Where we lose
This site is published by the sgit project, which is building the future-of-news stack it argues for. That is a participant's position, not a neutral one, and per this site's own argument about agenda as context, disclosing it is not optional.
Two years of disclosed model co-authorship
This site argues that authorship and agenda should be disclosed structurally, not left for a reader to infer. It can point to its own track record on exactly that point. Across the ten-article library this site cites: most core articles credit "Dinis Cruz" and "ChatGPT Deep Research" jointly, several add Claude 3.5 or Claude 3.7, and the practice of naming an AI co-author in structured front matter runs back to the first article in February 2025 — eighteen months before this site existed. Preserving that credit verbatim on every republished or synthesised page here is not a courtesy; dropping a co-author on republication would be exactly the attribution failure this site exists to argue against. A site arguing for provenance and disclosed agenda, which has a two-year dated record of disclosing its own model co-authorship, is putting that record on this page rather than treating it as an awkwardness to manage.
Where this is not, and cannot yet be, what it argues for
1 · Nothing here runs
The provenance pages, the decision graph, the £8.40 ledger, Trust-as-a-Service — all specified, none built. The line, stated in full.
2 · The metering mechanism is unspecified
Theme 4 is well argued and mechanically open: how a read of a cited source is detected, attributed, priced and settled has no answer anywhere in the source material. Stated as open, not solved.
3 · The self-critique is real, not decorative
Agenda-tagging is a tool that can discount a true and inconvenient finding as easily as a false and convenient one. This site names that rather than hiding it — but naming a risk is not the same as having solved it.
4 · It inherits a real technical weakness
The vault delivery substrate this site would eventually run on can be invisible to crawlers. For a news site that is close to existential. Stated on the network page, unresolved.
The licence position, stated rather than assumed
| Material | Licence |
|---|---|
This site's own content — every page, llms.txt, index.md, the brief pack in briefs/ | CC BY 4.0 |
| This site's code — build tooling, stylesheet, workflow | Apache 2.0 |
Source material on docs.diniscruz.ai (the library) | CC0 1.0 Universal at source — a public-domain dedication, no attribution required. Stated per page per the provenance contract |
These are different grants over closely related material. CC0 is the more permissive of the two, so republishing CC0 material under this site's CC BY 4.0 is legally sound — but this site's pages cannot make the original more restrictive than it already is, which is why every provenance block states the source licence explicitly rather than defaulting silently to CC BY 4.0. The cleanest resolution, and the one this site argues for: align docs.diniscruz.ai to CC BY 4.0 going forward, leaving already-published articles as CC0. A site about attribution that dedicates its own foundational work to the public domain with no attribution requirement is making an argument about itself it may not intend to make.
What was redacted from this site, and why
A small set of source material was deliberately excluded rather than republished: a named venture-capital firm and a planned meeting; a live product's B2B pricing list; an internal investor review; a source-protection design that would endanger a real source if published as a working claim before it exists. The full policy is on the documents page. This is disclosed here because a participant's disclosure page should say what was left out, not only what was included.