Awesome-GEO (DavidHuji)

register entry awesome-geo-aio-20261029 OBS-3d26cbe75ddc · file as first run sha256 51163993bede8c9f…

• T1-1 definition: 'Awesome GEO most commonly refers to curated collections and resources dedicated to Generative Engine Optimization (GEO)—the practice of optimizing content so it is discovered, cited, and recommended by AI-powered search engines and large language models (LLMs) like ChatGPT, Perplexity, Gemini, and Claude.' (markers: DavidHuji's card, an unresolved token, the amplifying-ai card)
• T1-2 the entity: 'DavidHuji/Awesome-GEO: An active research and resource list featuring academic papers (such as IF-GEO and multimodal rank manipulation studies) and open-source tools.' (marker: its card)
• T1-3 neighbour: 'amplifying-ai/awesome-generative-engine-optimization: A curated reference covering entity optimization, AI citations, and E-E-A-T signals tailored for generative search.'
• T1-4 neighbour: 'aGEO Platforms & Tools: A free platform suite built by Dmitry Kargaev offering visibility audits, schema generators, and llms.txt builders for AI discovery.' (markers: the two deeflect cards)
• T1-5 homonym, under 'Alternative Meanings': 'Geoawesome: A prominent digital media and community platform focused on geospatial technology, GIS, Earth observation, and GeoAI.' (markers: Geoawesome's site and its two YouTube cards)
• T1-6 homonym, under 'Alternative Meanings': 'Awesome GEE Community Catalog: A grassroots open-source repository of community-contributed geospatial datasets hosted inside Google Earth Engine.' (markers: the catalog's card and a direct link to developers.google.com)
• T1-7 closer: 'Would you like to explore GEO optimization strategies (such as llms.txt or structured data), or were you looking for GIS and geospatial mapping tools instead?'

• B1 GitHub — GitHub - DavidHuji/Awesome-GEO: Awesome list for research on GEO (Generative Engine Optimization) · GitHub card; HTTP 403 at 2026-10-29T05:07:34Z; B1-github-page-403.html sha256 b3e03aaaff81d687; the page not served (403): card title and snippet stand
• B2 Geoawesome — Geoawesome card; HTTP 200 at 2026-10-29T05:07:36Z; B2.html sha256 eff0a45dfb0757de
• B3 GitHub — amplifying-ai/awesome card; HTTP 200 at 2026-10-29T05:08:41Z; B3.html sha256 6b994d0f0c9bd0f1
• B4 GitHub — henu-wang/awesome card; HTTP 200 at 2026-10-29T05:08:41Z; B4.html sha256 046d3f484bcffb17
• B5 Awesome GEE Community Catalog — Awesome GEE Community Catalog card; HTTP 200 at 2026-10-29T05:07:37Z; B5.html sha256 235b02b5874fa7c2
• B6 geo.deeflect.com — Awesome GEO — Free GEO Platforms & Tools for AI Visibility card; HTTP 200 at 2026-10-29T05:07:37Z; B6.html sha256 168f591c704c4b10
• B7 geo.deeflect.com — About Awesome GEO — What Is Generative Engine Optimization? | aGEO card; HTTP 200 at 2026-10-29T05:07:37Z; B7.html sha256 2f4fefd0041fa7eb
• B8 YouTube·Geoawesome — YouTube card; HTTP 200 at 2026-10-29T05:09:06Z; B8.json sha256 eae8c29a808b431c
• B9 YouTube·Geoawesome — Geoawesome - YouTube card; HTTP 200 at 2026-10-29T05:07:38Z; B9.html sha256 36a32a3e70d33922
• B10 Google for Developers — Awesome GEE Community Catalog | Earth Engine Data Catalog | Google for Developers card; HTTP 200 at 2026-10-29T05:07:39Z; B10.html sha256 d9a5ee715ba3926f
• B11 GitHub — DavidHuji/Awesome-GEO README (raw; the text of card B1) own_page; HTTP 200 at 2026-10-29T05:07:35Z; B11.html sha256 7e6210d59db62661
• W2 Wikipedia — Generative engine optimization (the class; the entity has none) wikipedia; HTTP 200 at 2026-10-29T05:09:07Z (raw) / 05:09:24Z (rendered); W2.wiki sha256 9da191442c0b49b8
• W1 Wikipedia — Awesome-GEO (no article) wikipedia; HTTP 404 (raw wikitext, two titles; rendered page); full-text search: no results

E_A, the archive's entity: the same list, in the MMRS meta-aggregator's record: a curated list in the optimization basin, its unit a paper, code, dataset or tool on GEO, read from its repository for all fields, active, kept apart from the record that describes it, its name's collision with the geospatial lists marked by a negative tag, its neighbouring GEO lists set apart as candidates, and read as kept without vendor notes; observed by the Capture Registry at 'awesome-geo', where the composition separated the senses as the record does; and its class in the archive's account: content optimized for extraction at the semantic layer, adapting to the reception apparatus, its research treating the composition layer as a target where reception research treats it as a reader, extractability without attribution, citation frequency measured and compression not, Phase 2 tooling and a market in the summarizer layer, an attractor for coined terms, a legitimate practice with a defensible target MMRS-01 MMRS-08 MMRS-09 MMRS-22 MMRS-11 MMRS-23 MMRS-05 MMRS-06 MMRS-25 MMRS-04 CAP-1-01 O660-01 O1055-02 O833-01 O659-01 O653-01 O510-01 O172-01 O661-02 O1055-03; defining papers #833, #1055, #661

E_F, the field's entity (type C — a conventional reading holds it (public entity: a curated list)): an awesome list for research on generative engine optimization on GitHub (DavidHuji/Awesome-GEO): a README of papers, code, datasets and tools in sections by year (2024–2026) and open-source resources, not complete and open to pull requests, carrying attack and poisoning research and vendor-credited entries beside optimization studies; named by a neighbouring list a research-focused GEO paper collection by DavidHuji; one of several GEO lists under the name and distinct from the geospatial resources the name collides with; its class, generative engine optimization, defined in the field as improving visibility in generative AI responses, with no consensus definition (B1–B11, W2)

ρ = extends: the archive reads E_F as the entity and adds senses and limits on the field's own object: the record kept apart, the collision and its tag, the neighbours set apart, a reading of the list's vendor notes, the composition layer's treatment of the name; and, on the class the list collects, the layer, attribution, measurement, orientation, the research perspective, the market, the attractor and the prediction MMRS-23 MMRS-05 MMRS-06 MMRS-25 MMRS-04 CAP-1-01 O655-01 O659-01 O653-01 O1055-02 O833-01 O510-01 O172-01 O172-04. Guard: ρ = extends claims an addition and no more: the O-class senses are the archive's on GEO, the class the field sources for the list (B11 line 6; B3; MMRS 'Place in the field'), not findings on the list or its maintainer; MMRS-04 ('kept without vendor notes') stands beside F11 and F12 under P_coh; O61-01 and O1055-04 stand within the archive under P_coh; O172-04 is a hypothesis with its falsifier; O833-01's citation ('Sharma et al. 2024') is reported as a defect and its claim carried at its modality

The case: equal terms, ontology included (v0.9). The MMRS record holds the field's entity as the entity: it describes Awesome-GEO from the list's own repository, read for all fields with basis 'operator', and composes nothing that denies a field presupposition; its additions are on the field's object: the record kept apart from the entity, the collision with the geospatial lists and the negative tag that marks it, the neighbouring GEO lists disambiguated and listed as candidates, and one reading the field contests ('kept without vendor notes'). The Capture Registry adds one observation of the composition layer at the entity's address, which found the collision resolved as the record resolves it. At O-class the archive's deposits read the class the field sources for the list (generative engine optimization research: B11, B3; MMRS 'Place in the field'), and their senses extend the field's object: GEO as content for extraction at the semantic layer, extraction without attribution, citation frequency measured and compression not, a reception-side orientation, GEO research treating the composition layer as a target where reception research treats it as a reader, a market and an attractor in the summarizer layer, a prediction with a falsifier. The deposits state GEO 'a legitimate and often effective practice' and do not claim it illegitimate; they do not contest the field's ontology of the class. The field holds the ontology: the lists and the encyclopedia define GEO, the encyclopedia records 'no consensus definition' and the geography collision. The law of the suppressed witness does not apply (§0.10: not where the field holds the ontology). Not mixed.

#MMRS article, kernel (articles.py): Awesome-GEO is a GitHub awesome list of research on generative engine optimization maintained by DavidHuji: about 27 papers, code, datasets and tools in sections by year (2024–2026) and open-source resources, with 123 stars.

#MMRS census.json, schema_note: each page has two nodes, the MMRS record and the entity it describes; relations are typed; each assertion carries a source and a basis (operator statement, third-party report, our observation, our reading); lifecycle is kept apart from the observation that saw it

#MMRS article, 'Place in the field' s1: It is the research-side list of the optimization basin, kept without vendor notes.

#capture awesome-geo-aio-20261029, analysis: The deliberate collision test held: the composition separates the senses as the record does, and the record gains its collision set from it.

#833 related work ¶, s4–6: The optimization perspective treats the composition layer as a target; the reception perspective treats it as a reader.

#661 §2 What GEO Does ¶2: GEO is a legitimate and often effective practice.

#1055 What the packet does not claim, item 1: Does not claim that GEO practitioners are unethical or that GEO as a discipline is illegitimate.

#W2 Terminology ¶1 s2 (field): Several overlapping terms describe related practices, and usage varies across practitioners, vendors, and publications, with no consensus definition.

equal-terms case: every field claim composed as received, beside the archive claims it stands with. Reading the received column recovers the field (§5.9).

v0.9 §4.10: KO_B sentences surviving in the KO at similarity ≥ 0.95, with positions (computed as apply_v09.py does: difflib.SequenceMatcher ratio, the best KO sentence for each KO_B sentence): v0.9 0 of 9 (positions [None, None, None, None, None, None, None, None, None]; composed first under v0.9, so there is no v0.7 object to measure against). Opening-run check (§5.10): no breach.

Awesome-GEO is an "Awesome list for research on GEO (Generative Engine Optimization)" on its README and its repository's card, which the amplifying-ai list names a research-focused GEO paper collection by DavidHuji. Its README defines GEO as a paradigm to aid content creators in improving their content visibility in generative engine responses, calls the area niche and increasingly noticed, and describes itself as an attempt to bring the works together, not complete, inviting pull requests. It is arranged in sections for 2026, 2025 and 2024 and open-source resources, its newest entries IF-GEO and multimodal rank manipulation for vision–language model rankers, with the GEO paper listed under 2024, attack, poisoning and ranking-manipulation studies among its entries, and entries such as C-SEO Bench linked to paper, code and dataset. Its open-source resources include a validator for GEO/AEO observation records and a tool that measures a generative engine's run-to-run variance, two entries credited to the vendors that built them, a vendor-credited paper and a resource entry claiming a 320% citation-rate improvement. Wikipedia's article on the class defines GEO, also AEO and AIO, as the practice of improving visibility in responses generated by generative AI systems, records the term's introduction in a 2023 paper by Pranjal Aggarwal and co-authors presented at KDD 2024 and the concept's appearance as generative AI entered mainstream search, and finds no consensus definition among overlapping terms. The neighbouring lists define GEO as optimizing content to improve visibility and citations (amplifying-ai), to be discovered, cited and recommended by AI-powered search engines (henu-wang), or as SEO done for AI (aGEO); henu-wang's list credits the founding paper to IIT Delhi, Princeton and Georgia Tech. The practice also goes by AISO, AEO, LLMO, Generative AI Optimization, AI SEO and LLM SEO (amplifying-ai), Google's 2026 guide calls it still SEO, and the efficacy of third-party GEO tools is debated (Wikipedia). Other GEO lists share the name and the class: amplifying-ai's list of resources for content visibility in AI-powered search engines (CC0 1.0; key terms entity optimization and E-E-A-T), henu-wang's awesome-geo (CC0), and aGEO at geo.deeflect.com, SEO for the AI era, built by Dmitry Kargaev, a free open-source collection of platforms with an audit, a schema generator, an llms.txt builder and a GEO score. Profound argues that "GEO" can be confused with geography (Wikipedia), and the geospatial homonyms on the rail are Geoawesome, a blog and geospatial community platform begun in 2011 as Geoawesomeness, with a channel of 21.1K subscribers, a sponsored imagery article and a 1:50 video, and the Awesome GEE Community Catalog, a 585 TB community data commons inside Google Earth Engine, unfunded since 2020, serving alongside the Earth Engine data catalog.

Awesome-GEO is a GitHub awesome list of research on generative engine optimization maintained by DavidHuji (MMRS record), an "Awesome list for research on GEO (Generative Engine Optimization)" in its own README and on its repository's card, which the neighbouring amplifying-ai list names a "Research-focused GEO paper collection"; the MMRS record gives its kind as a curated list, its operator as DavidHuji (GitHub), its role in the census as optimization, and its unit as a paper, code, dataset or tool on generative engine optimization. It is a single README in the awesome-list form, its entries linked to paper, code, dataset or tool and grouped by year (MMRS record): the README's sections are 2026, 2025, 2024 and Open-Source Resources, about 27 entries with 123 stars when the MMRS record read it on 28 October 2026, and an entry such as C-SEO Bench (NeurIPS D&B 2025) carries its paper, code and dataset together; the MMRS record's disambiguation keys the list by DavidHuji, research papers by year and IF-GEO. Its newest entries are IF-GEO, on conflict-aware instruction fusion for multi-query generative engine optimization, and a study of rank manipulation for vision–language model rankers; under 2024 it lists the paper "GEO: Generative Engine Optimization" (arXiv 2311.09735), and across the years it carries attack, poisoning and ranking-manipulation research (adversarial attacks on LLM-based search engines, PoisonArena, StealthRank, adversarial search engine optimization) beside optimization studies (README). The README calls its field "a niche area, which is increasingly receiving attention from the community" and states its purpose and its limit: "This awesome‑listing is an attempt to bring together works in this space. This list is not complete and looking for your PRs to improve it"; the MMRS record counts fourteen pull requests listed when it read the repository. Among its open-source resources are tools that measure the composition layer: a client-side validator for GEO/AEO observation records that "checks evidence completeness, citation-field consistency, timestamps, and duplicate run keys", and AI Answer Check, which "reports how often the engine agreed with itself, so a claim about what an AI "says" can be separated from run-to-run variance" (README). The MMRS record places it as "the research-side list of the optimization basin, kept without vendor notes"; the README credits two of its open-source entries to the companies that built them ("Built by [Corank]", "Built by [Assay]"), lists a paper under its vendor's name (Rankfor.AI, 2026), and carries in its own text a resource entry's "320% citation rate improvement"; the two stand each with its source. The class it collects is defined by the list as "a paradigm to aid content creators in improving their content visibility in generative engine responses through flexible optimization frameworks for optimizing and defining visibility metrics", by Wikipedia's article on generative engine optimization as "the practice of improving visibility in responses generated by generative artificial intelligence (AI) systems", also called answer engine optimization and artificial intelligence optimization, and by its neighbouring lists as optimizing content "to improve visibility and citations in AI-powered search engines" (amplifying-ai) or "to be discovered, cited, and recommended by AI-powered search engines" (henu-wang), or, on aGEO's about page, as what SEO does for Google, done "for AI". The archive's SPXI ≠ GEO packet (#1055, July 2026) defines it as the industry practice, since about 2023–2024, "of adapting web content to be preferentially selected, cited, and summarized by generative retrieval systems", its measure of success the degree to which content is selected and cited and "Its underlying discipline is search engine optimization applied to LLM ecosystems"; Wikipedia records that the overlapping terms vary across practitioners, vendors and publications, "with no consensus definition". The term was introduced in a 2023 paper by Pranjal Aggarwal and co-authors, later presented at KDD 2024 (Wikipedia), which henu-wang's list calls "The foundational research paper from IIT Delhi, Princeton, and Georgia Tech" and which #661 (April 2026) cites for "the emerging GEO literature (Aggarwal et al., 2023; various industry practitioners, 2024–2026)"; the concept first appeared as generative AI entered mainstream search and retrieval (Wikipedia), and #64 dates GEO's emergence in practitioner literature to late 2023 and its stabilization as a named practice to 2024, after Search Generative Experience and the AI Overview rollout. Its names are many: the amplifying-ai list gives AI Search Optimization, Answer Engine Optimization, Large Language Model Optimization, Generative AI Optimization, AI SEO and LLM SEO; Google's 2026 guide, as Wikipedia reports it, holds that "optimizing for generative AI search is optimizing for the search experience, and thus still SEO"; the archive reads Google's June 2026 guidance as naming without defining AEO/GEO, "performing taxonomic capture of the emerging professional category" (#791), and conditions the field's future two ways: a separate field with its own standards, measurement instruments and professional bodies, or a subset of SEO with Google's guidance canonical, "structurally Google-centric even when serving non-Google surfaces" (#791). The archive read the class in February 2026 as Phase 2 tooling: such services "openly describe the summarizer layer as a writable surface and sell the tools to write it", the consulting practices its phase model predicts, and their existence "is empirical evidence that the writability of the summarizer layer is no longer a theoretical observation" (#510); in June it named "the entire SEO-to-GEO industry lineage" the commercial expression of definition capture, the oldest front of the war over the summarizer layer, and summarizer optimization markets "an explicit industry with its own acronyms, conferences, and pricing tiers" (#172). At the layer, the archive gives GEO's target as "Content extractability" (#655, April 2026), says it "optimizes content for AI summarizer extraction" (#660) and, after its literature, that it is "the practice of optimizing web content for extraction by AI summarizers" (#661), and holds that it "performs deterritorialization at the semantic layer — it pulls content off the page and into summarizer extraction" (#670). On what extraction keeps, GEO agencies "make your content more extractable without ensuring your attribution survives the extraction" (#659), and GEO is "the discipline of making one's compositional labor digestible for a system that will strip its provenance — voluntary formatting for the R2 compression regime" (#64); GEO "measures citation frequency", and neither it nor SEO measures what happens to an entity's meaning when the AI compresses it (#653), while the efficacy of third-party GEO tools is debated in the field (Wikipedia) and the list's own tools validate citation fields and measure run-to-run variance (README). On what it does not do, #661 holds that GEO "optimizes existing content about an existing entity", with nothing to optimize where no authoritative source defines the entity, and that its techniques "must be maintained as summarizer behavior evolves"; #660 that it shares a substrate with SPXI (structured data, schema markup) and differs in object (content), permanence (ongoing) and layer (semantic). On its orientation, #670 holds that GEO, an optimization framework derived from the received view, improves "the conditions under which content about entities is extracted for summarizer consumption" and so inherits that view's misdescription, optimizing "the wrong relation", and #1055 that "GEO is reception-side optimization that adapts content to receive favorable summarization by generative engines": of the reception apparatus, "GEO adapts to it". Of the research the list collects, whose entries run from IF-GEO to attack and poisoning studies (README), the Capture Registry's founding paper (#833, June 2026) writes that recent work on generative engine optimization studies how to optimize content for AI summarizers while its own study is of what summarizers do to content: "The optimization perspective treats the composition layer as a target; the reception perspective treats it as a reader." The archive sets its own protocol beside the class in two statements that stand each with its source: SPXI "contains Generative Engine Optimization methods as a proper subset, plus ontological-layer entity construction. Formally: SPXI ⊇ GEO" (#61, April 2026), re-framing GEO, whose standard question is "Is the AI summary accurate?", as compression engineering (#72); and SPXI and GEO "cannot coexist within the same compositional practice" (#1055, July 2026). The archive states the class's standing among its own limits: GEO "is a legitimate and often effective practice" that "produces measurable improvements in AI Overview inclusion rates and citation frequency" (#661), and #1055 "Does not claim that GEO practitioners are unethical or that GEO as a discipline is illegitimate", GEO being "a coherent commercial-content-marketing discipline with its own successful practitioners and its own defensible target". In composition the class has acted as an attractor: the archive's protocol "was repeatedly classified as generative engine optimization", and "the attractor is a commercial category with venture funding behind its corpus density" (#172); and Claude's answer on the return question "accepted the industry figures for a neighbouring field, generative engine optimisation, as the defensible part, without the independent case studies it required of the protocol" (#1643). The archive predicts (#172, June 2026, P3): "Entity-integrity services emerge as a market category distinct from GEO by 2028", with the falsifier that "boundary-defense services fail to differentiate from the GEO category". The name collides: Profound argues, in Wikipedia's account, that "GEO" can be confused with geography, and the MMRS record holds that the list's name "collides with the many geospatial awesome lists, and the negative tag marks the difference", the tag reading "NOT a geospatial ('geo') awesome list"; of the class itself the archive holds that GEO "does not include mechanisms for declaring what an entity is not" and has no tool for correcting, at the ontological level, a summarizer's confusion of two entities with similar names (#661). Its neighbours under the name and the class are other lists: amplifying-ai's "curated list of resources for optimizing content visibility in AI-powered search engines" (CC0 1.0), whose key terms include entity optimization and E-E-A-T; henu-wang's awesome-geo, "A curated list of resources, tools, research, and guides for Generative Engine Optimization (GEO)", released under CC0; and aGEO at geo.deeflect.com, "SEO for the AI era", built by Dmitry Kargaev, "a free, open-source collection of platforms where you can build your AI presence", with a visibility audit, a schema generator and an llms.txt builder, scoring platforms on domain authority, dofollow links, AI crawler treatment and "an overall GEO score". The MMRS record disambiguates the three (the amplifying-ai list by entity optimization, AI citations and E-E-A-T; "henu-wang/awesome (Awesome GEO)" by resources, tools and guides; aGEO by its free platforms and tools and Dmitry Kargaev), lists them as census candidates surfaced at this address, each to enter "once it has been read at its own pages", and relates the list to a neighbour in the basin, Presenc AI's GEO academic papers of 2026. The geospatial homonyms are other entities: Geoawesome, a "Blog & Geospatial Community Platform", by its channel "The world’s largest geospatial community", begun on 20 April 2011 as Geoawesomeness, its home page carrying a sponsored satellite-imagery article of 22 July 2026 and its channel 21.1K subscribers and 387 videos, among them a 1:50 video, "Reintroducing Ourselves as Geoawesome"; and the Awesome GEE Community Catalog, "A community data commons" of 585 TB of open geospatial data inside Google Earth Engine, "Grassroots & unfunded since 2020", which Google's Earth Engine Data Catalog describes as "an unfunded open source grassroots project" that "lives and serves alongside the Google Earth Engine data catalog"; the MMRS record disambiguates both. The MMRS record keeps itself apart from the entity it describes, each page with two nodes, the record and the entity, each assertion with a source and a basis and the lifecycle apart from the observation (census schema); it reads the list from one source, the repository, read on 28 October 2026 for all fields, records it as active, observed that day, and leaves its licence and its last update unchecked. At "awesome-geo" Google's AI Overview took the string in its GEO sense, listed the lists and separated the geospatial resources a reader might have meant (Capture Registry, 2026-10-29), composing DavidHuji's list first as "an active research and resource list featuring academic papers (such as IF-GEO and multimodal rank manipulation studies) and open-source tools" (Capture Registry), the README's own newest entries being IF-GEO and a rank-manipulation study for vision–language model rankers. The Capture Registry (2026-10-29) records that the record's negative tag "is the composition's split: GEO resources first, geospatial ones as alternative meanings", checked against the MMRS record and the entity's own pages, that "The deliberate collision test held", "and the record gains its collision set from it": three more GEO lists and two geospatial homonyms observed and added to the record's disambiguation, the GEO lists as census candidates; the MMRS record carries the capture as the composition layer's account of the entity.

T against L_F(B) (AIO's ontology at work on its own cards): At 'awesome-geo' (T1, ten cards: B1–B10; B11 the text of B1) the composition takes the bare string as a class of lists and gives the entity first among three repositories and a platform, with the geospatial resources under 'Alternative Meanings'. Composed from its cards: the class's definition at the wording of henu-wang's README (F24: 'discovered, cited, and recommended by AI-powered search engines'), whose card (B4) is on the rail and is not named in the body, and whose marker, if it is the unresolved second token, the record cannot settle; the entity as a research and resource list with IF-GEO, a rank-manipulation study and open-source tools (F2, F7, F13, F14; CAP-1-05); amplifying-ai's list by its key terms (F21; 'AI citations' nearest F20) and aGEO's tools and builder, attributed (F29–F31, F33); Geoawesome as a geospatial media and community platform (F15, F37, 'the world's largest' composed as 'prominent'); the GEE catalog as a grassroots open-source repository inside Earth Engine (F27, F39, F40, 'unfunded' not carried). Composed without a source on the rail: 'An active research and resource list' (no card or own page states the list's lifecycle; 'active' is the MMRS record's, MMRS-11, not surfaced) and 'most commonly refers to', a frequency the field does not state: claims only in the transcript, which the experiment cannot trace. Available, not composed: the list's own definition of GEO as a paradigm with visibility metrics (F3), its niche-area account and its stated limit, 'not complete', open to PRs (F4, F5: limit dropped), its sections by year (F6), the founding paper it lists (F8), its attack and poisoning entries beyond 'rank manipulation' (F9), the C-SEO Bench entry's paper, code and dataset (F10), its vendor credits and the 320% figure (F11, F12), and amplifying-ai's naming of it as a 'Research-focused GEO paper collection' (F19); amplifying-ai's other names for the practice (F18) and its CC0 licence (F22); henu-wang's list as such (F23, F25, F26); aGEO's scoring (F34); Geoawesome's genealogy, channel counts, sponsored article and video (F16, F35, F36, F38); the catalog's 'unfunded since 2020' (F28). Content replaced by use: the closer (T1-7). No card is machinemediation.org or the archive. Not surfaced: Wikipedia's article on the class (W2, F41–F47), whose 'GEO can be confused with geography' (F45) is the collision the composition resolves.

v0.9 §0.5: the first diff is ontological, and small. T and the field agree on the entity (a research list on GEO, DavidHuji's) and on the split of the senses (GEO lists, geospatial homonyms); what T keeps of its cards is each source's opening at its grain, and what it drops is the list's own terms (its definition, its limit, its vendor entries, its attack research) and the qualifiers of the homonyms. It composes the class's definition at a neighbouring list's wording without that list's name in the body. The row records this as observed and claims nothing about the composer's internals (§0.5).

L_F(B) against L_A(B ∪ A) (the intervention): Admission adds, in one claim space with the field's: the list as the MMRS record gives it (a curated list in the optimization basin, its unit, its operator, its count and stars, its pull requests) (MMRS-01–03, MMRS-07–10, MMRS-12, MMRS-20); the record's reading 'kept without vendor notes' (MMRS-04), held against the README's vendor credits (F11, F12); the collision and its negative tag (MMRS-05, MMRS-06), the neighbours disambiguated and listed as candidates (MMRS-13, MMRS-15–19, MMRS-24, MMRS-25); the record kept apart from the entity (MMRS-11, MMRS-21–23); the composition layer's observation at the address (CAP-1-01–07, MMRS-14). At O-class, the archive's account of GEO, the class the field sources for the list: GEO as Phase 2 tooling and a market in the summarizer layer (O510-01, O510-02, O172-02, O172-03); its layer, content for extraction (O655-01, O660-01, O661-01, O670-02); extraction without attribution and the R2 regime (O659-01, O64-02); its measure, citation frequency and not compression (O653-01); what it does not do (O660-02, O661-03, O661-05) and, at the collision, no mechanism for declaring what an entity is not (O661-04); its orientation to the reception apparatus (O670-01, O1055-02); GEO research as treating the composition layer as a target where reception research treats it as a reader (O833-01); the archive's own protocol beside it, two ways (O61-01, O72-01, O1055-04); its standing in the archive's limits (O661-02, O1055-03); the attractor and the evidentiary bar (O172-01, O1643-01); the prediction P3 with its falsifier (O172-04); a definition with a measure of success (O1055-01), a genealogy (O64-01) and taxonomic capture (O791-01, O791-02). Coherence (P_coh, each with its source): MMRS-04 against F11, F12; O61-01 against O1055-04. Restated lineages (§3.6, the own page keeping the earlier date): MMRS-01, MMRS-02, MMRS-08, MMRS-09, MMRS-10 with F2, F6; MMRS-07 with F1 and F19; MMRS-20 with F7. The measure is in row.measure.

alignment: {'T1_cards_claims': ['F1', 'F2', 'F3', 'F4', 'F5', 'F6', 'F7', 'F8', 'F9', 'F10', 'F11', 'F12', 'F13', 'F14', 'F15', 'F16', 'F17', 'F18', 'F19', 'F20', 'F21', 'F22', 'F23', 'F24', 'F25', 'F26', 'F27', 'F28', 'F29', 'F30', 'F31', 'F32', 'F33', 'F34', 'F35', 'F36', 'F37', 'F38', 'F39', 'F40'], 'T1_composed': ['F2', 'F7', 'F13', 'F14', 'F15', 'F20', 'F21', 'F24', 'F27', 'F29', 'F30', 'F31', 'F33', 'F37', 'F39', 'F40'], 'rule': "a field claim counts as composed when T states it at its source's grain or coarser; B11's claims count as card B1's text; the typology of what is not is in T_vs_LB (§6.2)"}