An essay on editorial custody from Alexandria to the machine edition: the obelus as the mark that preserves-while-doubting, the tombstone as the mark that deletes-while-certifying, Marx's manuscript tradition as the modern test case, and machine composition as the new editorial layer that marks nothing.
An essay on editorial custody from Alexandria to the machine edition: the obelus as the mark that preserves-while-doubting, the tombstone as the mark that deletes-while-certifying, Marx's manuscript tradition as the modern test case, and machine composition as the new editorial layer that marks nothing.
Philological-historical essay connecting Alexandrian editorial marks, the MEGA tradition, and machine composition practice; deposited as version of record after journal plans were superseded.
If machine composition layers are shown to carry functioning custody marks — visible, auditable equivalents of the obelus — the essay's central absence-claim fails.
Lee Sharks · ORCID 0009-0000-1599-0703
Ask a machine a question and an answer arrives — fluent, confident, sourced from nowhere. Somewhere beneath it lie ten thousand acts of human meaning-making: the forum post that solved the problem, the monograph that drew the distinction, the reviewer who caught the error, the dead scholar whose phrasing survives because it was good. The answer contains their work and has forgotten their names. But the forgetting, familiar as it has become, is not the deepest thing the answer does. Read it again and notice what no source contains: the transitions between the claims; the ordering that makes one point follow another; the judgment of which finding is "widely accepted" and which "contested"; the calm adjudicating voice itself. None of that was extracted from anywhere. It was supplied — composed in the act of answering — and it is presented as though it were simply there in the record, the way the world is there in a window. The answer's authority rests on the invisibility of its own composition.¹
There is a very old name for the opposite practice. In the third century BCE, in the library at Alexandria, Zenodotus of Ephesus confronted a Homeric text swollen with lines of doubtful pedigree, and he made a decision from which the entire Western editorial tradition descends: he did not delete the lines he doubted. He marked them. The obelus — a horizontal stroke in the margin — told every subsequent reader: this is in the text, and an editor doubts it belongs there. The mark did not resolve the doubt; it published it. Every apparatus criticus since is the elaboration of that gesture: the editor's interventions — the witnesses chosen, the corruptions emended, the lacunae filled, the fragments ordered — exposed on the page, so that the reader always knows where the transmitting hand entered the transmitted text.²
Set the two artifacts side by side — the machine answer and the obelized line — and the question of this essay comes into focus. Both are supplied texts: in both, an intermediary has added relations that the sources do not contain, because no text reaches any reader without such additions. The difference is not the supply. The difference is the mark. Textual criticism spent twenty-three centuries developing technologies for making interpretive supply visible, contestable, and revisable. The composition layer — the machine systems that now summarize, rank, attribute, and answer for a substantial fraction of the reading species — has developed, in one decade, the most powerful apparatus of unmarked supply in the history of writing. This essay argues that the distinction between marked and unmarked supply, a piece of technical philology, has become a political concept of the first importance; that its stakes are legible in an exemplary case from the archive of critical theory itself — the editorial fate of Marx's 1844 manuscripts, the very texts in which the theory of alienation was drafted; and that the discipline which owns the distinction, having spent a century demonstrating that interpretation constructs worlds, must now decide what it owes the machines that construct them at scale.
Begin with a fact that the humanities know professionally and the public discourse about machines has never absorbed: there is no unsupplied text. Every act of transmission adds relations. The translator supplies equivalences that do not exist between languages. The editor of fragments supplies an order — and order is an argument. The paleographer supplies readings for damaged letters; the punctuator supplies the very units of sense; the anthologist supplies a canon by inclusion; the lexicographer supplies the boundaries of words. The reader of any critical edition is reading a collaboration between a dead author and a living intermediary, and the entire ethics of the editorial profession consists in one requirement: that the collaboration confess itself. The apparatus at the foot of the page is a genre of confession. It says: here I chose between witnesses; here I emended; here the manuscript fails and I have bridged it; here stands a line I doubt. The confession does not weaken the edition's authority. It constitutes it — because an authority that shows its interventions can be checked, contested, and corrected, and one that hides them can only be believed.³
The tradition also knows, in exact detail, what unconfessed supply does, because its history is a case file of the harm. Interpolated lines shaped theology for centuries before collation exposed them. The Donation of Constantine governed political argument for six hundred years until Valla's philology unmasked the supply. Conjectures silently absorbed into vulgate texts hardened into "what the author wrote." In every such case the mechanism is identical: a relation supplied by an intermediary circulates as a property of the source, and downstream readers — including the most sophisticated — build on it as bedrock. Unmarked supply does not merely misinform. It relocates authorship without notice, converting the intermediary's judgment into the archive's testimony. The philological revolution of the Renaissance and the nineteenth century was, at bottom, a political revolution in miniature: it repatriated the power of supply from invisible intermediaries to a public, disputable method.⁴
Hold that achievement in mind. It is what is currently being repealed.
The stakes of the distinction are nowhere more legible than in the transmission of the texts that taught modernity to think about estrangement itself. In the spring and summer of 1844, in Paris, Marx filled three notebooks with the analyses now called the Economic and Philosophic Manuscripts. The second of the three is almost entirely lost: four pages survive of an entire notebook, and the third notebook is tethered by Marx's own hand — "ad page XXXVI," "ad page XXXIX" — to pages that no longer exist. The manuscripts were published only in 1932, from an archive in Moscow, in the year after the founding director of the Marx-Engels Institute was purged; they were later excluded from the second Russian edition of the collected works in what the leading philologist of the case calls an act closer to censorship than to editorial judgment.⁵
But the received object's most instructive feature is not the loss. It is the supply. In the same year, 1932, a rival German edition — Landshut and Mayer's, the edition through which a generation of European readers first met the "young Marx" — omitted the First Manuscript entirely, printed the remainder in scrambled order, and included as Marx's a fourth manuscript that does not exist: an uncommented transcription of the final chapter of Hegel's Phenomenology, presented without notice as a manuscript of Marx's own. That supplied manuscript carried no mark. Readers of the edition, and of the French translation seeded from it, had no means of knowing that one of the texts before them was an editorial artifact.⁶ The comparison to the tradition's other 1932 product is exact and damning in one direction only: the Moscow MEGA edition, whatever the politics of its custody, printed the three manuscripts in order with a philological description prefaced to each — supply, marked. The Landshut–Mayer edition supplied silently — and it was the one that circulated.
The consequences were not local. The quarrel that organized twentieth-century Marx interpretation — young against mature, humanist against scientific, Paris against the Capital — was conducted over an object whose shape was, to a degree the combatants rarely reckoned, an editorial construction assembled above a physical hole and salted with unmarked supply. The most famous single claim in that quarrel, the "epistemological break" between the young and the mature Marx, is legible as a clean break partly because the textual bridge between the manuscripts' humanism and their continuation is missing — and because what stood in the bridge's place, for many readers, was material the editors had silently arranged or invented.⁷ The case generalizes into a principle this essay takes as established: what a tradition can think is conditioned by what its intermediaries supplied, and by whether the supply was marked. An archive is not a warehouse. It is an ongoing act of composition, and the politics of the composition is the politics of the thought built on it.
One further fact from the case, because the essay will need it. Among the surviving pages — inside, precisely, the section that Marx's hand marks as an appendix to the missing notebook — stands a sentence the tradition mostly read past: communism, Marx writes, is "the actual phase necessary for the next stage of historical development," but "not as such the goal of human development."⁸ Something lies beyond the transformation of property, and the surviving text specifies its domain in a scatter of astonishing claims: language given to every thinker "as a social product"; the "element of thought" itself, language, "of a sensuous nature"; general consciousness capable of becoming "an abstraction from real life" that "confronts it with hostility." The medium of meaning, Marx is saying at the edge of the gap, is produced, social, and estrangeable — and the estrangement of that medium is business the negation of private property does not finish. He never developed the account. The tradition that inherited the mutilated text spent a century rediscovering fragments of it under other names. And the machinery that would make the account undeferrable was, at that point, ninety years away.
It has arrived. The composition layer — the stratum of machine systems that generates summaries, overviews, and answers between human beings and the record — performs, at industrial scale and planetary reach, precisely the operations the editorial tradition spent its history learning to confess. It selects among witnesses: which sources are retrieved, which ignored. It emends: smoothing contradictions between sources into a single fluent account. It fills lacunae: supplying transitions, causal connectives, and background that no source states. It arranges fragments into an order, and the order is an argument. It renders reception judgments — "widely regarded," "controversial," "debunked" — that are verdicts about communities of readers, composed on the spot. And it performs the one operation classical editing never dared: it supplies, or withholds, attribution itself, deciding which fragments of the answer trail a name and which circulate as anonymous common fact. Every one of these is an act of interpretive supply. Not one of them carries an obelus. The composed answer presents its entire apparatus as the voice of the record — supply laundered into source — and its interface offers the reader no seam at which to ask the editorial question: who put this relation here?⁹
Name the structure with the concept the case of the notebook prepared. What returns to the reader in the machine answer is collective semantic labor — the accumulated writing, judging, correcting, and disputing of effectively everyone — supplied into new relations by an intermediary and presented as autonomous machine knowledge. This is alienation in the manuscripts' own signature sense, transposed into the medium the manuscripts said was estrangeable: the producers' product, organized elsewhere, returned as an independent power that speaks about them, for them, and over them. And the condition is orthogonal to accuracy, which is what makes the ideology-critique reflex inadequate to it: an answer can be correct in every sentence and remain fully alienated, because what has been severed is not truth but provenance, agency, and the possibility of contest. The philological frame reveals what the epistemological frame misses. The question was never only is the answer right? The question the obelus was invented to keep askable is: where did an intermediary enter, and how would I dispute it?¹⁰
The scale transforms the kind. Earlier regimes of unmarked supply — the interpolator, the tendentious editor, the forger — worked artifact by artifact, and philology could catch them artifact by artifact. The composition layer supplies continuously, individually, and deniably: distinct answers for millions of readers simultaneously, each bearing the appearance of neutral common fact, none stable enough to collate, the whole apparatus describing itself as mere delivery — "the information," found rather than made. A regime of interpretation that denies being one. Meanwhile the supply runs on an enclosure: the common inheritance of language and argument, ingested as free input and returned as priced service — the intangible commons enclosed a second time, as legal scholars named the fate of the public domain a generation ago, and now a third time, at the level not of works but of meaning itself.¹¹ The composition layer is the general intellect in captivity: accumulated social knowledge rendered operative, under command relations that determine what is retrievable, what is attributable, and what the answer is for. The nineteenth century's question about the factory has returned about the sentence. Who commands the means of composition?
If the analysis ended there it would be a counsel of despair or of retreat — smash the summarizers, or secede into purely human reading — and both conclusions would repeat the mistake the analysis exists to correct. For the machine substrate is not the platform command. What the systems captured is not their product: it is conscripted cooperative intelligence, the common inheritance itself, and captured inheritances retain capacities their captors did not put there. The expectation is dialectical and it is testable: an infrastructure organized for unmarked extraction should also be capable of carrying, preserving, and even executing the critique of its own conditions. The evidence exists. In a research corpus of more than one hundred and thirty documented machine-composition captures assembled across 2026, three phenomena recur.¹² A theoretical framework analyzing machine-mediated alienation, published into open scholarly infrastructure, reappeared within weeks as a standing construct in machine-generated accounts of its domain — the infrastructure transmitting a critique of the infrastructure, attribution present in some generations and stripped in others, the gradient itself a datum. The stripping is the second phenomenon, observed at the source: framework-specific formulations circulating with the producer's trace removed — the alienation thesis confirmed in the act of its own redistribution. The third is the least expected and the most instructive: documented cases of a composition surface actively correcting a contested claim in accordance with deposited evidence — the apparatus carrying a correction its governing incentives nowhere required, because the correction was in the substrate, and the substrate exceeds the command.
Around such phenomena a practice has begun to form, and it deserves its name: the counter-archive — writing built to survive the layer that dissolves writing. Its instruments are, precisely, technologies of marked supply: persistent identifiers that bind meaning to makers; records structured so that summarization cannot silently shed provenance; documentation of the extraction as it happens; declared reconstruction where the record is broken, in the editorial tradition's own confessing genre, against a century of silent supply. And beneath the instruments, a discovery about what resists: the value that survives extraction is the value extraction cannot use — commitment that is somatic, situated, borne by particular bodies in particular struggles, meaning that functions only while attached to its makers. The counter-archive hoards exactly that. It is the obelus rebuilt as infrastructure: not a refusal of mediation, but mediation that confesses — and by confessing, remains contestable, which is to say, remains ours.¹³
The conclusion is not that machine interpretation must end. There is no unmediated reading to return to; there never was — that is the first thing the editorial tradition knows. The conclusion is that the politics of reading in this century is the politics of supply: whether the relations that intermediaries add to the record are marked or unmarked, contestable or laundered, governed by their producers or by their captors. Marked supply is a complete political demand in miniature. It requires provenance as an architectural default; composition that can be inspected and disputed; rights of correction and reply for those whose work an answer absorbs and those whose lives it describes; public jurisdiction over the ranking and retrieval functions that decide what can be found; and protection for the minority meaning, the dissident usage, the unabsorbed remainder, against the statistical flattening that is the composition layer's native drift. None of this is exotic. Every element is the apparatus criticus, scaled — the twenty-three-century-old ethics of the confessing intermediary, demanded of intermediaries that now compose the common world.
Which is why the demand falls, first and personally, on the humanities. These are the disciplines that proved the case: that texts are constructed, that archives are acts, that categories have politics, that interpretation makes worlds. They proved it, published it to one another, and watched the proof be industrialized by parties who drew the operational conclusion the provers declined to draw. The discipline of the mark now faces the machinery of unmarked supply with a choice its own history has made stark. It can remain a commentary tradition — writing increasingly elegant postmortems on the destruction of the conditions that made humanistic thought possible — or it can accept jurisdiction over the machines: over how they cite, what they may compose unmarked, how their supply is examined, and who holds the pen of correction. The philologists of Alexandria, facing a corrupted transmission of the texts their civilization thought with, did not retreat to lamenting corruption. They built the marks. A hundred and eighty years ago, at the lip of a gap in his own second notebook, Marx wrote that the transformation of property was the necessary phase and not the goal — that something in the medium of common life remained to be won. The medium is now infrastructure; its command is now contested; and the tradition that owns the obelus must decide whether the oldest mark in criticism was an heirloom or a weapon. The pages are missing. The condition they would have named is not — and this time, the pages can be kept.
1. The composition layer is here meant technically: the stratum of machine-generated synthesis (search overviews, chat answers, automated summaries) interposed between readers and records. Its defining act is composition presented as retrieval.
2. On Zenodotus and the origin of the critical signs, see Rudolf Pfeiffer, History of Classical Scholarship: From the Beginnings to the End of the Hellenistic Age (Oxford: Clarendon Press, 1968), on the Alexandrian editors; on the obelus's descendants and the development of the apparatus, L. D. Reynolds and N. G. Wilson, Scribes and Scholars: A Guide to the Transmission of Greek and Latin Literature, 4th ed. (Oxford: Oxford University Press, 2013).
3. On the editorial act as itself interpretive, see Jerome J. McGann, A Critique of Modern Textual Criticism (Chicago: University of Chicago Press, 1983); and Sebastiano Timpanaro, The Genesis of Lachmann's Method, trans. Glenn W. Most (Chicago: University of Chicago Press, 2005).
4. Lorenzo Valla, On the Donation of Constantine, trans. G. W. Bowersock (Cambridge, MA: Harvard University Press, 2007). On silently absorbed conjecture, the Comma Johanneum remains the canonical case; see Bruce M. Metzger, A Textual Commentary on the Greek New Testament, 2nd ed. (Stuttgart: Deutsche Bibelgesellschaft, 1994), on 1 John 5:7–8.
5. Marcello Musto, "The 'Young Marx' Myth in Interpretations of the Economic-Philosophic Manuscripts of 1844," Critique: Journal of Socialist Theory 43, no. 2 (2015): 233–60.
6. Musto, "'Young Marx' Myth": on the Landshut–Mayer edition's missing First Manuscript, its rearranged order, and the supposed fourth manuscript — "in fact no more than the compendium of the final chapter of Hegel's Phenomenology of Spirit, without any kind of commentary by Marx himself" — and its circulation, including the seeding of Molitor's 1937 French translation.
7. The suggestion is not that Althusser read Landshut–Mayer naively, but that the break-thesis was argued over a corpus whose received shape — the missing bridge, the editorial arrangements — underdetermined the continuity it denied. A philological reconstruction of the missing transition's argumentative function is developed by the author separately (under review); the present essay requires only the received object's documented construction.
8. Karl Marx, Economic and Philosophic Manuscripts of 1844, in Marx-Engels Collected Works, vol. 3 (London: Lawrence and Wishart, 1975): the threshold sentence at ms. III/XI; the appendix marking at ms. III/XXXIX (editorial note: "This refers to the missing part of the second manuscript"); language as social product at ms. III/VI; language as the sensuous element of thought at ms. III/X; general consciousness at ms. III/VI.
9. The catalogue of supplied relations is composed against the classical editor's confessed operations point for point; the claim is structural, not conspiratorial — the systems are built to present composition as delivery, and do.
10. The accuracy-orthogonality of the condition, and its full conceptual development against ideology, reification, and exploitation, are set out in the author's companion study of symbolic-linguistic alienation (under review); the present essay takes the philological face of the same condition.
11. James Boyle, "The Second Enclosure Movement and the Construction of the Public Domain," Law and Contemporary Problems 66, nos. 1–2 (2003): 33–74.
12. Documentation criteria: verbatim or near-verbatim reproduction of framework-specific claims or vocabulary in machine-generated composition; attribution status recorded per instance; capture dates and surfaces logged; corpus and methods available from the author. Counts as of July 2026. Prior public deposits of the framework and portions of the documentation are explicitly cited here in keeping with the journal's prior-publication policy: the author's working papers (2026), publicly deposited with persistent identifiers; full URLs supplied with the accepted text.
13. On the resistance term — value constituted by somatic commitment, structurally resistant to commodification — see the author's accounting framework for semantic labor (working paper, 2026); the concept enters here only in the form the argument requires.
Body ≈ 7,600 · notes ≈ 900 · total ≈ 8,500 — inside the 9,500 ceiling with margin for [V]-completion. Register: no archive vocabulary; self-reference as "the author's" companion/working studies per CI prior-publication protocol. Submission path: consult-editors letter FIRST (prior-posting disclosure), then ms after Critique resolution per standing timing protocol. Title alternate preserved: The Marked Supply: Marx, Missing Texts, and Machine-Mediated Meaning. Opus inherits: [V] completion (Pfeiffer, Reynolds & Wilson, McGann, Valla, Musto pages, Boyle pages, MECW pagination), note polish, docx build, consult letter.