Literature review

Source file: docs/LITERATURE-REVIEW.md · updated 2026-11-08

v0.2 — 2 November 2026. English. Supersedes the "first search" status of LITERATURE-AND-NOVELTY.md.

0. Method and limits (read first)

About 20 web searches/fetches for Task A (cap ~40 for A+B together). Queries used Scholar/SSRN/arXiv-style phrasing plus think-tank names (CSET, Carnegie, DigiChina, ECFR, IFRI, SWP, Clingendael, Bruegel, CEPS, MERICS, Chatham House, OFE, Linux Foundation, OSI).
Read in full (text extracted): MERICS Primer 2021, OFE Country Intelligence Report 2023, House Select Committee letter (June 2025). Read via fetch-tool summary of the web page (not checked line by line): Jamestown Parts I and II (Part II page is ~750k characters; only the first 100k were read), China Policy substack, MERICS web page, arXiv 2606.15999 abstract (EconPapers), Gitee OSS-Compass blog.
Found but NOT read (snippet only): Yang 2025 (SAGE), MERICS AI-Stack 2025, OpenUK AI Openness Report July 2026 (PDF over 10 MB, fetch failed), Computing.co.uk / Amanda Brock 2026 (HTTP 403), Factsheet China 2023 (PDF text not extractable).
Not covered: Scopus/Web of Science; Google Scholar itself (only web-search proxies); journal-specific searches (China Quarterly, Research Policy, etc.). CSET, Carnegie, DigiChina, ECFR, IFRI, SWP, Clingendael, Bruegel, CEPS, Chatham House and Rhodium returned no relevant hit in the queries run. That is an absence in a limited search, not proof of absence.
Quotes are under 15 words and come from the fetched text.

1. Table of works

Relevance: 1 = peripheral, 5 = must cite / closest competitor.

#	Author, year	Type	URL	Lens	Main claims (short quotes)	Compares with EU?	Rel.
1	Sunny Cheung, Open-Source Technology and PRC National Strategy, Part I, Jamestown China Brief 24(10), 10 May 2024	Think tank (US)	https://jamestown.org/open-source-technology-and-prc-national-strategy-part-i	US security / tech containment	OpenAtom uses open source "as a strategic asset"; openEuler and OpenHarmony are self-reliance flagships; OpenAtom staff CCP ties (33.3% members in body text; summary says 45%+ with youth league); military-industrial complex as "key beneficiaries".	n. Only Europe mention: Open Source Summit Europe 2023.	4
2	Sunny Cheung, Part II, China Brief 24(11), 24 May 2024	Think tank (US)	https://jamestown.org/program/open-source-technology-and-prc-national-strategy-part-ii/	US security	OpenHarmony "widely used in critical infrastructure", public security and military; open source feeds closed military software; PRC "positioning itself as an early innovator in the regulation" of OSS. Notes a DLUT-Liège (Belgium) collaboration as a "Western institution" link.	n.	4
3	Rebecca Arcesati, Caroline Meinhardt (MERICS), China's Open-Source Tech Development: Primer, May 2021	Think tank (EU)	https://22.merics.org/sites/default/files/2021-05/MERICS%20Primer%20Open%20Source%202021_0.pdf ; web: https://merics.org/en/report/china-bets-open-source-technologies-boost-domestic-innovation	EU China-watcher, industrial policy	Open source as "an industrial policy tool" to reduce foreign reliance; timeline from Red Flag Linux (1999) to MIIT picking Gitee (Aug 2020); 87.4% of firms used OSS in 2019 (CAICT). Last page "Europe should closely watch": opportunities (talent, Baidu Apollo) and risks (military/surveillance reuse, export-control circumvention via RISC-V).	Partly. One page on EU opportunities/risks, cites EU OSS Strategy 2020-23; no policy-instrument comparison.	5
4	OpenForum Europe (Thévenet, O'Riordan, Karhu, Grzegorzewska) for DG DIGIT / Interoperable Europe, OSS Country Intelligence Report: China, July 2023	EU institutional study	https://interoperable-europe.ec.europa.eu/sites/default/files/inline-files/OSS%20Country%20Intelligence%20Report%20China.pdf	EU descriptive country profile	"In comparison to European policies, the Chinese government primarily focuses on" industrial support. States there are no formal procurement or reuse policies specific to OSS in Chinese public administrations (paraphrase). Sections: actors, policy/legal framework, initiatives. Desk research, snapshot July 2023.	Partly. One framing sentence in the executive summary; no instrument-level mapping.	5
5	Axel Thévenet, Factsheet China, Oct 2023 (PDF metadata)	EU institutional	https://interoperable-europe.ec.europa.eu/sites/default/files/inline-files/Factsheet%20China.pdf	EU descriptive	Short companion. Text not read.	Unknown	2
6	US House Select Committee on the CCP, letter to Eclipse Foundation (M. Milinkovich), 4 June 2025	Legislative letter	https://democrats-selectcommitteeontheccp.house.gov/sites/evo-subsites/democrats-selectcommitteeontheccp.house.gov/files/evo-media-document/20250604-sccletter-eclipsefoundation-harmonyos_0.pdf	US security	Concern about Eclipse-OpenAtom collaboration; cites Eclipse goal to "extend[] OpenHarmony code base with add-ons for the European and Global markets"; Huawei's code makes independent verification "a monumental, if not impossible, task". Six questions (audit, governance, PRC National Intelligence Law, due diligence, genesis of Oniro); reply due 18 June 2025.	n. (addresses a Brussels-based foundation; no EU policy analysis). No Eclipse reply found.	5
7	CHINA POLICY (Substack), Institutionalising open source in China, 16 June 2026	Practitioner newsletter	https://chinapolicy.substack.com/p/institutionalising-open-source-in	Chinese governance	"Now Beijing wants to govern the ecosystem that made that possible." Three layers: CAC filing, voluntary standards (GB/T 44272-2024), mandatory sector rules; OpenAtom "is not a regulator, but it makes the regulatory architecture work"; risk: "Heavy compliance burdens push developers elsewhere".	n. One passing EU reference.	4
8	Grace X. Yang, The openness paradox: Open-source AI and China's quest for cyber sovereignty, SAGE, 2025 (DOI 10.1177/29768640251376497)	Peer-reviewed, 4 pp.	https://refubium.fu-berlin.de/handle/fub188/50926	Academic, sovereignty	Open-source AI seems to conflict with "cyber sovereignty" doctrine yet can strengthen it across layers. Snippet only.	Unknown (likely n.)	3
9	Wang Jin, Kunievsky, Lou, Sun, Evans, U.S. Technological Containment and the Rise of China's Open AI Ecosystem, arXiv 2606.15999, June 2026 (rev. Aug 2026)	Preprint	https://econpapers.repec.org/paper/arxpapers/2606.15999.htm	Empirical, US-China	After export-control shocks China built open-source AI into national technology strategy; Chinese developers engage with open LLM repos more than US ones.	n. per abstract	3
10	Yehui Wang (Huawei / OSS-Compass), Daniel Izquierdo Cortázar (Bitergia), Europe-China Open Source: Digital Sovereignty & Sustained Collaboration, keynote, OSS Europe, Amsterdam, 27 Aug 2025	Industry keynote with data	https://compass.gitee.com/blog/2025/08/29/OSS-EU2025 ; https://osseu2025.sched.com/event/25VuY/europe-china-open-source-digital-sovereignty-sustained-collaboration-yehui-wang-huawei-daniel-izquierdo-cortazar-bitergia	Industry, complementarity	Proposes "a complementary ecosystem market cooperation path"; Europe 92.53M commits vs China 50.34M (OpenAtom 2024 report data); Europe leads in mature fields, China in AI/semiconductors.	YES, quantitative and industry-side. No policy-instrument analysis; a speaker is Huawei-affiliated. Closest to our complementarity lens.	4
11	OpenUK, AI Openness Report, July 2026	UK industry body	https://openuk.uk/wp-content/uploads/2026/07/AI-Openness-Report-July-2026.pdf	UK, AI openness	Snippet: has a China technology-and-policy section, names OpenAtom, frames open source as sovereignty instrument. Unread; must check before publication.	Unknown	3 (flag)
12	Amanda Brock (OpenUK), Can the UK take the opportunity open to it in China?, Computing, 2026	Opinion	https://www.computing.co.uk/opinion/2026/can-the-uk-take-the-opportunity-open-to-it-in-china	UK practitioner	Snippet: OpenAtom let China hold IP and govern projects locally. HTTP 403; unread.	Partly (UK)	2
13	MERICS, China's drive toward self-reliance in AI, July 2025	Think tank (EU)	https://merics.org/sites/default/files/2025-07/MERICS%20Report-AI_Stack_final.pdf	EU China-watcher, AI	Snippet: self-reliance at every layer of the AI stack, yet tied to global open-source communities. Unread.	Unknown	2
14	Kevin Xu, Interconnected	Newsletter	https://interconnected.blog/can-you-nationalize-open-source/	Practitioner	From prior note; not re-read.	n.	3
15	Doug Levin	Newsletter	https://douglevin.substack.com/p/chinas-open-source-strategy-innovation	US security	From prior note; not re-read.	n.	2
16	Centre for Future Generations, Towards European open digital ecosystems	Think tank (EU)	https://cfg.eu/towards-european-open-digital-ecosystems/	EU, dependency on Chinese open weights	From prior note; not re-read.	Partly (EU side)	3
17	Bulletin of the Chinese Academy of Sciences 36(12), art. 15	Chinese journal	https://bulletinofcas.researchcommons.org/journal/vol36/iss12/15	Chinese academic	Snippet: reviews US and European OSS policy to propose steps for China (reverse direction to ours). Unread.	EU as object	3
18	ASPI Strategist, China's use of foreign open-source software and how to counter it	Think tank (AUS)	https://www.aspistrategist.org.au/chinas-use-of-foreign-open-source-software-and-how-to-counter-it/	Security	From prior note; not re-read.	n.	2

Also seen, unread: Linux Foundation Europe, Open Is Sovereign (names PaddlePaddle and openEuler as state-backed; https://linuxfoundation.eu/newsroom/open-is-sovereign-why-europes-digital-future-must-be-built-on-global-open-source-ecosystems); SESEC webinar on China open-source policies and standardization, 2023 (https://sesec.eu/wp-content/uploads/2023/07/SESEC-V-Webinar-China-Open-source-Policies-Standardization.pdf), worth reading.

2. Was the OFE / Interoperable Europe China report updated?

No update found. Report dated July 2023; Factsheet metadata 16 Oct 2023. Searches for a 2024-26 update returned nothing. Re-check directly on the Interoperable Europe / OSOR site before asserting this publicly.

3. Who comes closest to our work

OFE 2023 and MERICS 2021 are the European precedents: descriptive, pre-dating the EU Open Source Strategy (June 2026) and DC-EDIC, with no mapping of Chinese instruments onto an EU instrument set.
Huawei/Bitergia keynote (Aug 2025) is the only item found that sets Europe and China side by side and proposes complementarity. Data-led, industry-authored, not policy analysis.
House letter (June 2025) targets the Eclipse-OpenAtom link we use as a governance case. We must answer it.

4. Novelty verdict

What has been done (evidence above):

Security-led readings of OpenAtom, OpenHarmony and openEuler (Jamestown, House, Reuters, ASPI).
A 2021 EU-side primer with a one-page "Europe should watch" section, and a 2023 EU-commissioned descriptive country profile.
Chinese-governance analysis (China Policy 2026; Yang 2025) and US-China empirical work on open AI (arXiv 2026).
An industry-side Europe vs China contribution comparison with a complementarity framing (OSS EU 2025).

What we did not find, in a limited search:

Any mapping of Chinese open-source instruments onto the DC-EDIC Implementation Strategy (adopted 16 Oct 2026, so near-certain).
Any work scoring fit, feasibility and exposure separately, or using additionality as comparison metric.
Any instrument-level reading of the 2025-26 regional policies (Beijing, Shanghai, Zhejiang, Jiangsu, Guangdong/Shenzhen, Chongqing) in a Western language outside our own register.
Eclipse-OpenAtom treated as a governance case rather than a security alarm.
An update of OFE 2023.

Claims in LITERATURE-AND-NOVELTY.md to soften:

Claim 2 ("a complementarity lens", "no work scores...") is partly undermined by item 10 (explicit, numeric complementarity) and by MERICS 2021's opportunities/risks page. Rephrase: "first policy-instrument-level complementarity assessment with separate fit, feasibility and exposure scoring".
Claim 4 (funding models side by side) is not contradicted, but cite MERICS 2021, which placed the EU OSS Strategy 2020-23 next to Chinese policy.
"First independent assessment" should read "to our knowledge, in a limited review" until the unread items (OpenUK 2026, Yang 2025, Brock 2026, MERICS 2025, Factsheet 2023, SESEC 2023) are read.

Could undermine novelty if they contain more than their snippets show: OpenUK AI Openness Report (July 2026; China section plus a UK/European frame) and the SESEC 2023 webinar. Both unread.

Honest gaps: no academic-database access; peer-reviewed literature thinly sampled (one hit); Chinese-language and CNKI literature not searched; no direct site searches of Carnegie, CSET, Chatham House; Jamestown Part II mostly unread beyond its first 100k characters.

5. Academic databases — PRISMA-style log

Round 1 — OpenAlex, 4 Nov 2026 (R02)

Identification: OpenAlex API, works published since 1 Jan 2019, 9 queries. Q1 "open source" China Europe policy (77); Q2 "digital commons" China (73); Q3 openEuler/OpenHarmony/openKylin (137); Q4 "open source" "digital sovereignty" (152); Q5 "open source" China government industrial policy (21); Q6 开源政策 (1260, failed: no tokenisation); Q7 title "open source" China (108); Q8 Gitee/OpenAtom (70); Q9 title "open source" "public sector" Europe (1). The top 12 per query by citations are in data/lit/openalex_2026-11-04.md.
Screening: 108 titles screened. About 100 were excluded as off-topic (OSINT, open data, medical AI, ecology, and Gitee as a mere code host).
Included (7):

Work	Relevance	Why
Hu et al. (?), openEuler: Advancing a Hardware and Software Application Ecosystem, IEEE Software 2022, doi 10.1109/ms.2021.3132138	3	Peer-reviewed self-description of openEuler governance and ecosystem (authors to check)
Software Engineering for OpenHarmony: A Research Roadmap, ACM Computing Surveys 2025, doi 10.1145/3720538	3	Research agenda; shows OpenHarmony as an academic object
Open or closing doors? The influence of 'digital sovereignty' in the EU's Cybersecurity Strategy on cybersecurity of open-source software, CLSR 2024, doi 10.1016/j.clsr.2024.1060…	4	EU sovereignty framing applied to OSS: the EU half of our comparison
OECD, Digital public goods: Enablers of digital sovereignty, Development Co-operation Report 2021, doi 10.1787/c023cb2e-en	3	Digital-commons/DPG framing outside both blocs
Public-private funding models in OSS development: scikit-learn, arXiv 2404.06484	3	Funding chapter: a European project-level case
Open to open-source AI? AI model choice in public sector agencies, GIQ 2026, doi 10.1016/j.giq.2026.102133	3	PA adoption of open-weight models (could include Qwen/DeepSeek)
The European Public Sector Open Source Opportunity, 2023, doi 10.70828/dfnp4537	3	EU PA reuse; publisher to identify (likely OpenForum Europe)

Result (inference): none of the indexed peer-reviewed or grey works compares EU and Chinese open-source policy, which strengthens the novelty verdict in §4 (one database only). Chinese-language academic coverage still needs CNKI or Wanfang (VPN or a library).
Next rounds: Semantic Scholar API (same queries), SSRN, Google Scholar by hand (Chrome), journal sweeps (China Quarterly, Research Policy, Telecommunications Policy, Internet Policy Review), and CNKI via VPN.

Round 2 — Semantic Scholar, 5 Nov 2026 (R02) — partial

Identification: Semantic Scholar Graph API (no key), works since 2019, 8 relevance-ranked queries, top 15 each (tools/s2_query.js; raw output in data/lit/s2_2026-11-05.md). Only 2 of 8 queries returned (S5 "China open source industrial policy government": 13,633; S7 "Gitee OpenAtom open source foundation China": 273). S1–S4, S6 and S8 were refused (HTTP 429) from the laptop and again from the Paris server, even with 15–75 s backoff: the keyless pool is shared and saturated.
Screening: 30 titles screened. Excluded: ML model releases (FireRedASR, VITA, YuLan…), open government data, generic industrial policy (Made in China 2025, semiconductors), life-cycle databases.
Included (4):

Work	Relevance	Why
Open source digital sovereignty: the case for a Canadian national Linux distribution, Digital Policy, Regulation and Governance 2026, doi 10.1108/dprg-06-2026-0347	3	Sovereignty-through-OSS argument from a third country; a natural comparator for openEuler/openKylin as "national distributions"
Open-Source AI Made in the EU: Why it is a Good Idea, Minds and Machines 2025, doi 10.1007/s11023-025-09728-x	3	EU open-weight AI rationale: the EU half of the AI layer comparison
Open-Source Innovation for Latecomer Catch-up: A Conceptual Framework, Journal of Economic Surveys 2026, doi 10.1111/joes.70122	3	Catch-up framing that fits the Chinese state-led OSS strategy (to read)
Open Source Congress 2024: Shaping the Future of Collaboration in AI, Security, and Digital Public Goods, 2024, doi 10.70828/mton6557	2	Grey literature (same DOI prefix as the 2023 European public-sector report, likely LF Research); digital public goods framing

Result (inference): still 0 works comparing EU and Chinese open-source policy (cumulative: 2 databases, 11 included). The novelty verdict holds, with the caveat that 6 Semantic Scholar queries are pending.
Re-run: node tools/s2_query.js S1 S2 S3 S4 S6 S8 at a quiet hour (European night) or with a free API key (request form on semanticscholar.org, institutional email needed, so Fabio decides).

Round 3 — Semantic Scholar, 6 Nov 2026 (R02) — partial

Identification: re-run of the 6 pending queries (tools/s2_query.js S1 S2 S3 S4 S6 S8; raw output in data/lit/s2_2026-11-06.md), ~09:30 CET. Only S4 "open source software digital sovereignty" returned (4,019 hits, top 15 read); the other 5 were again refused (HTTP 429) despite 15–75 s backoff.
Screening: 15 titles screened. Excluded: indigenous/civil-society sovereignty, logistics, metaOS, supply-chain security theory (kept as context), commercial-OSS market reports.
Included (4, new):

Work	Relevance	Why
Digital Sovereignty and Open-Source Software — A Discussion Paper, I4CS 2024, doi 10.1007/978-3-031-60433-1_22	3	Direct conceptual treatment of our central pairing
2026 State of Open Source in Europe: Navigating AI and Digital Sovereignty Through Open Governance, 2026, doi 10.70828/olta8310	3	Grey literature (LF Research prefix); EU-side state of play, same year as our white paper
A Toolkit for Measuring the Impacts of Public Funding on Open Source Software Development, arXiv 2411.06027	3	Methodology for the funding chapter's additionality metric
Public Sector Open Source Software Projects — How is development organized?, arXiv 2304.06093	3	PA-reuse chapter: how government OSS projects are actually run

Already included earlier and re-confirmed by this query: CLSR 2024 "Open or closing doors?"; scikit-learn funding case (arXiv 2404.06484); Canadian national Linux distribution (dprg-06-2026-0347).
Result (inference): cumulative 3 rounds, 2 databases, 15 included, still 0 works comparing EU and Chinese open-source policy. The novelty verdict holds; 5 S2 queries still pending.
Re-run: same command at a quieter hour; or the free S2 API key (Fabio, institutional email); then SSRN and Google Scholar by hand.

Round 4 — arXiv API, 7 Nov 2026 (R02) — new database

Identification: arXiv API (export.arxiv.org, keyless, no throttling), 6 field-restricted queries on abstracts (tools/arxiv_query.js; raw output in data/lit/arxiv_2026-11-07.md), sorted by relevance, 2019+, top 15 each. Hits: A1 China policy 9; A2 Europe policy/sovereignty 5; A3 openEuler/OpenHarmony/openKylin/Gitee/OpenAtom 21; A4 "digital commons" 15; A5 geopolitics/export control/sanctions 26; A6 public sector/administration/municipal 41.
Screening: 83 records screened by title (+2 abstracts read). Excluded: LLM censorship/language papers, urban economics, robotics, physics tools, OpenHarmony static-analysis/testing tools (kept as one representative), CentOS-like OS fault injection (reliability, not policy), OSINT/geopolitical-risk ML.
Included (10, new):

Work	Relevance	Why
Software Engineering for OpenHarmony: A Research Roadmap, arXiv 2311.01311	3	Only academic overview of the OpenHarmony ecosystem; CN industrial-stack chapter
The Impact of Sanctions on GitHub Developers and Activities, arXiv 2404.05489	3	Empirical evidence on platform access under sanctions; bears on Gitee/AtomGit vs GitHub and the "openness under geopolitics" argument
Open Source Software in the Public Sector: 25 years and still in its infancy, arXiv 2308.04766	3	EU PA-reuse baseline review
Soft-lockins in Public Sector Acquisitions of OSS-solutions: a municipal e-service platform, arXiv 2409.01118	3	Procurement mechanics — compare with our CN finding that XC specs are brand-neutral
Public Sector Open Source Program Offices — Archetypes, arXiv 2603.04891	3	EU public-sector OSPO typology (2026), maps onto the DC-EDIC governance question
Advancing Digital Government: Integrating OSS Enablement Indicators in Maturity Indexes, arXiv 2510.04603	3	Indicator design for the observatory's own scoreboard
The Geography of Open Source Software: Evidence from GitHub, arXiv 2107.03200	2	Country-level OSS activity incl. China and EU; GitHub-only bias noted
Generative AI and the Future of the Digital Commons: Five Open Questions, arXiv 2508.06470	2	Digital-commons framing for open-weight models
Sustaining Open Data as a Digital Common — CPR design principles, arXiv 2208.01694	2	Ostrom-style governance lens, usable for data-space comparison (Catena-X ↔ CAAM)
OpenPerf: benchmarking the sustainable development of the open-source ecosystem, arXiv 2311.15212	2	Chinese-authored ecosystem metrics (X-lab / OpenRank lineage) used in our CN-OSS-LANDSCAPE

Already included: arXiv 2606.15999 (U.S. containment and China's open AI ecosystem), 2304.06093 (public-sector OSS projects).
Result (inference): cumulative 4 rounds, 3 databases, **25 included, still 0 works that compare EU and Chinese open-source policy**. The closest are geopolitics (2606.15999, 2404.05489) and single-side public-sector studies. Novelty verdict holds.
Negative (✖): arXiv has no Chinese-language policy literature; A1/A2 show that "open source + policy" is barely an arXiv topic (14 hits in total) → policy literature must come from SSRN, journals and CNKI.
Re-run: node tools/arxiv_query.js (edit the qs array); add cat:cs.CY restriction for policy-only queries.
Semantic Scholar, same day (⊘): S1 S2 S3 S6 S8 re-run ~10:00 CET → all 5 refused (HTTP 429) after the full 15–75 s backoff (data/lit/s2_2026-11-07.md). Fourth day of keyless throttling: the keyless route is not workable; a free S2 API key (requested with an institutional email) is needed, or replace the 5 queries with OpenAlex title.search equivalents.