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.

The Story of the Rewriting

How the Trilogy Went from Essay to Edition

“Not ‘the essay was fragile’. The essay carried its repairs scattered about; the edition brings them to their locus.”

— Note on the Re-edition, second edition

A Finished Trilogy

In the summer of 2026, Tried by the Flow had three finished tomes. They follow the datum from the act that constitutes it to the reader who receives it, by way of its circulation. A synthesis accompanied them, along with a lexicon and a translation charter. The images held together, from the river and the canal to the reader’s window.

The book had been written in layers, as The Story of the Essay recounts: each document called for another. This way of working had given it its momentum. It had also left marks. Passages came back from one tome to another almost identically. A caution set down in one chapter disappeared in the next, so that a proposition qualified here reappeared further on without its qualification. Some images promised more than the argument demonstrated. Several internal cross-references, finally, pointed to the wrong document.

It was not a matter of writing another book. The intention was to take this one up again so that its distinctions would hold and its progression would read better.

The Consolidation Document

The starting point is a critical text dated August 2026, Consolidating the Processual Ontology of the Datum. It rereads the three tomes in the light of other disciplines and draws up the list of the objections that seem decisive to it.

This text was reread in its turn. The review that came out of it checks its sources, retains lines of work and proposes the plan of a second edition, down to its name. One rule dates from that moment, and it held until the end of the work: no change of substance enters the text without the author’s arbitration.

Words First

The first sessions touched no chapter. They produced instruments.

The rewriting charter sets the rules of the work. The first is simple: a concept is demonstrated in a single place, and everywhere else the text recalls it in one sentence before referring to that place. The charter also lists sixteen strong formulations of the first edition, to which three related motifs are added, and decides for each the form that the edition accepts.

The new terms were recorded in a lexical register, then arbitrated one by one: datum of origin, occurrence, filiation, provenance, the five states of the datum, validity for a use, the responsibility of the chain, among others. Twenty-one became canonical. Three remained confined to the formal machinery.

The synthesis gave way to the Architecture Volume, which gathers the theses, the overall diagram and the formalisms. Freed from this machinery, the tomes could keep their essay language and their images. This separation is among the choices that give the edition its form.

The Third Tome First

The reworking did not follow the order of the tomes. It began where the stake was heaviest, in the third, on the relation between the trace, the datum and reading.

The first edition implied that a datum that no one reads becomes a trace again. The second separates the existence of a datum from its actualization by a reader, and describes five states: constituted, accessible, intelligible, actualized, operative. For want of a reader, a datum does not cease to be constituted. It can remain out of reach, or not understood; it can also simply be left aside. Reading actualizes something for someone. It does not make the datum exist anew each time.

This correction spread through the whole trilogy. What was said about “living” data, about automata, about circulation and about the threat of the trace had to be taken up again. The image of the trace remained, with another scope: it now designates the extreme case, the datum that no one any longer knows how to read, and not the fate of every datum momentarily neglected.

The Chapters, Batch by Batch

Next came the chapters, taken up in batches. For each one, the author had the text before and after the correction in front of them; the author validated, refused, or asked for another wording.

In the first tome, the datum was distinguished more clearly from the act that produces it: it is an entity constituted by a process, which has a history and which relations maintain. Observational drawing had until then served as the implicit model for every datum. It lost this privilege, for a temperature entry and a marriage inscribed in the civil register are not born of the same mechanism, any more than an average calculated on other figures. The text now distinguishes observational, institutional and derived data. Lapse was made more precise in the same movement: a datum does not become bad by ageing; its validity is judged for a given use.

The second tome gained in precision about what circulates. Two files can carry occurrences of one and the same datum, whereas a transformation produces a new one, attached to the previous ones by a filiation. Exchange no longer “accomplishes” the datum: it widens its availability and what can be done with it. The torrent remains the image of excessive circulation, but the text now separates in it throughputs, processing capacities and the window of the one who reads.

In the third, reading by a machine is treated with more caution. The text no longer pronounces on the question of meaning in the machine. It says that a machine occupies a position from which it handles or processes a datum, which is not enough to make it a reader in the strong sense. The emphasis is placed on responsibility: delegating a processing exempts no one from answering for the choices made in it. The end of the trilogy also develops the loop that leads from observation to action. Data serve to decide; what is decided changes the real, which in its turn leaves new traces. Responsibility then extends to the whole chain, and no longer only on the reader.

The Assembly

One session brought the three tomes together. The methodological prefaces, which repeated themselves from one tome to another, became a single Preface to the Edition, and each tome now opens on a one-page Threshold. The “Landmarks” sections, at the end of each document, carry the reminders and credit what is established elsewhere. The cross-references by document number were checked one by one.

The following session completed the apparatus of references. The annotated bibliographies of the first edition were taken up and completed: twenty-five cross-references left without a destination find their target there. The lexicon was consolidated, the translation charter updated. An insertion note specified what, on the site, would replace what.

The rewriting had deleted, moved and added. It had not only shortened; the material was redistributed so that each passage would have a precise function.

At that moment, the edition seemed finished.

The Audit

It was not. An external editorial audit reread the thirteen pieces delivered, and its verdict begins thus: “The edition is not publishable as it stands.”

It pointed out four contradictions in the definitions themselves. The datum was in them sometimes an entity, sometimes an act. In the first tome, the act of reading constituted a datum from a trace, whereas the third reserved it for the actualization of a datum already constituted. The observational model passed for the foundation of every datum whereas the three families contradicted it. Translation, finally, appeared both among copies and among filiations. To this were added factual assertions that went beyond their conditions of validity: particular cases had become laws.

The audit also drew up the list of what must not be undone. The architecture held. What contradicted itself were certain statements.

The Corrective Branch

The audit opened a second campaign, numbered from S18 to S23.

S18 first made the audit verifiable. Twenty-two findings were examined one by one. The exact state of the pieces was fixed in an integrity manifest, the decisions to be taken were gathered in an arbitration file, and a plan divided up what followed.

The text was corrected in the course of the four following sessions. S19 went back to the doctrinal foundations of the first tome and of the Architecture Volume. In S20, the task was to make the acts, identity and circulation consistent, down to the lexicons and the translation. Factual assertions were bounded to their conditions in S21, and readability occupied S22. This represents about a hundred and twenty substitutions and eight additions, all documented.

S23 brought together the last checks, the remaining repairs and the preparation of the delivery.

The Proof in Question

This last phase lasted, because the checks themselves were in question. Some were too literal: they looked for a string of characters where an idea had to be recognized. Others concluded too broadly. Instruments had to be corrected, judgements revised, and what was established separated from what remained in reserve.

A bibliographical entry in the name of Lehmans, for example, referred to no verifiable reference. It was withdrawn, with the author’s agreement. The mention of Anne Lehmans and Camille Capelle, which refers to an identified work, remained in place.

The Closure

Six test benches were passed, from formulations to facts and the apparatus of references, by way of the lexicon, the images and doctrinal consistency. Five are compliant. The sixth is compliant with eight explicit reservations: quotations to be checked in their original edition (Bergson, Naisbitt, the accounting standard IAS 38, a letter from Peter Naur), and checks that could be made only once the edition was published.

The delivery counts fifty files, whose hashes have been checked. Not all are meant to be read: beside the texts of the edition, it keeps the states of the first edition on which the work was done, and the documents of the work itself.

On 2 October 2026, the author ascertained compliance with reservations and declared the closure.

From One Edition to the Other

The second edition takes the same way as the first, from the spring to the reader, with the same images and the same language of the canal. What changes is precision. Operations that the first confused are now separated, with their conditions. The demonstrations are better distributed, and what is established is distinguished more clearly from what remains open. This work of consolidation is not presented for all that as the definitive proof of the theses.

The first edition remains available in the archive, as it appeared, with the story of its birth.

The Author and the Models

The essay had been written with a single language model. The rewriting called on two, with separate roles. One, in the role of the writer, wrote the whole text of the second edition. All the checks fell to the other, the controller, which examined the writer’s deliverables and formulated recommendations. It is on these recommendations that the author decided, whether it was a matter of the new terms, of the batches of corrections, of the withdrawal of the Lehmans entry or of the closure. The credit of the edition says it in one line: conceived and supervised by David Bories, directed and written by generative artificial intelligence.

Landmarks

This story relies on the synthesis of the rewriting written at the end of the work and on the pieces of the delivery — the consolidation document and its review, the charter and the lexical register, the external editorial audit, the session notes S1 to S23, the progress review of 13 September 2026 and the author’s decision of 2 October 2026. These pieces are published with the second edition, each with its hash.

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.

• 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 and data as its plural.
• The five states of the datum: constituted, accessible, intelligible, actualized, operative. Actualized (actualisée) names the act by which a reading produces a meaning, never the updating of a file.
• Drawing (prélèvement): the gesture of drawing from the flow; entry renders the inscribed value (un relevé de température).
• Lapse (péremption): a datum’s validity is judged for a use; neglected (délaissée) says that no reader actualizes it.
• Locus (plural loci) renders foyer, the single place where a concept is demonstrated; Landmarks (Repères) is the closing section; Threshold (Seuil) is the opening page of each tome; the formal machinery (l’appareil formel) belongs to the Architecture Volume.
• Finding renders constat; the author ascertained (a constaté) compliance.
• Titles. Tried by the Flow renders À l’épreuve du flux; this volume is The Story of the Rewriting; The Story of the Essay renders Le récit de l’essai, the story of the first edition. The working documents named in the text are in French; their titles are translated (Consolidating the Processual Ontology of the Datum for Consolider l’ontologie processuelle de la donnée).