> Proposed terms for reusing Hadal measurement artifacts and dataset descriptors: CC BY 4.0, how to attribute, and the limits. Not yet ratified. - Canonical: https://hadalinstruments.com/licence/ --- Legal # Data licence and reuse terms These are the terms proposed to govern reuse of the measurement artifacts Hadal publishes — the hashed raw files behind a figure, the dataset descriptors that document them, and the proof bundles that let you recompute a result. They were published before the corpus existed, so that anyone deciding whether to build on it could read the terms first rather than discover them afterwards. **Eight hash-verified measurement artifacts are published on this site today**, each offered beside the figure it evidences with its digest printed there. The terms below are the licence proposed to govern their reuse, and they are **proposed, not ratified**. Ratifying a licence is a decision I have not yet taken, so no reuse terms have yet been granted. ## What the licence will cover Four classes of published material: - **Raw measurement artifacts.** The content-hashed files a published figure was computed from — the bytes, not a summary of them. - **Dataset descriptors.** Schemas, field definitions, units, collection windows and the provenance record attached to each artifact. - **Proof-corpus bundles.** The recomputation packages that let a third party re-derive a published result independently, which is the only reason any of this is worth downloading. - **The machine-readable site surfaces** —`/llms.txt`,`/rag-corpus.jsonl` and `/sparql-dump.ttl` — which exist today and are already opened to every crawler group in `/robots.txt`. Bringing them under an explicit licence rather than an implied permission is part of this proposal. ## What it will not cover A permissive data licence is not a general permission slip, and the edges are stated here rather than left for a dispute. - **The instrument software.** Licences to run a Hadal instrument are commercial and separate; see [refunds and cancellation](https://hadalinstruments.com/refunds/) for the terms of sale. The software licence agreement itself is NOT YET PUBLISHED . - **Trade marks.** The Hadal name and marks are not licensed by anything on this page. Public licences of the Creative Commons family say so explicitly, and it is repeated here because people miss it: you may state that data came from us, and you may not present your work as ours. - **Editorial content.** Research articles, glossary entries and documentation are not part of the artifact corpus. Where a page grants reuse of its own text, it will say so on the page. - **Third-party data we do not own.** Where a measurement derives from a venue or vendor feed whose redistribution is restricted, the artifact will carry a derived-data notice and that restriction travels with it. We will not sub-licence what we have no right to licence, and an artifact we cannot publish cleanly will not be published at all. ## The proposed licences, by name Named, off-the-shelf public licences — not terms invented here. A bespoke licence would be unreadable, incompatible with the tools that consume open data, and impossible for a machine to check. A named one has a canonical text you can read without asking us what we meant. - **Measurement artifacts and data** — CC BY 4.0 — PROPOSED - **Schemas and field definitions** — CC0 1.0 — PROPOSED - **Ratification** — NOT YET RATIFIED ### Data and artifacts: Creative Commons Attribution 4.0 [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) permits copying, redistribution, adaptation and commercial use, on one condition: credit the source, link the licence, and say if you changed anything. It is the least restrictive licence that still keeps the trail back to the measurement intact — which matters more here than it does for most data, because a Hadal figure is only worth anything alongside the method that produced it. ### Schemas and definitions: CC0 1.0 Field definitions and schemas are proposed under [CC0 1.0](https://creativecommons.org/publicdomain/zero/1.0/), which waives rights entirely. A schema is a vocabulary. Attaching an attribution obligation to a vocabulary discourages the exact thing a vocabulary is for — being adopted, embedded and forgotten about. ### Why this page proposes rather than declares A public licence, once applied to a file someone has downloaded, cannot be withdrawn from that copy. It is a one-way decision binding every future user of the corpus, so it is mine to take deliberately rather than the page's to imply. Until ratification, no reuse grant attaches to the artifacts already published — a download made today is made ahead of the licence, not under it, and ratification will apply forward rather than being backdated. ## How to attribute, when there is something to attribute CC BY asks for the title, the source, the licence and a note of any changes. The form below is what a Hadal artifact citation will look like. It is an **EXAMPLE FORMAT**: every bracketed field is a placeholder, no artifact with these fields exists, and no digest is printed here because there is no file to have digested. ``` Hadal Instruments, [artifact title] ([artifact identifier]). Retrieved from hadalinstruments.com. Licensed under CC BY 4.0. Content digest: [sha256 as published with the artifact]. Modifications: [none, or what you changed]. ``` Carrying the digest through is not a formality. It is what lets your reader confirm they hold the same bytes you analysed, which is the entire reason the artifacts are hashed in the first place. ## What the licence does not give you Permission to reuse is not agreement with what you do next. The bounds are part of the offer. - **No warranty.** CC BY 4.0 is offered as-is and disclaims warranties so far as the law allows. We are not going to imply otherwise on the way past. - **No endorsement.** Attribution names the source. It does not say we agree with your analysis, and CC BY expressly forbids suggesting that we do. If you publish a conclusion drawn from our data, the conclusion is yours. - **The method travels with the data.** Every artifact publishes alongside its pre-registered methodology and its stated limits. Strip those and you are left with numbers detached from what they measured — permitted by the licence, and the fastest way to be wrong in public. - **Removing the digest destroys the point.** Republishing an artifact without its content hash is allowed and is a poor idea: the hash is the only property that made the file worth more than a screenshot. - **A licence is not a guarantee of fitness.** The data describes what was measured, under the conditions stated. It does not certify that the measurement answers your question. ## The state of the corpus, stated exactly The [methodology page](https://hadalinstruments.com/methodology/) sets out what has to be true before an artifact publishes; the corpus lives on the [research pages](https://hadalinstruments.com/research/) today, and the [observatory](https://hadalinstruments.com/observatory/) is where broker portraits will land. - **Artifacts available for download** — 8 - **Dataset descriptors published** — ONE PER ARTIFACT — beside the file it documents - **Proof bundles published** — NONE - **Licence status** — PROPOSED · NOT YET RATIFIED - **Machine-readable licence declaration** — NOT YET PUBLISHED - **Independent legal review** — NOT YET PUBLISHED - **Terms drafted** — 2026-08-01 When the first artifact publishes, its licence will also be declared where a machine can read it — in the dataset descriptor and in the page metadata — so an automated consumer never has to infer permission from prose. That declaration does not exist yet, which is why the row above says so instead of describing one. [Return home](https://hadalinstruments.com/)[Methodology](https://hadalinstruments.com/methodology/)