# Licensing One policy across the research programme: the seven Greek market and agent papers, the twelve LLM measurement papers, and the data and code published with them. Version 1.0, 15 August 2026. Author and rights holder: **Nikolaos Broikos**, Athens, Greece. Contact: broikos.nikolaos@gmail.com **Scope.** Research only. Nothing on this page applies to any other work in this folder. The papers already carry a licence line in their text. This file is what the licence line points at: the actual grant, its exact scope, the parts it deliberately does not cover, and the notice a reuser is expected to reproduce. --- ## The grant, by asset family | Family | What it is | Licence | |---|---|---| | Research papers | The seven Greek market and agent papers, the twelve LLM measurement papers, in every format they are published in: markdown, HTML, PDF | **CC BY 4.0** | | Research data | Claim ledgers, source logs, census files, appraisal grids, benchmark result files, trial data, extracted statistical tables | **CC BY 4.0** | | Research code | Crawlers, graders, gates, runners, analysis and extraction scripts, harness adapters, build and audit scripts | **MIT** | | Charts and diagrams | Exhibits generated by the published scripts | **CC BY 4.0** | Full texts: [`LICENSE-CC-BY-4.0.txt`](LICENSE-CC-BY-4.0.txt) (canonical Creative Commons legal code, verbatim) and [`LICENSE-MIT.txt`](LICENSE-MIT.txt). **Why code is not CC BY.** Creative Commons itself recommends against applying its licences to software, because they carry no patent grant and no source availability terms. Splitting text and data from code costs nothing and removes the ambiguity a developer would otherwise have to resolve before reusing a grader. --- ## The notice to reproduce Anybody reusing the papers or the data satisfies attribution with this: > Nikolaos Broikos, *[paper title]*, version [n], [date]. Published at > https://broikos.gr/research/ under CC BY 4.0. Add the DOI once a Zenodo deposit exists. Attribution must survive into derivative works, including charts redrawn from the published tables and figures quoted in a slide deck. --- ## What the grant does not cover The CC BY grant covers material authored here. It does not and cannot cover material belonging to somebody else that happens to sit in the same directory. See [`THIRD-PARTY-NOTICES.md`](THIRD-PARTY-NOTICES.md) for the itemised list. In summary, it excludes: 1. Third-party PDFs captured as evidence, including the courier tariff, the EU Official Journal texts, the ELSTAT, EETT and GRECA documents, and the captured arXiv paper. 2. Vendor documentation captured for coding, including the prompt engineering guidance snapshot in `llm-series/lab/captures/`. 3. Third-party source code and its own licence terms, including the harness adapters' upstream projects. 4. Statistical data whose publisher imposes its own reuse terms. Eurostat and EU material are reusable with attribution under Commission Decision 2011/833/EU; ELSTAT and EETT material is reproduced as facts and cited, not relicensed. 5. Trade marks. Product, model and company names appear for identification only. No endorsement is claimed and no trade mark licence is granted or received. **Facts are not licensed because facts are not copyrightable.** A price, a market share and a benchmark score can be quoted freely by anyone. What CC BY grants is the right to reuse the expression, the tables as compiled, and the database as arranged, which is the part that took the work. --- ## Withheld material, and why Not everything in this folder is publishable, and the withholding is deliberate rather than an oversight. These never leave this machine and belong in the first `.gitignore` entry if any of it is ever version controlled. | File or class | Reason | |---|---| | `whitepapers/02-greek-ecommerce-market/MERCHANT-KEY-PRIVATE.md` | Links merchant codes A, B and C to real businesses. Publishing it breaks a confidentiality undertaking and undoes the anonymisation of every merchant figure in the programme | | `whitepapers/ACCOUNTS-PRIVATE.md` | Account and access material | | Raw crawl output, `gr_raw2.jsonl` and equivalents | Contains the domain for every measured row. The published table is the anonymised derivative | | Absolute merchant revenue and order counts | Reported only as rates and indices, per the standing anonymisation rules | | Client and employer material | None is used in any paper. This line exists so the absence is on the record | --- ## Declarations that travel with every paper Each paper carries these already. They are restated here so the policy is in one place. - **Funding.** Unfunded. No sponsor, client or commissioning party paid for, commissioned, reviewed or approved any of this research. - **Conflict of interest.** The author operates e-commerce businesses in Greece and builds agent harnesses. Where that experience is the source of a paper's data it is also the conflict, and both are stated in the same place. - **Corrections.** Errors are corrected in the public change log with the date and the reason, never silently edited. - **Tooling.** The research was produced with AI assistance under the author's direction. The measurements, the instruments and the errors are the author's own, and every figure ships with the data and the code that produced it so a reader can check rather than trust. See the note below on what this means for copyright. --- ## The AI authorship question Worth stating plainly rather than leaving for somebody else to raise, and for a CC BY programme it changes nothing that matters. CC BY is a permissive grant. Even if a portion of the expression turned out not to be protectable, because the US Copyright Office requires human authorship and protects only the human-authored contributions in a work containing AI-generated material, the practical effect on a licence that already permits commercial reuse is nil. The attribution request stands on its own as a scholarly norm rather than as an enforcement threat. What is protectable here is not in doubt in any case: the selection of what to measure, the design of the instruments, the arrangement of the data, the coding schemes and the analysis. That is the part that took the work and it is human. --- ## Applying this Placement, once the site work happens: - `LICENSE-CC-BY-4.0.txt` and `LICENSE-MIT.txt` served at the research root. - Each paper's licence line links to the served copy, not only to creativecommons.org. - Each data directory carries a one line `LICENSE` pointer stating which of the two applies to which files in it, because a reuser lands on a CSV, not on this page. - `THIRD-PARTY-NOTICES.md` served alongside, linked from every licence line. - Machine readable: `` in the head, and `"license"` in the JSON-LD for each paper and each dataset.