The corpusThe licenceThe 21The aggregateThe graphSourcesHow to reach themThe vaultMethod
Twenty-one op-eds, one permission, no licence
For World News Day on 28 September 2026, WAN-IFRA and the Canadian Journalism Foundation published twenty-one commissioned opinion pieces by editors, publishers and directors from Manila to Managua, and made them free to republish. That is a generous act, and an unusual one. This section takes it seriously enough to ask what it actually amounts to for anyone — a newsroom, a translator, an archive, a machine — who wants to act on it.
We fetched all twenty-one, froze the bytes, hashed them, and asked three questions.
Under what licence?
“Free to republish” is a permission. A licence is a permission with a name, a version and an address, so that a machine can find it and a lawyer can read it. Which of these is on the pieces?
The answer, in the bytes →In what form?
Twenty thousand words of licensed material is a corpus. Is any of it published in a form a machine can consume — a licence field, a feed, a manifest, a graph?
What is already there →What do they argue?
Twenty-one of the industry’s most senior people answered the same question in the same week. Counted together, what did they reach for — and what did none of them say?
The aggregate →Beta, agent-produced. The fetching, extraction, classification and drafting here are done by software agents against frozen bytes. Nothing on this page is a legal opinion, and there has been no legal review. If you are deciding whether you may republish one of these pieces, read the terms on the piece itself and ask the publisher — that is the point we are making.
We hold all twenty-one pieces and republish none of them. Every article is linked to its original on worldnewsday.org. What is published here is the description: who wrote it, under what terms, what it links to, what words it contains. The prose belongs to the people who wrote it, and the gate on this section fails the build if twelve consecutive words of any piece appear on any page we generate.
Every number walks back to bytes we hold. Each page was fetched once, frozen to a dated snapshot in this repository and hashed with SHA-256. The counts are re-derived from those bytes by a second program before anything ships. The register → · The method →
The short version
That gap is not pedantry. A newsroom in São Paulo that wants to translate one of these pieces has to read a sentence, interpret “faithful to the original meaning”, and hope. An archive that wants to keep them has nothing to record in its rights field. A machine that wants to know whether it may index, translate or train on them finds no statement at all, because every one of the twenty-one already ships structured data and not one of them uses it to say this. The generosity is real. The infrastructure for it is missing, and it is one line of JSON long.
Where to start
The argument, in one pieceFree to republish is not a licence
Twenty-one of the most senior people in news gave their work away on the same day. Not one of them said, in a form a machine can read, on what terms — and the line that would fix it is already half-written on every page. Written to be sent to them, and free to republish under a licence with a name.
Read it →Under what licence, exactly?
Twenty-one pieces, twenty-one prose permissions, two wordings, and no licence with a name. What the sentence actually permits, what it leaves open, and the one line of JSON that would fix it.
Open →The corpusThe twenty-one
Every piece: who wrote it, where they work, how long it is, what it links to, what it is about. Linked to the original, never reproduced.
Open →The aggregateWhat they said together
Twenty-one of the most powerful voices in news, asked the same question in the same week. What they all reached for, what only two mentioned, and what none of them gave their readers.
Open →The graph160 nodes, 427 edges
The same corpus as a graph you can walk: articles, authors, organisations, themes, cited sources, licences and the frozen files underneath. Every verb has a named inverse and a Portuguese form.
Open →The fractal layer1,070 sentences, typed
Each piece is its own graph of claims — assertions, questions, hypotheses, proposals — connected to the others by a shared vocabulary. Same shape at every zoom level. And four places in 1,070 sentences where anybody says what a word means.
Open →The outreachHow to reach them
Twenty-three authors at twenty-six organisations, and a corpus that gives you no route to any of them. What going to their own sites produced, under rules published before the looking started.
Open →The evidenceEvery file, every hash
43 frozen files with their SHA-256, the dated snapshot they belong to, and what this section has already had to correct about them in public.
Open →The methodHow this is built, and what it refuses
Fetch, freeze, hash, extract, derive, gate. The published lexicon. The four things this section will not do, enforced by a program rather than a promise.
Open →Whose work this is
The op-eds are not ours and are not reproduced here. They were commissioned by World News Day, a campaign run by WAN-IFRA with the Canadian Journalism Foundation, and every one of them is a click away from the corpus page. Read them there. What this section adds is the layer underneath: the terms, the structure, the graph and the evidence trail — the things this publication has argued for since its first page, applied to somebody else’s corpus rather than our own.
For an agent
Everything on this page is derived from four files you should fetch instead of parsing HTML: data/corpus.json (the twenty-one, described), data/licences.json (the terms, counted), data/analysis.json (the aggregate) and data/graph.json with data/ontology.json (the same thing as a typed graph, plus data/triples.nt as N-Triples). data/register.json maps every claim to a frozen file and its SHA-256. The prose of the op-eds is deliberately not in any of them — follow the url on each article to the publisher.