The torrent is not resolved from below, by the mere sum of good individual arbitrations; it is resolved — if it is resolved — from above, by a wisdom that arbitrates, at the scale of a collectivity and over time, what these arbitrations, federated, compose.

Abstention Is Not Neutral

Hence the culminating point, inherited as a criterion: abstention is not neutral. Self-Amplification fixed its condition of validity — abstention exists as a distinct position only where the choice is exposed and the capacity to take part in it acquired; elsewhere, not weighing in cannot be told apart from being out of the game — and the reason: not weighing in does not leave a balanced vacuum, but leaves the place to a slanted wisdom, that of the best-organized present interests. Applied to reading, this shifts the meaning of the three postures: framing well, delegating well, accepting well is not enough if one abstains from weighing in, where one could weigh in, on what produces the throughput. The fourth thing to do, which is not a reading posture but a civic requirement, is not to leave arbitration to the interests alone that never abstain. The tome does not settle how — that belongs to each person’s present, and the essay does not dictate it. It sets out only that abstention, where it is possible, has a price, and that this price is paid by all.

It remains to lead this responsibility to its end — the final posture of the self-knowing reader, where the probity of the reader and the ethics of reading find their formulation. This is the object of the last document, which closes the tome and, with it, the trilogy.

Landmarks

For the references used in this document — Epictetus’s epigraph and the divide between what depends on us and what does not, the observational optimum transposed to reading, old delegation and its forms —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the secretary who summarizes the press and the contemporary large language model belong to the status of dated illustrations — the substance of delegation is timeless. The concepts taken up refer back to their loci: the observational optimum and relative fidelity to criteria, to Digging the Canal; the weighing of the undrawn, to The Flow and the Annotation; the selectivity of reading, to The Properties of Reading; the scope of impossible classification — no exhaustive classification at finite cost — and the delegation of reading announced on the path of incapacity, to The Over-Plus; the fusion of the three roles — library, librarian, storyteller —, to Digging the Canal; the status of the machine that reads — a backup that diverges, an instrument in a chain, one more situated position of handling or processing, whose situation one inherits — and the measure of delegation by the freshness of the canals and the intention of those who dig them, to Human Reading, Automatic Reading, which carries the credits of the first tome; the dream of the universal reader, to The Situated Reader; the resultant and “we set the speed”, to The Torrent; over-throughput relative to the window and the portion of what dug, to The Self-Knowing Torrent; self-amplification, collective wisdom, the aggregated hold, the two levels of decision, non-neutral abstention and its condition of validity — exposed choice, acquired capacity —, to Self-Amplification. The referent, used here to gauge delegation, will be defined in the last document.

This document carries, for the trilogy, the three postures — framing: moving or narrowing one’s window, at the cost of effort and of the out-of-frame; delegating: inheriting the situation of another reader, at the cost of a dependence whose probity is knowing its bounds; accepting: the lucid and responsible bearing, neither vain resistance nor resignation, which the safeguard of the resultant keeps from fatalism — and the systemic side of ethics: the thread of collective wisdom taken up from diagnosis to criterion, the two levels transposed to reading — the arbitration of postures below, deliberation on what a society makes legible and tends above —, and the civic requirement not to leave arbitration to the interests alone that never abstain. Everywhere else, these established results circulate without being demonstrated again.

The Self-Knowing Reader

Probity, Minimal Ethics, and the End of the Trilogy

Tome 3 — Document 8

“The world ends up in no book. But it has produced a being who, reading the traces the world leaves of itself, writes books — and this is one of them.”

— The Trace and the Reader, first tome

Warning

This document closes the third tome, and with it the whole trilogy. It occupies, in the tome of reception, the place that “the self-knowing torrent” occupied in the tome of movement: that of the final posture. The echo between the two titles is intended. There, it was the torrent that knew itself — a regime grasped again from the situation of whoever must receive something from it. Here, it is the reader who knows themselves — a finite gaze that recognizes itself as overwhelmed, and that, knowing itself to be so, stands otherwise. The previous tome handed over by saying: what this changes in what can be required of the reader, it is for the next tome to say. This document says it, and closes the arch.

Preamble — What a Reader Can Know of Themselves

The previous document unfolded the three postures — framing, delegating, accepting — and showed that none is a salvation, that each has its cost, and that none is enough if one abstains from weighing in on what produces the throughput. It remains to lead this responsibility to its end. What can a reader know of themselves, in the torrent? And what falls to them, once they know it?

This document answers by taking on what the first two tomes forbade themselves: an ethics. Not because the matter has changed its nature — it remains descriptive in substance —, but because the three postures presuppose a criterion for judging them, and a criterion of this kind is ethical. The tome therefore takes on an ethics of reading; but it bounds it with care, for a badly bounded ethics would go stale along with the customs that had inspired it. This ethics will be minimal, timeless and systemic. Setting it out is saying what the trilogy has, in the last resort, to say to the reader it addresses — and it is, for the trilogy, its way of ending.

Part I — Why an Ethics, and of What Kind

From the Descriptive to the Propositional

The first tome kept a strictly descriptive discipline: it said what is true of the datum, never what must be done. What would belong to the prescriptive — the charter of a craft, the probity of an artisan — belongs to other works, distinct from the essay, which it calls for without containing them. The second tome kept the same discipline, while taking on only a more marked share of openness. The question, for the third, was: which way does it lean?

The answer is that it takes on an ethics, and one must say why without apologizing for it. First because its propositionality requires it: as soon as one describes what the reader can do with what overwhelms them — framing, delegating, accepting —, one presupposes a criterion for distinguishing good arbitration from bad, and this criterion is not descriptive. Next because the reflections gathered throughout the trilogy — collective wisdom and abstention that is not neutral (Self-Amplification), the responsibility of the resultant (The Torrent), the first probity of the reader (The Situated Reader) — are already ethical statements, not descriptive ones. The tome only recognizes what it was carrying. It becomes propositional because reception, unlike making and movement, cannot be thought without the question arising of what one must do with it.

Minimal, Timeless, Systemic

But the ethics that the tome takes on is carefully bounded, and its three bounds are what keep it from going stale.

It is minimal. It comes down to a single duty: not to pass off as read what one has merely seen — the first probity that The Situated Reader set out, raised to the rank of single duty. Everything else follows from it, and nothing is added to it. This duty does not require reading everything — The Properties of Reading and The Over-Plus established it: reading is a bottleneck, and not all unread data are a failure. It requires only not confusing the datum seen and the datum read, not taking skimming for deciphering, not taking for granted what one has merely glimpsed. It is little, and it is everything.

It is timeless. This duty held for the scribe drowned under tablets as for the contemporary drowned under notifications; it depends on no technique. The form of what submerges changes — tablets, registers, screens —, but the duty remains the same: not to pass off as read what one has merely seen. This is what distinguishes it from a judgement of form, which is dated: it judges no technique; it states what the act of reading engages, in all its forms and at all times.

It is systemic — extending the side that The Three Postures established along the thread of collective wisdom, and this is the point that the next part has the charge of establishing. It does not rest on the individual alone, summoned to know how to read everything. For requiring each person to know how to read everything would be repeating, on the ground of reading, an error that Reading, Writing, Understanding criticized.

Part II — The Probity of the Reader: Reading or Getting Help to Read

Why It Cannot Be “Knowing How to Read Everything”

The probity of the reader is the counterpart of the probity of drawing that Digging the Canal regulated in the optimum, and of transport probity — acknowledged condensation — that The Properties of Movement named. But it cannot be stated “each person must know how to read everything”. That would be repeating, on the ground of reading, the rut that Reading, Writing, Understanding criticized in connection with literacy — managerial co-optation, which reduces the stake to a problem of individual training — coupled with the dream of the universal reader, who would know how to read everything from everywhere, whose contradiction The Situated Reader showed. Each individual is trained, and it is not enough, because the torrent exceeds any individual training. No reader, however well trained, will know how to read everything the torrent sets down before them; and the foundation of this is established: every deciphering has a depth relative to the reader’s corpus — The Act of Reading established it —, and no corpus is universal — The Situated Reader demonstrated it.

The probity of the reader is therefore stated otherwise:

The reader must read honestly, or honestly call on a referent capable of reading for them. Probity includes the discernment to know when one does not know how to read, and whom to turn to. The duty to read includes the duty to have recourse.

This referent the trilogy already knows — The Act of Reading set out its foundation with the call for delegated deciphering, Human Reading, Automatic Reading named it, The Three Postures used it to gauge delegation — and it must now be defined broadly: a referent, before a datum, is anyone who knows more than the reader who stumbles on the subject to which it refers — the knower to whom the non-knower turns so that the deciphering may be continued. The referent is neither a rare nor an instituted function: it is a relative position, which each person occupies for someone and about something. Probity is therefore not knowing how to read everything; it is reading what one knows how to read, and knowing whom to turn to for the rest — honestly, that is, without passing off someone else’s deciphering as one’s own, nor having recourse out of laziness where one could read oneself.

Among referents, there is one whose place the tome must name — and recognize that it is, for the time being, vacant. The ordinary referent knows a subject of the world: the pharmacist knows mushrooms, the doctor the tracing. But who knows the datum itself — its making, its circulation, its reading, what the whole trilogy has tried to think? Who is a referent, not on such an object that the datum describes, but on the datum as datum: on the way of drawing it with probity, of judging its freshness, of framing a window, of recognizing a datum that neglect lies in wait for? This referent the trilogy calls the data artisan — Transport and Propagation announced the way to it, at the end of the work of opening paths: not a keeper who dispossesses, but a referent who makes one more autonomous. But one must be exact about its status: this artisan does not yet exist as a constituted craft. It is not an established result of the trilogy; it is a proposal of it. The tome does not possess it; it deduces its necessity, and calls for creating it. What its charter, its training, its own probity would be — all this remains to be produced, in a work distinct from the essay, which the essay makes thinkable without accomplishing it.

The Non-Knower / Knower Relation

This probity brings into play the relation that The Situated Reader set out and bequeathed to this document: the non-knower turns to the knower — a difference of depth of deciphering, not a hierarchy of value, reciprocal in its principle. What the referent adds to it is this: its recognition institutes the relation, by designating whom to turn to — recourse ceases to be left to the chance of encounters. The data artisan, when it exists, will be the referent so recognized for the proper domain of the datum; but the relation, for its part, does not wait for this artisan in order to function — it already operates, wherever a recognized knower can read what another does not know how to read. Delegated deciphering therefore calls for no separate concept: it is entirely covered by this probity — reading, or calling on a referent. And one finds here again, on the side of reading, the involuntary impediment that Transport and Propagation divided on the side of transport: the referent is the one who opens the path — not by forcing secrets, which belong to a will, but by remedying the incapacity to read, where a corpus is lacking.

The Stake Is Systemic

It is in this way that the ethics of reading is systemic, and not individual. A data-literate society is not a society where everyone knows how to read everything — that is impossible, and wanting it would be the managerial rut. It is a society that tends the referents to whom non-specialist readers can turn. The duty to read includes the duty to have recourse, and recourse presupposes a fabric of referents that the collectivity must tend.

This is consistent with the systemic side that The Three Postures established along the thread of collective wisdom. The involvement of each person does not mean that each person does everything, but that each person should either read, or know which referent to turn to — and that society should tend these referents. Tending a fabric of referents is therefore not an extra: it is the very condition of a collective reading with probity. A society that let its referents wither — that left them to the chance of encounters, without recognition or tending — would leave its non-specialist readers alone before the torrent, summoned to a universal competence that no one can have. The ethics of reading thus moves the duty from the individual towards the system, without relieving the individual of it: to each, reading or having recourse; to all, tending what one has recourse to.

And the tending of the fabric of referents is precisely one of the objects of the collective and reflexive level that The Three Postures transposed to reading: “which referents to tend” already appeared there among what a society deliberates on. Tending the referents is therefore not a charity towards non-specialists; it is the act by which a society keeps for itself the collective capacity to read, and therefore to govern itself, instead of abandoning it to those who would know how to read for it. One meets here, from the other end, the risk that Human Reading, Automatic Reading named: the capacity to read, if it is not tended as a commons, becomes concentrated — and a society that no longer reads what determines it relies on others, or on machines, to do so.

It is also at this level that the question of the data artisan left open above arises. If the ordinary referent — the one who knows a subject of the world — already exists and calls only for being recognized and tended, the referent of the proper domain of the datum, for its part, remains to be instituted. A society that takes seriously over-throughput and the risk it poses to its capacity for reading would have, among the things to tend, this one to create: a data craft, with its training and its probity, to which the ordinary reader could turn when what overwhelms them is not a particular subject, but the datum itself. The tome does not describe this craft — that would be prescribing a form, and form belongs to another work. It sets out only that its place is vacant, and that leaving it vacant has a cost, paid in collective capacity to read.

Part III — The Self-Knowing Reader

The Final Posture

The final posture can now be named, the one that closes the tome and the trilogy: the self-knowing reader. Knowing oneself, for a reader, is knowing three things that the three tomes have established.

It is knowing that one is situated — that every reading is done from a bounded window, that one deciphers only to the depth of one’s corpus, and that the first probity is to know the bound of one’s own window. It is knowing that one is overwhelmed — that the torrent sets down more than any window can receive, that reading is a bottleneck, and that one must therefore frame, delegate or accept, without any of these postures being a salvation. And it is knowing that one is responsible — that the torrent is not bad weather but our resultant, that abstention is not neutral, and that one answers, as a portion of what digs the current, for what one feeds and for what one lets pass.

The self-knowing reader is therefore neither the one who claims to read everything — they know it is impossible —, nor the one who lets themselves be swept along while calling themselves a victim — they know that we set the speed. It is the one who holds together their finitude and their responsibility: they read what they can, have recourse for what they cannot, accept letting pass what could rightly be let pass, and do not abstain from weighing in on what produces the throughput. It is not an instruction; it is a lucidity, and the responsibility that comes with it.

The Loop Closes

One piece of knowledge is still lacking to this reader, and it is the last: what their reading, when it passes into the world, does to the world. The scale announced it at its last degree — the operative datum, of which The Trace or the Datum said that the loop, at the close of the tome, would close there. Here it is closed. A reading feeds an inference, a coordination or an action; the action modifies the real; the modified real leaves new traces, which new drawings will constitute as data — and the last gesture joins the first. Digging the Canal set out this loop without demonstrating it — observing in order to decide, deciding in order to act, an action that modifies what is observed —; Self-Amplification demonstrated its upstream side; its closing is demonstrated here: the reader is not the end point of the chain; they are a link in it — the link where it turns back on itself.

One must still measure what this crossing engages, for one does not pass from reading to action in a single movement. Three relations to the real follow one another in it, and their bridge must be held: observation, which ascertains what has left a trace; inference, which extends to the unobserved what the observed makes it possible to conclude; intervention, which modifies what will no longer be observed as it was. Each crossing adds hypotheses, for which someone must answer. A prediction is not a fresh datum about the future: it is a derived datum, whose guarantee comes from a model, not from the real. And the time that separates calculation from application can make harmful a decision that was correct — the loop does not forgive the reader for forgetting that it turns while they conclude.

When the loop is closed, answering for one’s reading alone is no longer enough to describe what is at stake. Responsibility runs along the chain — from the regulated apparatus to the act that constitutes, from transformation to reading, from reading to decision — and lodges in no single link: each link answers for its gesture and transmits to the next what it needs to answer. Provenance, constitutive since The Properties of Movement, is the documentary form of this “transmitting what is needed to answer”. This is the responsibility of the chain.

And the probity of the artisan, which the trilogy carried in dispersed form, can then be decomposed. Seven places stand out, where one can answer — or fail — separately: constitution — the probity of drawing, which Digging the Canal regulated in the optimum; preservation — what persisting requires, which The Two Times of the Datum established; selection — the honesty of sorting, which The Properties of Reading announced for this document; transformation — the acknowledged condensation of The Properties of Movement; presentation — the rule that turns the raw into the presented, a political decision that The Torrent named; inference — the bridge that has just been set out; use — the imputability of arbitration, which Human Reading, Automatic Reading fixed. A chain can have probity in constitution and be at fault in presentation; this is decomposed probity, and the evidentiary regime — chain of custody, corroboration — is its practical horizon. The probity of the reader, established above, finds its exact place in it: the single duty is its core, in the places of selection and use — one place among seven, neither more nor less, but the one where this tome will have done its share.

What the trilogy called a processual ontology thus ends as an ontology of the responsibility of the chain — not of the mere materiality of its links. The self-knowing reader is the link where this chain knows itself.

What the Trilogy Will Have Said

The trilogy ends here; let us look, to finish, one last time at what it will have said. The first tome showed that the datum is not an object that one picks, but the individuated result of an operation of constitution, which inscription makes persistent without constitution and inscription being confused; drawing on a trace that the real leaves of itself is only its observational route, one of the three along with the act that institutes and the operation that derives. On this route, the datum is born at the meeting of a legible real and of someone who draws, whom the real has produced; reading, afterwards, actualizes it without founding it again. The second tome showed that the datum can circulate according to a frequent and structuring regime, without this circulation being its universal destination, that it navigates because it is fixed, and that its circulation, by racing, produces a torrent that sets down more than any window can receive. The third showed that what arrives in this way is not yet read, that without reading nothing is actualized, and that the reader, situated and overwhelmed, has no other way than to read with probity, to have recourse with discernment, and to answer for what they feed.

Three acts, a single movement. Inscription makes the real pass to the status of datum; exchange brings the datum within reach; reading makes a living meaning of it — or does not, and the datum remains neglected, the trace naming no more, at the end of this slope, than a horizon. The datum can circulate between these three acts, in a frequent and structuring regime, not as a universal destination; the situated reader is what accomplishes all three, for the one who inscribes first reads a trace, the one who transmits has first read, and the one who reads will be able, in their turn, to inscribe and transmit. The trilogy has, in substance, described only one thing: the wave that the flow of the real has raised, and that, knowing itself made of the water of the ocean, reads the ocean, transmits it, and answers for it.

The Last Sentence of the Trilogy

The first tome ended on an image: a wave that would like to possess the ocean, and that is nonetheless only water of the ocean momentarily raised. The second ended on a turning around: the instrument humans had given themselves to hold the world threatened, through sheer abundance, to let it slip away again. The third ends on the self-knowing reader. For what decides, in the last resort, that the torrent does not let everything slip away is neither circulation, which has done its share, nor mass, which knows nothing of itself: it is that a situated reader, somewhere, actualizes a part of what passes — and knows, of the other part, what they have let pass, and why. The wave will never possess the ocean. But it can, for the time it stands raised, read a fragment of it with probity, transmit it to another wave, and answer for what it will have known, or neglected, to read. It is little, at the scale of the ocean. It is everything, at the scale of a wave that knows itself.

“The world is made to end up in a book.” No: the world ends up in no book, and no torrent becomes knowledge of its own accord. But it has produced a being who, reading the traces the world leaves of itself, makes them into actualized data — or neglected ones, for want of being read. Everything, from now on, depends on their knowing which ones they make.

Landmarks

For the references used in this document — the critique of individualist literacy, the evidentiary regime — chain of custody, corroboration — as the practical horizon of probity, the taking up again of the ending of the first tome —, see the Annotated Bibliography of Tome 3. The methodological status is set out in the Preface to the Edition; the scribe and the contemporary belong to the status of illustrations. The concepts taken up refer back to their loci: managerial co-optation and wisdom as judicious application, to Reading, Writing, Understanding; the probity of drawing and the optimum, as well as the loop set out without demonstration, to Digging the Canal; what persisting requires, to The Two Times of the Datum; the ending of the wave, to The Trace and the Reader, and its unfolding, to The Wave and the Ocean; acknowledged condensation and constitutive provenance, to The Properties of Movement; the involuntary impediment, the opening of paths and the announcement of a data craft, to Transport and Propagation; the decision of what is presented, to The Torrent; the turning around of the instrument, to The Self-Knowing Torrent; the upstream side of the loop, collective wisdom and non-neutral abstention, to Self-Amplification; the scale and the announcement of the closing at the degree of the operative, to The Trace or the Datum; the relative depth of deciphering and the call for delegated deciphering, to The Act of Reading; the first probity and the relation of the non-knower to the knower, to The Situated Reader; the imputability of arbitration, concentration and the reversal of the referent, to Human Reading, Automatic Reading; the bottleneck and the honesty of sorting, to The Properties of Reading; the drama of the torrent, to The Over-Plus; the three postures and the systemic side, to The Three Postures. The layout of the looped diagram and the formalization live in the Architecture Volume.

This document carries, for the trilogy, the closing of the loop — the closing demonstrated at the reader’s link, the bridge of the three relations to the real — observation, inference, intervention: a prediction is not a fresh datum about the future —, the responsibility of the chain — which lodges in no single link, and whose documentary form is provenance — and decomposed probity in its seven places — constitution, preservation, selection, transformation, presentation, inference, use —, a credited bringing-back of the probities that the trilogy carried in dispersed form; the ethics of reading — minimal: the single duty, the first probity raised; timeless: independent of any technique; systemic: the fabric of referents that a collectivity tends —, the referent defined — a relative position, not an instituted function, whose recognition institutes recourse — fulfilling the announcement of Human Reading, Automatic Reading, and the data artisan — the referent of the proper domain of the datum, a proposal whose necessity the essay deduces without describing its form; finally the self-knowing reader — situated, overwhelmed, responsible —, the final posture by which the trilogy ends, not on an instruction, but on a lucidity, and the responsibility that comes with it. Everywhere else, these established results circulate without being demonstrated again.

Thus ends the third tome, The Trace or the Datum — The Situated Reader Tried by the Flow, and with it the trilogy Tried by the Flow. The making, the circulation and the reception of the datum have been led in it as the three acts of a single movement: inscription that draws, exchange that transports, reading that actualizes a meaning — or, failing that, neglect, up to the horizon where the trace would regain its rights. What the self-knowing reader will do, in their present, with what the trilogy will have established for the long term no longer belongs to the essay: it belongs to its reader.

Annotated Bibliography of Tome 3

Apparatus of References of Tome 3

This bibliography brings together the references proper to the third tome of the essay. It extends the Annotated Bibliographies of Tomes 1 and 2, to which it refers for the whole common foundation (Bergson, Whitehead, Peirce, Uexküll, Ackoff, processual ontology, the history of the datum, literacy, the movement of the datum). It does not repeat their content: it adds the references that the reception of the datum — the act of reading, the situated reader, submersion, the ethics of reading — calls for in its own right. Like the previous ones, it is organized to remain useful over time.

What Will Not Change

At the Philosophical Foundations

Peirce, Charles Sanders — Collected Papers, on the semiotics of the index and of the conventional sign. Already used in the first tome to distinguish the trace (a caused sign, physically linked to what it marks) from the arbitrary sign, he is used here to found the distinction of the present tome between interpreting (the genus: understanding that a mark refers to something other than itself, true of every mark) and deciphering (the restricted species: interpreting by means of a conventional code). The deciphering of the tome is interpretation applied to the conventional sign.

Uexküll, Jakob von — Umwelt und Innenwelt der Tiere, 1909; Streifzüge durch die Umwelten von Tieren und Menschen, 1934 (French translation Milieu animal et milieu humain). The notion of Umwelt — the world of its own that an organism constitutes to the measure of its senses — is the direct underpinning of the situated reader and of their window. The tome retains its principle from it: every reading is situated, and therefore partial, and the window is not a prison but the condition of any view.

Bergson, Henri — Essai sur les données immédiates de la conscience, 1889; L’Évolution créatrice, 1907. For duration as continuous flowing, the underpinning of the temporality of reading: the act of reading takes time, and this time has not passed below any threshold, unlike transport. See the bibliographies of the previous tomes for the detail.

On Attention and Reading

Note on the economy of attention — The idea that attention is a finite resource, and that an abundance of information produces a scarcity of attention, is commonly associated with Herbert Simon: Designing Organizations for an Information-Rich World, in Computers, Communication, and the Public Interest, Johns Hopkins Press, 1971. The essay retains from it the principle of substance — attention is finite and not extensible at will, which makes reading the bottleneck of the torrent — without calling on its economic formalism, and, in accordance with the Preface to the Edition, without making a dated diagnosis of it.

Note on the difficulty of reading and on attention-capturing devices — Contemporary phenomena (observed reading difficulties, hooking devices, gamification, accessibility as a new regime of availability) belong to form, which is dated, and will find their place in the work that puts the tome into practice, not in the tome itself. The tome states what will long remain true of the act of reading; it does not judge the forms of the present.

Ethics and Politics of Reading

Note on managerial co-optation and the universal reader — The critique according to which literacy, as disseminated by commercial publishers and consultancy firms, depoliticizes the stake by reducing it to a problem of individual training is established in Reading, Writing, Understanding the World through Data; the critique of the dream of a universal reader has its locus in The Situated Reader. These two critiques found, in the present tome, the refusal of a probity of the reader understood as “knowing how to read everything”, and the systemic character of the ethics of reading.

Ostrom, Elinor — Governing the Commons, 1990. Already used in the second tome for the political dimension of transport, she sheds light here on the collective dimension of the ethics of reading: the tending of a fabric of referents is a matter of collective governance, which cannot be reduced to the sum of individual trainings.

The Evidentiary Regime

Anderson, Terence, Schum, David A. & Twining, William — Analysis of Evidence, 2nd edn, Cambridge University Press, 2005, for the chain of custody and corroboration as regulated requirements. Langlois, Charles-Victor & Seignobos, Charles — Introduction aux études historiques, Hachette, 1898; Bloch, Marc — Apologie pour l’histoire ou Métier d’historien, Armand Colin, 1949, for the criticism of sources. Two independent traditions — the law of evidence, historical method — established separately that an assertion is judged on the chain that carries it. The tome makes this the practical horizon of decomposed probity, without importing its procedures.

Sources of the Epigraphs

Johnson, Samuel — a formula attributed to him; the inquiry of Quote Investigator (8 November 2014, online) locates the earliest attribution it found in Seward, William, Biographiana, 1799, p. 260: an open attribution, for the epigraph of the first document, The Trace or the Datum.

The absent voice — the epigraph of The Act of Reading is given without an author, “after an intuition attributed to many readers”: an open attribution, acknowledged as such.

Uexküll, Jakob von — for the epigraph of The Situated Reader (entry above).

Skinner, B. F. — a quip attributed to Skinner, as the epigraph of Human Reading, Automatic Reading; an open attribution, it too.

Epictetus — Enchiridion, I, for the epigraph of The Three Postures: the divide between what depends on us and what does not.

The self-quotation — the epigraph of The Self-Knowing Reader is borrowed from The Trace and the Reader, first tome: an internal reference, not an external source.

The three “Working formulas” have no source: they belong to the edition.

Neighbours of the Thesis

The Architecture Volume carries the map of neighbours (§7); two families touch on the proper subject of this tome: their references are below; those of the other families are in the Annotated Bibliography of Tome 1 and the Annotated Bibliography of Tome 2.

Relational datum — Leonelli, Sabina, Data-Centric Biology: A Philosophical Study, University of Chicago Press, 2016. Overlap: dependence on context, on use and on practice. Fruitful tension: an object can acquire the status of datum through its later evidential role, which contests a status entirely fixed at inscription. The edition answers it by distinguishing constitution and actualization, and through the intention of use that recruits.

Datum as process — Jones, Matthew, What we talk about when we talk about (big) data, The Journal of Strategic Information Systems, vol. 28, no. 1, 2019, pp. 3-16. Overlap: the rejection of the datum as an objective and stable resource, held to be partial and contingent, produced by situated practices of conceptualization, recording and use. Difference: this work belongs to organization theory and does not establish the overall architecture — but it forbids asserting that no existing framework thinks the datum in a processual regime.

Internal Cross-References of the Essay

The third tome relies continually on the first two. The following concepts are established there and are not redefined here:

• object ontology and its critique, the datum as a constituted entity of which drawing is the observational route, the backup metaphor and the language model as a backup that diverges, the three ages of the datum — The Flow and the Annotation (first tome);
• the flow of observation, the observational optimum, the extended chain (flow of the real → flow of observation → elementary data → consolidated data → information → knowledge → wisdom), the machine as an instrument in a chain where humans keep intention, the LLM as the fusion of the library, the librarian and the storyteller — Digging the Canal (first tome);
• the two times of the datum, the act of inscription and fixation, the two deaths (lapse and disappearance), the paradox of duplication and the uniqueness of the act, freshness that cannot be recovered, the speeds of lapse — The Two Times of the Datum (first tome);
• the legibility of the real, the caused trace opposed to the intended datum, the threshold that nothing crosses on its own, the five faculties of the reader (perceive, distinguish, interpret, aim, inscribe), the situated reader, the window as a band of the spectrum of times, humans as a portion of the flow, the wave and the ocean — The Trace and the Reader (first tome);
• the torrent and the rule that points out the shoals without describing them, responsible lucidity — The Wave and the Ocean (first tome);
• the “privilege → common” movement and the history of libraries, the movement “entrusting to the tool what is repetitive” — From the Notch to the Algorithm (first tome);
• the DIKW pyramid and wisdom as “judicious” application, managerial co-optation — Reading, Writing, Understanding the World through Data (first tome);
• the act of exchange and the transport medium, double navigation (spatial and temporal) and the image of the spiral, transport as a relation between two movements and its criterion of accomplishment (the reader’s share made practicable), the inaccessible datum as an unaccomplished transport, propagation as the aggregate of transports, the properties of movement (fidelity, temporality, multiplication) and transport probity, the abolition of transport ageing by instantaneity, stability as a ratio of speeds, the torrent as the regime of the real marked by us, self-amplification and its two engines (the loop of automation and the coexistence of regimes), collective wisdom, the aggregated individual hold (power through federation), the two levels of decision (binary individual and reflexive collective, the latter metacognitive in nature) and non-neutral abstention with its condition of validity (exposed choice and acquired capacity), over-throughput relative to the window and its two sources (speed and the coexistence of regimes), the voluntary (the secret) and involuntary (incapacity) impediment of transport and the work of opening paths, irrigation — From Spring to Torrent (second tome), each concept referring to the document indicated in its section “Concepts Introduced by the Present Tome”, Annotated Bibliography of Tome 2;
• the substance/form distinction, the status of illustrations and guiding images, the status of the “flow of the real”, the role of AI — Preface to the Edition.

Concepts Introduced by the Present Tome

The reference definitions are in the Canonical Lexicon for the concepts that have an entry there or a correspondence of heading; the groups, images, sub-concepts and articulations named here for the record have no entry of their own there and are demonstrated in the document indicated after each element, the precisions they carry belonging to the same document. For the record, the concepts proper to the third tome are: the scale of the datum — constituted, accessible, intelligible, actualized, operative —, of which this tome is the locus for the trilogy, in The Trace or the Datum; the act of reading (with the structure perceive → distinguish → decipher, shorter than inscription, and bearing only on data, never on the raw trace), in The Act of Reading; deciphering as the restricted species of interpreting, at a depth relative to the corpus, in The Act of Reading; the exit threshold (inscribing, calculating, saying, acting, or nothing — later acts distinct from reading), in The Act of Reading; the two inheritances of the reader (the faculties, inherited from the flow; the code, inherited from transmission), in The Act of Reading; the distinction between the mute datum (relative, reversible) and the dead datum (universal second death, limit-ideal), in The Trace or the Datum; the return cycle of the first tome, reserved for the limit case, in The Trace or the Datum; the bounds of the window (perceptual, attentional, interpretive), in The Situated Reader; the re-circulating datum (a datum really produced by automated inscription, but which no human reads, and which turns between machines without crossing the threshold of reading) and the risk of far-reaching delegation (humans moving away from the datum, the atrophy of their capacity to read, and the two slopes that follow from it — the concentration of the capacity to read in a few, and the reversal of the referent towards the machine, up to the risk that a society can no longer read what governs it), in Human Reading, Automatic Reading; the properties of reading (selectivity, temporality, cost and fatigue) and reading as the bottleneck of the torrent, in The Properties of Reading; the over-plus of reading and the impossible library, in The Over-Plus; the three paths of the title (incapacity, disinterest, rapid lapse), the last of which establishes that not all unread data are a failure, in The Over-Plus; the three postures (framing, delegating, accepting) and delegation as the inheritance of the situation of another reader, in The Three Postures; the probity of the reader (reading or calling on a referent), in The Self-Knowing Reader, and the non-knower / knower relation, in The Situated Reader; the referent as a general category (anyone who knows more, on the subject to which a datum refers, than the reader who stumbles — a relative position and not an instituted function) and the data artisan as a particular case proposed and not established (the referent of the proper domain of the datum, a craft that does not yet exist and whose creation the tome calls for without describing its form), in The Self-Knowing Reader; the minimal, timeless and systemic ethics of reading, in The Self-Knowing Reader; finally, at the close, the responsibility of the chain — which lodges in no single link, and whose documentary form is provenance — and decomposed probity in its seven places: constitution, preservation, selection, transformation, presentation, inference, use, in The Self-Knowing Reader.

The tome also takes up from the previous tome, as a criterion and not as a new concept, collective wisdom and its established results — the aggregated individual hold, the two levels of decision including the reflexive collective level, and non-neutral abstention under its condition of validity (exposed choice and acquired capacity) —, applying them to the three postures of the reader and to the collective tending of the fabric of referents. What it adds in its own right is not a concept but an articulation: the tending of referents as an object of the reflexive level, tied to the risk of concentration of the capacity to read established in Human Reading, Automatic Reading.

Over-throughput relative to the window, established as a threshold by the previous tome, is taken up again here and turned around: the reception face of a concept whose production face the tome of movement had set out. It is the weld between the two tomes.

The data craft — the craft of the referent whose proper domain would be the datum itself, to which Transport and Propagation opens the way — does not yet exist as a constituted craft. The present tome does not present it as an established result, but as a proposal: it establishes its necessity from the probity of the reader and the non-knower / knower relation, and calls for its creation. Its elaboration — what its charter, its training, its own probity would be — belongs to a work distinct from the essay, which the essay makes thinkable without accomplishing it, and cannot be taken as established on the sole strength of these pages.

Translation choices

These notes say how to read the English words that render the terms of the corpus; they are fixed for the whole English translation by the Translation Charter of the Corpus. The notes of the first two tomes remain valid here; those below repeat the main ones and add the terms of reception.

• datum / data. The corpus treats the datum as an individual: constituted by an act, dated, carried by media. The translation therefore says datum in the singular — the Latin for “what is given” — and data as its plural, construed as a plural.
• drawing / to draw (prélèvement, prélever): the gesture of drawing from the flow, as one draws water. Draw is kept for this gesture; derive renders the other senses of tirer. To measure, entry and survey render relever and relevé; reading is kept for the act of reading.
• the scale of the datum (l’échelle de la donnée): five cumulative states — constituted, accessible, intelligible, actualized, operative —, each relative to a reader and to an instant. A rung renders barreau.
• actualized (actualisée): the act by which a reading produces a meaning, never the updating of a file.
• neglect; neglected datum (délaissement; donnée délaissée): the constituted datum, often intelligible, that no one reads for want of attention: neglected, not mute. Mute (muette) says that a datum is unintelligible for one reader and alive for another; dead (morte), that its interpretation is lost for all — the limit case, where the datum falls back into trace.
• lapse; lapsed; stale (péremption; péremptée; périmée): stale says the gap from the real, which grows continuously; lapsed says the state that a work gives to this gap when it exceeds what use tolerates. The use-by limit (durée limite d’utilisation) is the threshold of that state. The assessment renders le bilan; recounting (récit, technical sense) is the new inscription by which a reader makes a lapsed datum rereadable.
• perceive, distinguish, decipher (percevoir, distinguer, déchiffrer): the three moments of the act of reading. Interpret is the genus; decipher, the species that applies to conventional inscription. The non-knower and the knower (non-connaisseur, connaisseur) differ in depth of deciphering, not in worth; connoisseur is avoided.
• exit threshold (seuil de sortie): what follows reading without being part of it — inscribing, calculating, saying, acting, or nothing.
• window; bounds (fenêtre; bornes): the reader’s window is closed by three bounds — perceptual, attentional, interpretive. Reading is the bottleneck (goulot) of the torrent.
• situated position of handling or processing (position située de reprise ou de traitement): a functional and bounded place in the chain, from which an arrived datum is handled, processed or relayed; a human or a machine may occupy it, and a machine is an actual reader (lecteur effectif) only if a meaning is actualized for it. The re-circulating datum turns between machines without being read. Blind ingestion and delegated ingestion keep apart absorption without criteria and a mandated drawing.
• over-plus (sur-plus): the submersion of the reader, formed like over-throughput (sur-débit); never surplus. “One does not shelve the torrent” renders On ne range pas le torrent.
• the three paths (les trois chemins): incapacity, disinterest, rapid lapse.
• the three postures (les trois postures): framing, delegating, accepting. What framing leaves outside is the out-of-frame (hors-champ); accepting is a bearing (une tenue), a way of holding oneself, not a technique; being swept along renders l’emportement.
• referent; fabric of referents (référent; tissu de référents): anyone who knows more than the reader who stumbles, on the subject to which a datum refers. The data artisan (artisan de la donnée) is the referent whose domain would be the datum itself; the data craft (artisanat de la donnée) is that craft, still to be created.
• decomposed probity; responsibility of the chain (probité décomposée; responsabilité de la chaîne); the evidentiary regime (régime de preuve), with chain of custody and corroboration.
• to ascertain; finding (constater; constat): establishing a fact. Observe and observation are kept for observation in the sense of the corpus (the flow of observation; observation, inference, intervention).
• gaze renders regard when it is the eye that follows; record renders trace in its ordinary sense (“the records of the work”, “a record of the choice”). The trace of the corpus is always the mark that the flow of the real leaves on matter.
• Tome names each of the three tomes; volume, any other book of the corpus. “The tome of movement” and “the tome of the reader” are Tomes 2 and 3. Threshold (Seuil) is the opening text of a tome; Landmarks (Repères) is the closing section of each document; locus (plural loci) renders foyer.
• Quotations. The sentence attributed to Samuel Johnson and the quip attributed to B. F. Skinner are given in the English wording that circulates under their names; Epictetus in a received English translation of the Enchiridion. The epigraph after Uexküll, the absent voice and the three working formulas are translated from the French. The epigraph of The Properties of Reading varies a well-known sentence of Saint-Exupéry (“One sees clearly only with the heart”). The quotations of the first tome, and Mallarmé’s sentence, are given as in the English translation of the first tome.
• Titles. Tried by the Flow renders À l’épreuve du flux: the datum put to the test of the flow. This tome is The Trace or the Datum — The Situated Reader Tried by the Flow.