How did writing transform thought?

Writing did not merely record thought: it transformed it. By externalising memory, it freed the mind from the burden of retaining everything; by fixing words, it made thought inspectable, comparable, open to criticism. Each major medium that followed — the printing press, the screen, artificial intelligence — has replayed this shift: each time, it is not only the tool that changes, but what thinking means.

This page follows the journey that L’Odyssée des idées (in French; 2026) unfolds along its path “Les signes” (“The Signs”): from the first signs to AI, the media first, thought second.

Before writing: thinking with the voice

For most of its history, humanity thought without writing. An oral culture is not a culture without knowledge — it is a culture in which knowledge has to fit within living memories: hence rhythm, formula, verse, repetition. Milman Parry’s work on Homer showed it: the epic is not a recited text, it is a mnemonic machine — the poet recomposes at each performance with calibrated formulas. Thinking orally means thinking in flow, in front of witnesses, with no possibility of going back.

The shift: when memory leaves the body

Writing was born in Mesopotamia some 5,000 years ago — not for poetry, but for counting: sheep, jars, debts. This founding detail says the essential: to write, first of all, is to create a binding record — accounting and law before literature.

Then the trace reveals its powers. The anthropologist Jack Goody (The Domestication of the Savage Mind; in French, La Raison graphique) showed that writing does not merely record pre-existing thoughts: it creates new intellectual operations. The list, the table, the column do not exist orally — and it is these that make classification, term-by-term comparison and formal logic possible. Walter Ong (Orality and Literacy) summed it up in a formula: writing restructures consciousness. A text, unlike speech, can be reread: one can return to it, confront it with another, find the contradiction in it — the critical mind is a child of writing.

The shift had its trial, and it has remained famous: in Plato’s Phaedrus, King Thamus predicts that writing will produce forgetfulness — people will trust the external sign rather than living memory, and written words, when questioned, “preserve a solemn silence” (275d, tr. Harold N. Fowler). The irony is double: this criticism has reached us only in writing, and Socrates — the philosopher without books, who never wrote a line — still thinks in us today only through the texts of his students. Each subsequent medium would replay exactly this trial.

The printing press: thought changes scale

Gutenberg (around 1450) does not change the nature of the text — he changes its regime of existence. Reproduced identically by the thousand, the text becomes standardised: copying errors stop accumulating, editions can be compared, and scholars in different cities can at last discuss the same text. The historian Elizabeth Eisenstein showed what modern science owes to this fixity: reproducible tables, maps and figures, on which one person’s observation can correct another’s.

And print produces something else besides: a public. The same text read everywhere at the same time — Luther’s Reformation was the first massive demonstration — creates the new entity that would come to be called public opinion. The press, then radio and television, would only accelerate this manufacture.

Digital media and AI: the third shift

The screen picks up the sequence: unlimited memory, instant access, fluid text that is searched more than it is read. Then artificial intelligence moves the frontier one notch further: after memory (writing) and diffusion (printing), it is a part of judgement that is externalised — formulating, summarising, answering, choosing.

The lesson of the first two shifts holds for the third. Thamus’s trial is replayed point for point — AI will produce forgetfulness, laziness, dependence — and it has the same share of truth and the same blind spot: a medium does not abolish thought, it redistributes it. The real question is not whether to bless or curse the tool: it is to see what is freed, what atrophies, and who controls the medium. On this last point, the author’s anthropic framework extends the analysis: when cognitive externalisation saturates, the burden returns to attention and judgement — this is the anthropic loop.

What this history teaches

Three regularities run through the three shifts:

1. The medium precedes the content — accounting before literature, almanacs before encyclopaedias, entertainment before online knowledge: the nobler uses come second.
2. Each externalisation both frees AND atrophies — writing did weaken the living memory that Thamus defended; it also made Euclid possible. The balance sheet of a medium is never read in a single column.
3. Each shift has its trial — and the trial is rarely wrong about the losses, almost always wrong about the balance.

The complete journey

This page follows a single thread. The book L’Odyssée des idées (in French; new 2026 edition, 696 pages) unfolds it among 250 stopovers — a work, a life, an invention, a shift — linked by seven reading paths, among them “Les signes” (“The Signs”), which runs from the first signs to artificial intelligence. Each stopover follows the same movement: a scene, a shift, a closing note. The one on Socrates — “the philosopher without a book” — ends like this (our translation):

No definitive answer. No closed doctrine. Nothing but a way of not letting words sleep in peace. […] twenty-four centuries later, he does not teach us what to think. He stops us from believing too quickly that we are thinking.

Further reading

• L’Odyssée des idées — the book’s page (in French; 250 stopovers, seven paths, excerpt).
• Which general-knowledge book to give as a gift? (in French) — if the journey is meant for someone else.
• What is the anthropic loop? — what cognitive externalisation becomes when it saturates.

All resources: data, explainers and reference material →