Zenodo deposit license: CC-BY-NC-SA-4.0 set as the cautious default — open to CC-BY pending a source-license survey

[rdhyee]

🤖 rbotyee (Claude, operated by @rdhyee) — Raymond's intent (2026-08-27): set the deposit's license to the more restrictive option now, as the cautious move in the absence of firm guidance, and flag it here as open to relaxation. His words: "CC-BY is probably the right long-term choice but attaching SA-NC is for the cautious move in the absence of firm guidance." What I did: set Zenodo draft 21288719 ("iSamples Query Substrate — Snapshot 202601") to CC-BY-NC-SA-4.0 via the deposit API (was cc-by-4.0, a placeholder); opened this issue; started a survey of the four sources' data-use terms.

Why NC-SA, and why it is not settled

• The parent record — the April 2025 complete export, 15278211 — is CC-BY-NC-SA-4.0. It was deposited from Raymond's Zenodo account on 2025-04-25 while he was "waiting on Dave and others to tell me how to proceed in terms of license" (his journal, 2025-04-28). No recorded answer to that question has been found (email, Slack, notes). So NC-SA on the parent is a cautious placeholder, not a rights-holder requirement we know of.
• The February 2025 OpenContext-only export (14927593) is CC0.
• The records themselves carry no rights statement (dc_rights is NULL on all 6.68 M rows across SESAR, OpenContext, GEOME, Smithsonian).
• Rule applied: a derived deposit should not be more permissive than its source without authority from the rights holders. Matching the parent is the defensible floor today.

What would let us relax it

A survey of what each source actually permits for its metadata records (SESAR / IEDA, Open Context per-project licenses, GEOME, Smithsonian Open Access), then a decision by the PIs (@athomer, Kerstin). Survey results will be posted here. Zenodo allows editing the license of a published record without a new version, so publishing now under NC-SA does not lock this in; and since 15278211 is also on Raymond's account, the parent can be relaxed at the same time.

Related

• Zenodo deposition for iSamples query substrate — 202601 snapshot #139 (the deposition), Closeout task: add Andrea as Zenodo owners #147 (Andrea now has manage access on the draft).
• Bundled pqg code has no license file yet — README says "see LICENSE" but no LICENSE file exists — which license? pqg#25 (separate question: code, not data).

[rdhyee]

🤖 rbotyee (Claude, operated by @rdhyee) — Raymond's intent: survey what the four sources actually permit for their metadata, so the NC-SA-vs-CC-BY question can be decided on facts. What I did: delegated the web survey to a subagent (2026-08-27), spot-checked its key quotes live (SESAR footer, GBIF license fields for NMNH/GEOME datasets, Open Context IP page), and am posting the summary. Full report with URLs and verbatim quotes: SOURCE_LICENSE_SURVEY_2026-08-27.md in the coordination repo (local; will travel with the Zenodo bundle or a later PR).

Source (share of deposit)	Stated license for metadata	Confidence
Smithsonian NMNH (322 K, 5 %)	CC0 1.0 — NMNH's own Darwin Core resource page: "the publisher has waived all rights to these data and has dedicated them to the Public Domain (CC0 1.0)." GBIF lists NMNH Extant Specimen Records as CC0.	high
GEOME (606 K, 9 %)	CC0 for public metadata — Deck et al. 2018 (PLOS Biol): "To maximize open access, metadata are available under a Creative Commons Zero license (CC0)." GEOME's own policy page didn't render to the fetcher.	high (secondary source)
Open Context (1.11 M, 17 %)	Per project, contributor's choice — CC-BY 4.0 by default, CC0 allowed, and "Open Context does permit use of license variants that restrict commercial use" for culturally sensitive material. So the OC slice is not uniform; some projects may be NC.	high that it varies; the mix in our 1.1 M is unknown
SESAR / IGSN (4.69 M, 69 %)	Unclear. The only license text on geosamples.org is the site footer "Site licensed under Creative Commons Attribution-Noncommercial-Share Alike 3.0", which re3data also records as SESAR's data license. Nothing SESAR-authored says whether that binds sample metadata or just the website. Cutting the other way: IGSN metadata registered via DataCite is CC0 at the DataCite level.	low — this is the pivotal unknown
iSamples itself	No license asserted anywhere on the aggregate (consistent with dc_rights NULL on every row).	—

What it implies. CC0 sources (NMNH, GEOME) are compatible with anything. CC-BY-4.0 for the whole deposit is clean only if (a) SESAR's footer does not govern its metadata, and (b) the OC projects in our export are all CC-BY/CC0. Neither is established. NC-SA-4.0 is therefore the defensible floor today — which is what the draft now carries — and nothing found requires it either.

Two questions for people (in priority order):

1. SESAR / IEDA (info@geosamples.org, or Kerstin directly): does the CC BY-NC-SA 3.0 notice apply to sample metadata served via the SESAR API, or only to the website? This one answer decides 69 % of the deposit.
2. Open Context / @ekansa: Open Context's API exposes a per-project dc-terms:license; the iSamples export dropped it. A per-project license pull for the projects present in our 1.1 M OC records would settle the OC slice from data rather than by assumption.

Not done here: no one has been contacted; this comment is the material for that ask.