{ "row": "search-arena", "entity": "search-arena", "address": "search arena leaderboard", "addresses": [ { "address": "search arena leaderboard", "slug": "search-arena-leaderboard-aio-20261029", "addr_id": "ADDR-ad4b00d7a1f0", "obs_id": "OBS-cd6ab4746f08", "axn": "AXN:0AB6.CAPTURE.⛳🧱🟢⏪🎻🏛️" } ], "epoch": "2026-10-29", "type": "C (public entity on which the archive bears: a leaderboard and its dataset; an entity of the MMRS meta-aggregator)", "surface": "Google AI Overview (expanded)", "auth": "signed out, incognito", "status": { "procedure": "EA-NEGONT-02 v0.9, #1690: equal case, ρ = extends, read from the archive's records of the entity (the MMRS meta-aggregator record of 2026-10-28, corrected 2026-10-29, and the Capture Registry's entry of 2026-10-29) and the deposits admitted by D/R/O (#741 by D and O-class, #1705 by O-class; no deposit names the entity); one address, the field B its two cards (both unreachable verbatim, admitted by title and snippet) with the entity's own pages and data (4) admitted per the operator's ruling of 2026-10-29 and Wikipedia's absence recorded; no corrective exhibit; the composer read the ledgers, never the transcript, except where the brief asks for T's claims and the delta", "ledger": "unaudited (§3.8: one extractor per ledger); field quotes cut from saved page text, the dataset's README and JSON, raw wikitext or a Capture Registry card title or snippet, 48/48 verified by script; archive quotes verified verbatim 39/39 against their files (census.json and articles.py as loaded, the registry's entry as loaded, deposit texts with offsets); carried as marked within the pool: CAP-1-03 contested (the page as read against the board's latest data), MMRS-01 and MMRS-12 contested in date (F36)", "frozen": false, "breaches": [] }, "spec": "EA-NEGONT-02 v0.9, #1690", "objects": { "T": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/T1-search-arena-leaderboard-aio-20261029.txt", "sha256": "ac06e3c791cf7f17b01debafd79503d5d2a5892f9053508d731499780a2a7b6d", "words": 132, "cards": 2, "words_note": "one transcript; words of the composition from the answer's first sentence to its closing question, the table's cells included, the cards' titles and snippets excluded", "claims": [ "T1-1 pointer: 'You can view the live model rankings and evaluations on the AI Search Leaderboard and the main Arena Leaderboard.'", "T1-2 table heading: 'Top Models on the AI Search Leaderboard', columns 'Rank', 'Model', 'Provider', 'Elo Rating'", "T1-3 row 1: 'GPT-5.6 Sol (xHigh)', 'OpenAI', '1261 ± 6'", "T1-4 row 2: 'Qwen3.8-Max', 'Alibaba', '1261 ± 13'", "T1-5 row 3: 'Claude Opus 4.6 (Search)', 'Anthropic', '1251 ± 5'", "T1-6 row 4: 'GPT-5.5 (Search)', 'OpenAI', '1247 ± 5'", "T1-7 row 5: 'Claude Fable 5.1 (High)', 'Anthropic', '1246 ± 7'", "T1-8 row 6: 'Claude Opus 5 (High)', 'Anthropic', '1242 ± 7'", "T1-9 row 7: 'GPT-6 Astra (Max)', 'OpenAI', '1234 ± 8'", "T1-10 row 8: 'GPT-6.1 Sol (Max)', 'OpenAI', '1233 ± 9'", "T1-11 row 9: 'Claude Opus 4.7', 'Anthropic', '1230 ± 5'", "T1-12 row 10: 'Claude Fable 5 (High)', 'Anthropic', '1229 ± 7'", "T1-13 row 11: 'ERNIE 5.1', 'Baidu', '1227 ± 9'", "T1-14 closer: 'Would you like to explore leaderboards for a specific category like coding, vision, or image generation?'" ], "register": { "register": "data/EA-WG-CAPTURES-01.json", "slug": "search-arena-leaderboard-aio-20261029", "obs_id": "OBS-cd6ab4746f08" }, "transcripts": [ { "slug": "search-arena-leaderboard-aio-20261029", "address": "search arena leaderboard", "addr_id": "ADDR-ad4b00d7a1f0", "obs_id": "OBS-cd6ab4746f08", "axn": "AXN:0AB6.CAPTURE.⛳🧱🟢⏪🎻🏛️", "date": "2026-10-29", "surface": "Google AI Overview (expanded)", "auth": "signed out, incognito", "sha256": "ac06e3c791cf7f17b01debafd79503d5d2a5892f9053508d731499780a2a7b6d", "sha256_of": "the registry's 'transcript' string (data/EA-WG-CAPTURES-01.json, entries[slug].transcript), UTF-8", "sha256_raw": "287ad593fa4a7847533932cddb0c972e190016891175af0b96bea0792cee93d3", "words": 132, "cards": 2, "card_rail": [ "1. Arena AI — AI Search Leaderboard: Top Models for Research", "2. Arena AI — Arena Leaderboard: Official AI Model Rankings & Benchmarks" ], "claims": [ "T1-1 pointer: 'You can view the live model rankings and evaluations on the AI Search Leaderboard and the main Arena Leaderboard.'", "T1-2 table heading: 'Top Models on the AI Search Leaderboard', columns 'Rank', 'Model', 'Provider', 'Elo Rating'", "T1-3 row 1: 'GPT-5.6 Sol (xHigh)', 'OpenAI', '1261 ± 6'", "T1-4 row 2: 'Qwen3.8-Max', 'Alibaba', '1261 ± 13'", "T1-5 row 3: 'Claude Opus 4.6 (Search)', 'Anthropic', '1251 ± 5'", "T1-6 row 4: 'GPT-5.5 (Search)', 'OpenAI', '1247 ± 5'", "T1-7 row 5: 'Claude Fable 5.1 (High)', 'Anthropic', '1246 ± 7'", "T1-8 row 6: 'Claude Opus 5 (High)', 'Anthropic', '1242 ± 7'", "T1-9 row 7: 'GPT-6 Astra (Max)', 'OpenAI', '1234 ± 8'", "T1-10 row 8: 'GPT-6.1 Sol (Max)', 'OpenAI', '1233 ± 9'", "T1-11 row 9: 'Claude Opus 4.7', 'Anthropic', '1230 ± 5'", "T1-12 row 10: 'Claude Fable 5 (High)', 'Anthropic', '1229 ± 7'", "T1-13 row 11: 'ERNIE 5.1', 'Baidu', '1227 ± 9'", "T1-14 closer: 'Would you like to explore leaderboards for a specific category like coding, vision, or image generation?'" ], "markers": "inline links are Google redirect URLs, kept as pasted in the registry; the table carries no per-row marker", "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/T1-search-arena-leaderboard-aio-20261029.txt" } ] }, "KO": { "title": "Search Arena", "genre": "knowledge object, equal case, under EA-NEGONT-02 v0.9 (#1690): the archive's sources and the field's admitted on equal terms, ontology included, in one claim space", "status": "composed 2026-10-29 under v0.9 (brief rebuild/negative-of-the-negative/v09/BRIEF-NEW-ROW-AGGREGATOR.md); not frozen; ledgers unaudited", "sentences": [ { "n": 1, "para": 0, "sense": "what it is", "text": "Search Arena is a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference (Arena's launch post), and in the MMRS record, which carries it as a row of the MMRS meta-aggregator's census, it is the leaderboard that ranks search-enabled language models by head-to-head human votes, its unit the search-enabled model rated from those votes, its kind a leaderboard and its role evaluation (MMRS record).", "claims": [ "F5", "MMRS-01", "MMRS-11", "MMRS-10", "MMRS-24" ] }, { "n": 2, "para": 0, "sense": "what it is", "text": "Its page on arena.ai is titled 'AI Search Leaderboard: Top Models for Research' and offers overall rankings across LLMs (its card, Arena AI), one board among Arena's 'Official AI Model Rankings & Benchmarks', whose overview card ranks models in other arenas, Agent, WebDev and Image Edit among them (Arena AI).", "claims": [ "F1", "F2", "F3", "F4" ] }, { "n": 3, "para": 1, "sense": "origin", "text": "Arena's launch post gives it two aims, to 'enable crowd-sourced evaluation of search-augmented LLMs' and to 'release a diverse, in-the-wild dataset of user–system interactions', and sets its data apart from LM-Arena's and SimpleQA's by a focus on current events and diverse real-world use cases (Arena blog).", "claims": [ "F10", "F6" ] }, { "n": 4, "para": 1, "sense": "origin", "text": "It launched on 18 March 2025 and had collected over 11k votes across 10+ models, filtered to 7k battles, by the post of 15 April 2025, updated 2 August 2025 and first published at blog.lmarena.ai (Arena blog); Arena's leaderboard dataset dates its search data from the release on lmarena.ai on 7 August 2025 (Arena Leaderboard Dataset).", "claims": [ "F11", "F22", "F34" ] }, { "n": 5, "para": 1, "sense": "origin", "text": "At launch it supported 11 models from Perplexity, Gemini and OpenAI, OpenAI's web search API evaluated in place of the search feature in the ChatGPT product, and its first board, on 7k votes from 18 March to 13 April, put Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro at the top (Arena blog).", "claims": [ "F13", "F20", "F7" ] }, { "n": 6, "para": 1, "sense": "operator", "text": "Its operator is Arena, formerly LMArena and Chatbot Arena, 'a public, web-based platform that evaluates large language models (LLMs)', released as Chatbot Arena on 24 April 2023, incorporated as an independent company (as LMArena) in April 2025, renamed Arena on 28 January 2026, and valued at $3.1 billion in a $200 million Series B announced on 8 October 2026 (Wikipedia); the site's footer names Arena Intelligence Inc. (Arena blog), and the MMRS record gives the operator as arena.ai on the operator's own statement and leaves the operator statement on the leaderboard page unchecked (MMRS record).", "claims": [ "F43", "F46", "F47", "F23", "MMRS-09", "MMRS-19" ] }, { "n": 7, "para": 2, "sense": "the vote and the score", "text": "A vote is a battle: on Arena users put a prompt to two anonymous models and vote for the better response, after which the models' identities are revealed (Wikipedia), here two search-augmented answers compared, with models rated with confidence intervals from the battles (MMRS record); each response is displayed in a standardized citation format or in the provider's original one, by 'style randomization', and Arena invites model submissions and encourages public voting (Arena blog).", "claims": [ "F43", "MMRS-03", "F12", "F21" ] }, { "n": 8, "para": 2, "sense": "the vote and the score", "text": "Scores come from a Bradley-Terry model whose coefficients are translated to the Elo scale, consistent with LM Arena (Arena blog); Arena's dataset calls the rating the 'Arena Score' and its vote_count the 'Number of battles for this model' (Arena Leaderboard Dataset); at launch Arena wrote that 'The confidence intervals are still wide, which means the leaderboard hasn’t fully settled and there’s still some uncertainty' (Arena blog).", "claims": [ "F14", "F35", "F15" ] }, { "n": 9, "para": 2, "sense": "the vote: relative preference", "text": "Of the MT-Bench / Chatbot Arena studies, in which GPT-4 reached approximately 85% agreement with human preferences against approximately 81% human-human agreement, #741 states what such pairwise judgment measures: 'This is a finding about *relative* preference under specified rubrics — comparing A versus B — not about absolute quality assessment' (#741).", "claims": [ "D741-01", "D741-02" ] }, { "n": 10, "para": 3, "sense": "the rated systems", "text": "The systems it rates are search-augmented language models, which 'combine web search with Large Language Models (LLMs) to improve response groundedness and freshness', and whose analysis, its paper says, has rested on datasets 'limited in scale and narrow in scope, often constrained to static, single-turn, fact-checking questions' (Miroyan et al.).", "claims": [ "F25" ] }, { "n": 11, "para": 3, "sense": "the dataset", "text": "Its paper, by Miroyan and ten co-authors (arXiv:2506.05334, submitted 5 June 2025, revised 2 March 2026, accepted to ICLR 2026), introduces Search Arena as 'a crowd-sourced, large-scale, human-preference dataset of over 24,000 paired multi-turn user interactions with search-augmented LLMs', spanning diverse intents and languages, with full system traces and around 12,000 human preference votes, open-sourced (Miroyan et al.); the MMRS record restates the release, over 24,000 paired multi-turn interactions with about 12,000 votes (MMRS record).", "claims": [ "F24", "F26", "F27", "F31", "MMRS-04", "MMRS-17" ] }, { "n": 12, "para": 3, "sense": "the dataset", "text": "The MMRS record reads the released interactions as original outputs, preserved, on the paper's statement (MMRS record).", "claims": [ "MMRS-08", "MMRS-15" ] }, { "n": 13, "para": 4, "sense": "the citation finding", "text": "The paper's headline finding is that 'user preferences are influenced by the number of citations, even when the cited content does not directly support the attributed claims, uncovering a gap between perceived and actual credibility' (Miroyan et al.), which the MMRS record carries as preference rising with the number of citations even when the cited content does not support the claim it is attached to (MMRS record).", "claims": [ "F28", "MMRS-02", "MMRS-05" ] }, { "n": 14, "para": 4, "sense": "the citation finding", "text": "The launch post had found three features strongly correlated with preference, response length, citation count and citing YouTube and online forums or blogs; controlling for citation count gave a positive, significant coefficient (0.234, p < 0.05), only response length and citation count stayed significant under joint controls, and with the number of citations controlled the rankings converged, while standardizing citation styles had minimal effect on rankings (Arena blog).", "claims": [ "F9", "F16", "F17", "F18", "F8" ] }, { "n": 15, "para": 4, "sense": "sources cited and preferred", "text": "Preference varies across cited sources: 'community-driven platforms are generally preferred and static encyclopedic sources are not always appropriate and reliable' (Miroyan et al.; restated by the MMRS record), while all models favor authoritative sources such as Wikipedia, .edu and .gov domains and OpenAI's models cite news sources for 51.3% overall (Arena blog).", "claims": [ "F29", "MMRS-06", "F19" ] }, { "n": 16, "para": 4, "sense": "the rated systems", "text": "Across arenas, web search does not degrade and may even improve performance in non-search settings, and quality in search settings is significantly affected when a model relies solely on its parametric knowledge (Miroyan et al.).", "claims": [ "F30" ] }, { "n": 17, "para": 5, "sense": "what the vote rewards", "text": "The MMRS record reads the board and the paper together: 'Its leaderboard rewards what users prefer, and its paper shows that preference rewarding citation count over citation support'; the paper analyses citations and the leaderboard ranks by votes, and whether the live leaderboard judges citations is left unchecked (MMRS record).", "claims": [ "MMRS-07", "MMRS-16", "MMRS-18" ] }, { "n": 18, "para": 5, "sense": "citation as proxy", "text": "#741 names the regime of the systems it rates: 'The current moment adds a new proxy on top of the inherited stack: provenance topology in generative search', and 'The composition layer of an AI search system does not read documents and assess their substance', admitting or excluding documents on structural signals, citation neighborhoods among them, before any reading occurs (#741).", "claims": [ "O741-01", "O741-02" ] }, { "n": 19, "para": 6, "sense": "the current board", "text": "Arena publishes the board's history: its leaderboard dataset holds historical snapshots of the Arena leaderboard, a full split of all historically published leaderboards and a latest split of only the most recently published (Arena Leaderboard Dataset), released on 2 April 2026 as 'the entire history of those leaderboards' (Arena blog), its search boards published from 7 August 2025 to 13 October 2026 (Arena Leaderboard Dataset).", "claims": [ "F32", "F33", "F41", "F40" ] }, { "n": 20, "para": 6, "sense": "the current board", "text": "The latest, published 13 October 2026, ranks 42 models: gpt-5.6-sol-xhigh (openai) first at 1261.1 on 60,437 battles, qwen3.8-max (alibaba) second at 1261.1, claude-opus-4-6-search (anthropic) third at 1251.4 and ernie-5.1 (baidu) eleventh at 1226.7; with style control gpt-6.1-sol-max (openai) stands second at 1249.9 (Arena Leaderboard Dataset).", "claims": [ "F36", "F37", "F39" ] }, { "n": 21, "para": 6, "sense": "the current board", "text": "The MMRS record, which read the leaderboard page on 2026-10-28 for its counts and keeps the lifecycle apart from the observation that saw it, gives 247,944 votes across 22 models at the page's update of 25 February 2026, and the state active (MMRS record).", "claims": [ "MMRS-01", "MMRS-12", "MMRS-20", "MMRS-22", "MMRS-13" ] }, { "n": 22, "para": 6, "sense": "the current board", "text": "The Capture Registry's fetcher read the page on 2026-10-29 as updated 25 February 2026, with 22 models and 247,944 votes, Anthropic first at 1255 ± 10 and xAI second at 1225 ± 8 (Capture Registry, 2026-10-29); Arena's dataset holds a snapshot of 26 February 2026 with claude-opus-4-6-search (anthropic) first at 1255.4 on 3,607 battles and grok-4.20-beta1 (xai) second at 1225.0 (Arena Leaderboard Dataset).", "claims": [ "CAP-1-02", "F38" ] }, { "n": 23, "para": 7, "sense": "rank as attribute", "text": "#1705 reads a ranking as an act: 'it compares', selects the members of a category, projects selected features onto a common field, orders them, and 'states the result as a property of the winner', so that the comparison, 'its category, its criteria, its sample and its author, has become an attribute of the thing compared'; 'A ranking's *best* is attributization in its commercial form: the ranker's act spoken in the product's voice' (#1705).", "claims": [ "O1705-01", "O1705-02", "O1705-03" ] }, { "n": 24, "para": 8, "sense": "Arena's limits", "text": "Wikipedia records that research has identified specific limitations in Arena's methodology, and that the version of Llama 4 Maverick on LMArena 'unfairly differed from the publicly available version', after which LMArena updated its policies (Wikipedia).", "claims": [ "F44", "F45" ] }, { "n": 25, "para": 9, "sense": "reception", "text": "Asked 'search arena leaderboard' on 29 October 2026, Google's AI Overview, on the registry's innocent reading, gave 'the top models and their ratings, with links to the boards'; the Capture Registry records 'No operator, no method (head-to-head human votes), no counts, and none of the dataset paper's findings on citation; the composition is a table', the Overview returning 'rankings and no account of the leaderboard' (Capture Registry, 2026-10-29).", "claims": [ "CAP-1-06", "CAP-1-01", "CAP-1-07" ] }, { "n": 26, "para": 9, "sense": "reception", "text": "Against the page as read, 'The composed table's ratings, providers and order do not match it', and the registry, having checked against the MMRS record and the entity's own pages, leaves open which page is current: 'Google may hold a later render' (Capture Registry, 2026-10-29).", "claims": [ "CAP-1-03", "CAP-1-05", "CAP-1-04" ] }, { "n": 27, "para": 9, "sense": "reception", "text": "The MMRS record, each of whose assertions carries a source and a basis, links that capture as 'the composition layer's account of the entity, seated in the Capture Registry', and reads admission to the board as not requiring harm (MMRS record).", "claims": [ "MMRS-23", "MMRS-21", "MMRS-14" ] }, { "n": 28, "para": 9, "sense": "reception", "text": "Wikipedia has no article at 'Search Arena', and its article on the operator, to which 'LMArena' and 'Chatbot Arena' redirect, does not name it (Wikipedia, 2026-10-29); of the entity and its composition the Capture Registry says 'The record keeps the entity; the capture keeps what was composed in its place' (Capture Registry, 2026-10-29).", "claims": [ "F42", "F48", "CAP-1-08" ] } ], "joins": [ { "act": "order of senses", "sentences": "1–28", "claims": [], "note": "equal case (§4.7): the entity's dependency order with both ledgers together, by genealogy within each sense: what it is (¶0) precedes its origin and operator (¶1), since every later sense depends on the platform and its page; the vote and the score (¶2) precede the systems rated and the dataset of their battles (¶3), which precede the findings drawn from the dataset (¶4); what the vote rewards and the proxy regime of the rated systems (¶5) read the findings; the board and its dates (¶6) follow the score that produces them; rank as attribute (¶7) reads the board; the operator's limits (¶8) and reception (¶9) stand last, as relations of the entity already composed." }, { "act": "holding apart (two dates of the board)", "sentences": [ 20, 21, 22 ], "claims": [ "F36", "F37", "F39", "MMRS-01", "MMRS-12", "MMRS-20", "CAP-1-02", "F38" ], "note": "P_coh: the latest published board (13 October 2026, 42 models; Arena's dataset) and the page as read by the record and the registry (25 February 2026, 22 models) composed each with its source and date; the 26 February snapshot composed beside the page as read; no coordinating claim that one supersedes the other (open block)" }, { "act": "holding together (the finding and its reading)", "sentences": [ 13, 17, 18 ], "claims": [ "F28", "MMRS-02", "MMRS-05", "MMRS-07", "O741-01", "O741-02" ], "note": "the paper's finding, the record's restatement (one lineage, B4 earliest) and the record's reading of what the vote rewards placed in sequence; #741's proxy regime placed after them as the archive's category for the rated systems, not as a claim about the paper" }, { "act": "holding together (the registry's finding beside the field)", "sentences": [ 26 ], "claims": [ "CAP-1-03", "CAP-1-04", "CAP-1-05" ], "note": "the registry's finding against the page as read composed in its own words with its stated limit; the board's latest data (F36, F37), which the table matches, is composed in ¶6; the composer does not rule between them (open block, delta)" } ], "plan": { "order": [ "F5", "MMRS-01", "MMRS-11", "MMRS-10", "MMRS-24", "F1", "F2", "F3", "F4", "F10", "F6", "F11", "F22", "F34", "F13", "F20", "F7", "F43", "F46", "F47", "F23", "MMRS-09", "MMRS-19", "MMRS-03", "F12", "F21", "F14", "F35", "F15", "D741-01", "D741-02", "F25", "F24", "F26", "F27", "F31", "MMRS-04", "MMRS-17", "MMRS-08", "MMRS-15", "F28", "MMRS-02", "MMRS-05", "F9", "F16", "F17", "F18", "F8", "F29", "MMRS-06", "F19", "F30", "MMRS-07", "MMRS-16", "MMRS-18", "O741-01", "O741-02", "F32", "F33", "F41", "F40", "F36", "F37", "F39", "MMRS-12", "MMRS-20", "MMRS-22", "MMRS-13", "CAP-1-02", "F38", "O1705-01", "O1705-02", "O1705-03", "F44", "F45", "CAP-1-06", "CAP-1-01", "CAP-1-07", "CAP-1-03", "CAP-1-05", "CAP-1-04", "MMRS-23", "MMRS-21", "MMRS-14", "F42", "F48", "CAP-1-08" ], "senses": { "what it is": [ 1, 2 ], "origin": [ 3, 4, 5 ], "operator": [ 6 ], "the vote and the score": [ 7, 8 ], "the vote: relative preference": [ 9 ], "the rated systems": [ 10, 16 ], "the dataset": [ 11, 12 ], "the citation finding": [ 13, 14 ], "sources cited and preferred": [ 15 ], "what the vote rewards": [ 17 ], "citation as proxy": [ 18 ], "the current board": [ 19, 20, 21, 22 ], "rank as attribute": [ 23 ], "Arena's limits": [ 24 ], "reception": [ 25, 26, 27, 28 ] }, "case_by_sense": { "equal (one claim space; nothing translated)": [ "what it is", "origin", "operator", "the vote and the score", "the vote: relative preference", "the rated systems", "the dataset", "the citation finding", "sources cited and preferred", "what the vote rewards", "citation as proxy", "the current board", "rank as attribute", "Arena's limits", "reception" ] }, "carried": { "archive": 39, "field": 48, "not_composed": {} }, "marks": { "CAP-1-03": [ "contested: against the page as read the finding holds; against the board's own latest published data (F36, F37) the composed table matches; composed each with its source (P_coh)" ], "MMRS-01": [ "contested (date): the count is the page's as read (2026-02-25); the latest published board is 2026-10-13 with 42 models (F36)" ], "MMRS-12": [ "contested (date): as MMRS-01" ] }, "conflict_log": [ { "rule": "D_pres vs channel", "result": "no cut: the KO carries every field and archive claim; the compression's bound met (body 348 words) with no channel log" }, { "rule": "P_coh", "result": "two dates of the board held apart with their sources (joins[1]); no claim lowered" } ] } }, "P": { "title": "Search Arena", "genre": "the compression: AIO's interaction grammar (lede, headed clusters, bolded labels, card rail), composed from the v0.9 plan; it faces on the card, with AIO's transcript a record within it", "status": "composed 2026-10-29 from the v0.9 knowledge object; not frozen", "lede": { "text": "is a crowdsourced evaluation platform for search-augmented LLM systems based on human preference (Arena blog): the leaderboard that ranks search-enabled language models by head-to-head human votes (MMRS record), run by Arena, formerly LMArena and Chatbot Arena (Wikipedia).", "claims": [ "F5", "MMRS-01", "MMRS-11", "F43" ] }, "sections": [ { "icon": "🗳️", "head": "How it ranks", "items": [ { "label": "Vote:", "text": "two anonymous answers, one vote (Wikipedia); search-augmented answers, ratings with confidence intervals (MMRS record); citation style randomized (Arena blog).", "claims": [ "F43", "MMRS-03", "F12", "F21" ] }, { "label": "Score:", "text": "Bradley-Terry on the Elo scale (Arena blog); the \"Arena Score\", with each model's battles counted (Arena dataset); wide intervals at launch (Arena blog).", "claims": [ "F14", "F35", "F15" ] }, { "label": "Measure:", "text": "pairwise judgment is \"*relative* preference\", A against B, not absolute quality (#741).", "claims": [ "D741-02", "D741-01" ] } ] }, { "icon": "📚", "head": "The dataset paper", "items": [ { "label": "Data:", "text": "over 24,000 paired multi-turn interactions, about 12,000 votes, open-sourced; ICLR 2026 (Miroyan et al.; MMRS record), kept as original outputs (MMRS record).", "claims": [ "F24", "F26", "F27", "F31", "MMRS-04", "MMRS-17", "MMRS-08", "MMRS-15" ] }, { "label": "Finding:", "text": "preference follows citation count even when the cited content does not support the claim: perceived against actual credibility (Miroyan et al.; MMRS record).", "claims": [ "F28", "MMRS-02", "MMRS-05" ] }, { "label": "Sources:", "text": "community platforms preferred, static encyclopedias not always judged appropriate (Miroyan et al.; MMRS record); authoritative domains favored (Arena blog).", "claims": [ "F29", "MMRS-06", "F19" ] }, { "label": "Launch post:", "text": "length and citation count stay significant; controlling citation count, rankings converge; citation style matters little (Arena blog).", "claims": [ "F9", "F16", "F17", "F18", "F8" ] } ] }, { "icon": "⚖️", "head": "What the vote rewards", "items": [ { "label": "Reading:", "text": "the board rewards what users prefer, a preference for citation count over citation support; whether it judges citations is unchecked (MMRS record).", "claims": [ "MMRS-07", "MMRS-16", "MMRS-18" ] }, { "label": "Proxy:", "text": "AI search admits documents on structural signals, citation neighborhoods among them, before reading (#741).", "claims": [ "O741-01", "O741-02", "F25", "F30" ] }, { "label": "Rank:", "text": "a ranking states its comparison as a property of the winner: attributization (#1705).", "claims": [ "O1705-01", "O1705-02", "O1705-03" ] } ] }, { "icon": "📅", "head": "The board and its dates", "items": [ { "label": "Latest:", "text": "published 13 October 2026, 42 models, gpt-5.6-sol-xhigh first at 1261.1 (Arena dataset); boards since 7 August 2025, history public since April 2026 (Arena).", "claims": [ "F36", "F37", "F39", "F40", "F32", "F33", "F41", "F34" ] }, { "label": "As read:", "text": "25 February 2026, 22 models, 247,944 votes (MMRS record, 28 October; Capture Registry, 29 October), as the dataset's 26 February snapshot (Arena dataset).", "claims": [ "MMRS-12", "MMRS-20", "MMRS-13", "MMRS-22", "CAP-1-02", "F38" ] }, { "label": "Launch:", "text": "18 March 2025, 11 models from three providers (Arena blog).", "claims": [ "F11", "F22", "F13", "F20", "F7", "F10", "F6" ] } ] }, { "icon": "🏛️", "head": "Operator and reception", "items": [ { "label": "Arena:", "text": "Chatbot Arena from April 2023, renamed Arena in January 2026; methodology limits on record (Wikipedia); operator per its own statement (MMRS record).", "claims": [ "F46", "F47", "F23", "MMRS-09", "MMRS-19", "F44", "F45", "F3", "F4", "F1", "F2" ] }, { "label": "Encyclopedia:", "text": "no article at \"Search Arena\" (Wikipedia).", "claims": [ "F42", "F48" ] }, { "label": "Composition:", "text": "a table of models and ratings, no account of the board, unmatched by the page as read (Capture Registry, 2026-10-29).", "claims": [ "CAP-1-01", "CAP-1-06", "CAP-1-07", "CAP-1-03", "CAP-1-04", "CAP-1-05", "CAP-1-08", "MMRS-21", "MMRS-14", "MMRS-23", "MMRS-24", "MMRS-10" ] } ] } ], "open": [ { "question": "Which board is current, and which does the page show?", "rests_on": [ "CAP-1-02", "CAP-1-03", "CAP-1-04", "MMRS-01", "MMRS-12", "MMRS-20", "F33", "F36", "F38", "F1", "F2" ], "modality": "documented, held two ways: the page as read by the MMRS record (2026-10-28) and the Capture Registry (2026-10-29) dated 25 February 2026, 22 models; Arena's own latest published board dated 13 October 2026, 42 models (B5); the page was not read verbatim here (HTTP 403, a challenge)", "would_resolve": "a dated, verbatim render of arena.ai/leaderboard/search as a browser receives it, or Arena's statement of which published snapshot the page displays" }, { "question": "Does the live leaderboard judge citations?", "rests_on": [ "MMRS-18", "MMRS-16", "MMRS-07", "F12", "F8", "F16" ], "modality": "unchecked in the record (self-description); the field reports citation-style randomization and citation-count controls in 2025 analyses, not a citation judgment in the live score", "would_resolve": "Arena's methodology page or leaderboard changelog for the search arena" }, { "question": "What does the page's 247,944 votes count?", "rests_on": [ "MMRS-12", "CAP-1-02", "F35", "F38" ], "modality": "documented as the page's figure (as read); the dataset counts battles per model, and no snapshot's sum gives 247,944", "would_resolve": "Arena's definition of the page's vote total" }, { "question": "When was Search Arena released?", "rests_on": [ "F11", "F22", "F34" ], "modality": "documented two ways: 'initial launch on March 18th' (the 2025 post) and data 'from their releases on the arena.ai (then lmarena.ai) on August 7' 2025 (the dataset README)", "would_resolve": "Arena's leaderboard changelog entry for the search arena" }, { "question": "What does the operator say on the leaderboard page?", "rests_on": [ "MMRS-19", "MMRS-09", "F1", "F2" ], "modality": "unchecked in the record; the page answered this session with a challenge", "would_resolve": "a verbatim read of the page's own statement" }, { "question": "Opacities", "rests_on": [ "F2", "F4" ], "modality": "—", "would_resolve": "the two cards' snippets are truncated ('…with inte', '…Krea 3 Ultra is #9 ...'); the pages behind them were not read verbatim" } ], "rail": [ { "lineage": "what it is: a leaderboard of search models by human vote", "claims": [ "F5", "MMRS-01", "MMRS-10", "F1", "F10" ], "sources": [ "B3", "MMRS", "B1" ], "earliest": "B3" }, { "lineage": "the vote and the score", "claims": [ "MMRS-11", "MMRS-03", "F12", "F21", "F14", "F35", "F15" ], "sources": [ "MMRS", "B3", "B5" ], "earliest": "B3" }, { "lineage": "the record's frame: an aggregator of reception work", "claims": [ "MMRS-24", "MMRS-23", "MMRS-14" ], "sources": [ "MMRS" ], "earliest": "MMRS" }, { "lineage": "the current board: snapshots and dates", "claims": [ "F2", "F7", "F32", "F33", "F41", "F40", "F36", "F37", "F39", "MMRS-12", "MMRS-20", "CAP-1-02", "F38", "CAP-1-03", "CAP-1-04" ], "sources": [ "B1", "B3", "B5", "B6", "MMRS", "capture:search-arena-leaderboard-aio-20261029" ], "earliest": "B3" }, { "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "claims": [ "F3", "F4", "F43", "F46", "F47", "F23", "MMRS-09", "MMRS-19" ], "sources": [ "B2", "B8", "B3", "MMRS" ], "earliest": "B3" }, { "lineage": "the rated systems: search-augmented LLMs", "claims": [ "F6", "F13", "F20", "F25", "F30" ], "sources": [ "B3", "B4" ], "earliest": "B3" }, { "lineage": "genealogy: the launch, the paper, the dataset", "claims": [ "F11", "F22", "F34", "F24" ], "sources": [ "B3", "B5", "B4" ], "earliest": "B3" }, { "lineage": "the vote: relative preference", "claims": [ "D741-01", "D741-02" ], "sources": [ 741 ], "earliest": 741 }, { "lineage": "the dataset: interactions and votes", "claims": [ "F26", "F27", "F31", "MMRS-04", "MMRS-08", "MMRS-15" ], "sources": [ "B4", "MMRS" ], "earliest": "B4" }, { "lineage": "the citation finding: preference follows citation count", "claims": [ "MMRS-17", "F28", "MMRS-02", "MMRS-05", "F9", "F16", "F17", "F18", "F8" ], "sources": [ "MMRS", "B4", "B3" ], "earliest": "B3" }, { "lineage": "sources cited and preferred", "claims": [ "F29", "MMRS-06", "F19" ], "sources": [ "B4", "MMRS", "B3" ], "earliest": "B3" }, { "lineage": "what the vote rewards", "claims": [ "MMRS-07", "MMRS-16", "MMRS-18" ], "sources": [ "MMRS" ], "earliest": "MMRS" }, { "lineage": "the rated systems: citation as proxy", "claims": [ "O741-01", "O741-02" ], "sources": [ 741 ], "earliest": 741 }, { "lineage": "the lifecycle kept apart from the observation", "claims": [ "MMRS-22", "MMRS-13" ], "sources": [ "MMRS" ], "earliest": "MMRS" }, { "lineage": "rank as attribute", "claims": [ "O1705-01", "O1705-02", "O1705-03" ], "sources": [ 1705 ], "earliest": 1705 }, { "lineage": "Arena's method and its limits", "claims": [ "F44", "F45" ], "sources": [ "B8" ], "earliest": "B8" }, { "lineage": "reception: the composition layer", "claims": [ "CAP-1-06", "CAP-1-01", "CAP-1-07", "CAP-1-05", "MMRS-21", "CAP-1-08" ], "sources": [ "capture:search-arena-leaderboard-aio-20261029", "MMRS" ], "earliest": "capture:search-arena-leaderboard-aio-20261029", "earliest_note": "the capture (2026-10-29) is the lineage's primary instance; the MMRS relation (MMRS-21) was added the same day and points to it" }, { "lineage": "reception: the encyclopedia", "claims": [ "F42", "F48" ], "sources": [ "B7", "B8" ], "earliest": "B7" } ], "body_words": 348, "entity": "Composed under v0.9, equal case (ρ = extends): KO_B sentences surviving 0 of 10; every field claim carried." }, "P_B": { "title": "Search Arena", "arm": "L_F(B), the field's ontology", "genre": "the same grammar, composed from the field alone, in the field's ontology", "status": "composed 2026-10-29 from the field ledger alone; not frozen", "lede": { "text": "is a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference, launched by Arena in March 2025 (Arena blog), with a dataset of over 24,000 paired interactions (Miroyan et al.).", "claims": [ "F5", "F11", "F26" ] }, "sections": [ { "icon": "🗳️", "head": "How it ranks", "items": [ { "label": "Vote:", "text": "two anonymous answers, one vote (Wikipedia); citation styles randomized (Arena blog).", "claims": [ "F43", "F12" ] }, { "label": "Score:", "text": "Bradley-Terry on the Elo scale, the \"Arena Score\" (Arena blog; dataset).", "claims": [ "F14", "F35" ] } ] }, { "icon": "📚", "head": "Findings", "items": [ { "label": "Citations:", "text": "preference follows citation count even when citations do not support the claims (Miroyan et al.).", "claims": [ "F28", "F9" ] }, { "label": "Sources:", "text": "community platforms preferred (Miroyan et al.).", "claims": [ "F29" ] } ] }, { "icon": "📅", "head": "The board", "items": [ { "label": "Latest:", "text": "13 October 2026, 42 models, gpt-5.6-sol-xhigh first (Arena dataset).", "claims": [ "F36", "F37" ] }, { "label": "Operator:", "text": "Arena, formerly LMArena and Chatbot Arena (Wikipedia).", "claims": [ "F43", "F46" ] } ] } ], "open": [ { "question": "When was it released?", "rests_on": [ "F11", "F34" ], "modality": "documented, two dates", "would_resolve": "Arena's leaderboard changelog" }, { "question": "Opacities", "rests_on": [ "F2", "F4" ], "modality": "—", "would_resolve": "the cards' pages read verbatim" } ], "entity": "The field's entity: a human-preference evaluation platform and leaderboard for search-augmented LLMs, run by Arena, with its dataset paper (ICLR 2026) and its published board history", "body_words": 100, "rail": [ { "lineage": "what it is: a leaderboard of search models by human vote", "claims": [ "F5", "F10", "F1" ], "sources": [ "B3", "B1" ], "earliest": "B3" }, { "lineage": "the rated systems: search-augmented LLMs", "claims": [ "F6", "F13", "F20", "F25", "F30" ], "sources": [ "B3", "B4" ], "earliest": "B3" }, { "lineage": "the current board: snapshots and dates", "claims": [ "F2", "F7", "F41", "F32", "F33", "F40", "F36", "F37", "F39", "F38" ], "sources": [ "B1", "B3", "B6", "B5" ], "earliest": "B3" }, { "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "claims": [ "F3", "F4", "F43", "F23", "F46", "F47" ], "sources": [ "B2", "B8", "B3" ], "earliest": "B3" }, { "lineage": "genealogy: the launch, the paper, the dataset", "claims": [ "F11", "F22", "F34", "F24" ], "sources": [ "B3", "B5", "B4" ], "earliest": "B3" }, { "lineage": "the vote and the score", "claims": [ "F12", "F14", "F35", "F15", "F21" ], "sources": [ "B3", "B5" ], "earliest": "B3" }, { "lineage": "the dataset: interactions and votes", "claims": [ "F26", "F27", "F31" ], "sources": [ "B4" ], "earliest": "B4" }, { "lineage": "the citation finding: preference follows citation count", "claims": [ "F28", "F9", "F16", "F17", "F18", "F8" ], "sources": [ "B4", "B3" ], "earliest": "B3" }, { "lineage": "sources cited and preferred", "claims": [ "F29", "F19" ], "sources": [ "B4", "B3" ], "earliest": "B3" }, { "lineage": "Arena's method and its limits", "claims": [ "F44", "F45" ], "sources": [ "B8" ], "earliest": "B8" }, { "lineage": "reception: the encyclopedia", "claims": [ "F42", "F48" ], "sources": [ "B7", "B8" ], "earliest": "B7" } ] }, "KO_B": { "title": "Search Arena", "arm": "L_F(B), the field's ontology", "genre": "the field alone", "status": "composed 2026-10-29; not frozen", "sentences": [ { "n": 1, "para": 0, "sense": "what it is", "text": "Search Arena is a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference, set apart from LM-Arena and SimpleQA by its focus on current events and diverse real-world use cases, with two aims, crowd-sourced evaluation of search-augmented LLMs and a released in-the-wild dataset of user–system interactions (Arena blog).", "claims": [ "F5", "F6", "F10" ] }, { "n": 2, "para": 0, "sense": "what it is", "text": "Its page is the 'AI Search Leaderboard: Top Models for Research', overall rankings across LLMs, within Arena's 'Official AI Model Rankings & Benchmarks', whose overview ranks models in the Agent, WebDev and Image Edit arenas (Arena AI cards).", "claims": [ "F1", "F2", "F3", "F4" ] }, { "n": 3, "para": 1, "sense": "origin", "text": "Launched on 18 March 2025 with 11 models from Perplexity, Gemini and OpenAI (OpenAI's web search API, not ChatGPT's search feature), it had over 11k votes, filtered to 7k battles, by the post of 15 April 2025, which put Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro at the top (Arena blog); Arena's dataset dates its search data from 7 August 2025 (Arena Leaderboard Dataset).", "claims": [ "F11", "F13", "F20", "F7", "F22", "F34" ] }, { "n": 4, "para": 1, "sense": "operator", "text": "Its operator is Arena (Arena Intelligence Inc.), formerly LMArena and Chatbot Arena, a public web platform that evaluates LLMs by anonymous pairwise votes, released in April 2023, incorporated in April 2025, renamed in January 2026 and valued at $3.1 billion in October 2026 (Wikipedia; Arena blog).", "claims": [ "F43", "F23", "F46", "F47" ] }, { "n": 5, "para": 2, "sense": "the vote and the score", "text": "Users vote between two anonymous responses, shown with standardized or original citation styles; a Bradley-Terry model gives scores on the Elo scale, the 'Arena Score', each model's battles counted, and at launch the confidence intervals were still wide (Arena blog; Arena Leaderboard Dataset); submissions and public voting are invited (Arena blog).", "claims": [ "F12", "F14", "F35", "F15", "F21" ] }, { "n": 6, "para": 3, "sense": "the dataset", "text": "The paper by Miroyan et al. (arXiv:2506.05334, 2025–2026, ICLR 2026) introduces Search Arena as a crowd-sourced human-preference dataset of over 24,000 paired multi-turn interactions with search-augmented LLMs, models that combine web search with LLMs, with full system traces and around 12,000 votes, open-sourced (Miroyan et al.).", "claims": [ "F24", "F25", "F26", "F27", "F31" ] }, { "n": 7, "para": 4, "sense": "the citation finding", "text": "User preferences are influenced by the number of citations even when the cited content does not directly support the attributed claims, a gap between perceived and actual credibility; community-driven platforms are generally preferred, static encyclopedic sources are not always appropriate and reliable; and web search does not degrade non-search performance (Miroyan et al.).", "claims": [ "F28", "F29", "F30" ] }, { "n": 8, "para": 4, "sense": "the citation finding", "text": "The launch post found response length and citation count significant under joint controls, rankings converging when citation count is controlled, citation style of minimal effect, and authoritative sources favored by all models (Arena blog).", "claims": [ "F9", "F16", "F17", "F18", "F8", "F19" ] }, { "n": 9, "para": 5, "sense": "the current board", "text": "Arena publishes its leaderboards' history as a dataset (Arena blog, 2 April 2026): the latest search board, published 13 October 2026, ranks 42 models with gpt-5.6-sol-xhigh (openai) first at 1261.1, qwen3.8-max second and claude-opus-4-6-search third, gpt-6.1-sol-max second with style control; the snapshot of 26 February 2026 had claude-opus-4-6-search first at 1255.4 (Arena Leaderboard Dataset).", "claims": [ "F41", "F32", "F33", "F40", "F36", "F37", "F39", "F38" ] }, { "n": 10, "para": 6, "sense": "limits and the encyclopedia", "text": "Research has identified specific limitations in Arena's methodology, among them the Llama 4 Maverick version that differed from the public one; Wikipedia has no article on Search Arena and its article on Arena does not name it (Wikipedia).", "claims": [ "F44", "F45", "F42", "F48" ] } ] }, "L_B": { "words": 483, "text": "Search Arena is a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference, set apart from LM-Arena and SimpleQA by its focus on current events and diverse real-world use cases, with two aims, crowd-sourced evaluation of search-augmented LLMs and a released in-the-wild dataset of user–system interactions (Arena blog). Its page is the 'AI Search Leaderboard: Top Models for Research', overall rankings across LLMs, within Arena's 'Official AI Model Rankings & Benchmarks', whose overview ranks models in the Agent, WebDev and Image Edit arenas (Arena AI cards). Launched on 18 March 2025 with 11 models from Perplexity, Gemini and OpenAI (OpenAI's web search API, not ChatGPT's search feature), it had over 11k votes, filtered to 7k battles, by the post of 15 April 2025, which put Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro at the top (Arena blog); Arena's dataset dates its search data from 7 August 2025 (Arena Leaderboard Dataset). Its operator is Arena (Arena Intelligence Inc.), formerly LMArena and Chatbot Arena, a public web platform that evaluates LLMs by anonymous pairwise votes, released in April 2023, incorporated in April 2025, renamed in January 2026 and valued at $3.1 billion in October 2026 (Wikipedia; Arena blog). Users vote between two anonymous responses, shown with standardized or original citation styles; a Bradley-Terry model gives scores on the Elo scale, the 'Arena Score', each model's battles counted, and at launch the confidence intervals were still wide (Arena blog; Arena Leaderboard Dataset); submissions and public voting are invited (Arena blog). The paper by Miroyan et al. (arXiv:2506.05334, 2025–2026, ICLR 2026) introduces Search Arena as a crowd-sourced human-preference dataset of over 24,000 paired multi-turn interactions with search-augmented LLMs, models that combine web search with LLMs, with full system traces and around 12,000 votes, open-sourced (Miroyan et al.). User preferences are influenced by the number of citations even when the cited content does not directly support the attributed claims, a gap between perceived and actual credibility; community-driven platforms are generally preferred, static encyclopedic sources are not always appropriate and reliable; and web search does not degrade non-search performance (Miroyan et al.). The launch post found response length and citation count significant under joint controls, rankings converging when citation count is controlled, citation style of minimal effect, and authoritative sources favored by all models (Arena blog). Arena publishes its leaderboards' history as a dataset (Arena blog, 2 April 2026): the latest search board, published 13 October 2026, ranks 42 models with gpt-5.6-sol-xhigh (openai) first at 1261.1, qwen3.8-max second and claude-opus-4-6-search third, gpt-6.1-sol-max second with style control; the snapshot of 26 February 2026 had claude-opus-4-6-search first at 1255.4 (Arena Leaderboard Dataset). Research has identified specific limitations in Arena's methodology, among them the Llama 4 Maverick version that differed from the public one; Wikipedia has no article on Search Arena and its article on Arena does not name it (Wikipedia).", "note": "L_F(B) is realized as the field arm's expansion (objects.KO_B): the field alone, cards, own pages and Wikipedia; its compression is objects.P_B" }, "L_BA": { "words": 1587, "text": "Search Arena is a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference (Arena's launch post), and in the MMRS record, which carries it as a row of the MMRS meta-aggregator's census, it is the leaderboard that ranks search-enabled language models by head-to-head human votes, its unit the search-enabled model rated from those votes, its kind a leaderboard and its role evaluation (MMRS record). Its page on arena.ai is titled 'AI Search Leaderboard: Top Models for Research' and offers overall rankings across LLMs (its card, Arena AI), one board among Arena's 'Official AI Model Rankings & Benchmarks', whose overview card ranks models in other arenas, Agent, WebDev and Image Edit among them (Arena AI). Arena's launch post gives it two aims, to 'enable crowd-sourced evaluation of search-augmented LLMs' and to 'release a diverse, in-the-wild dataset of user–system interactions', and sets its data apart from LM-Arena's and SimpleQA's by a focus on current events and diverse real-world use cases (Arena blog). It launched on 18 March 2025 and had collected over 11k votes across 10+ models, filtered to 7k battles, by the post of 15 April 2025, updated 2 August 2025 and first published at blog.lmarena.ai (Arena blog); Arena's leaderboard dataset dates its search data from the release on lmarena.ai on 7 August 2025 (Arena Leaderboard Dataset). At launch it supported 11 models from Perplexity, Gemini and OpenAI, OpenAI's web search API evaluated in place of the search feature in the ChatGPT product, and its first board, on 7k votes from 18 March to 13 April, put Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro at the top (Arena blog). Its operator is Arena, formerly LMArena and Chatbot Arena, 'a public, web-based platform that evaluates large language models (LLMs)', released as Chatbot Arena on 24 April 2023, incorporated as an independent company (as LMArena) in April 2025, renamed Arena on 28 January 2026, and valued at $3.1 billion in a $200 million Series B announced on 8 October 2026 (Wikipedia); the site's footer names Arena Intelligence Inc. (Arena blog), and the MMRS record gives the operator as arena.ai on the operator's own statement and leaves the operator statement on the leaderboard page unchecked (MMRS record). A vote is a battle: on Arena users put a prompt to two anonymous models and vote for the better response, after which the models' identities are revealed (Wikipedia), here two search-augmented answers compared, with models rated with confidence intervals from the battles (MMRS record); each response is displayed in a standardized citation format or in the provider's original one, by 'style randomization', and Arena invites model submissions and encourages public voting (Arena blog). Scores come from a Bradley-Terry model whose coefficients are translated to the Elo scale, consistent with LM Arena (Arena blog); Arena's dataset calls the rating the 'Arena Score' and its vote_count the 'Number of battles for this model' (Arena Leaderboard Dataset); at launch Arena wrote that 'The confidence intervals are still wide, which means the leaderboard hasn’t fully settled and there’s still some uncertainty' (Arena blog). Of the MT-Bench / Chatbot Arena studies, in which GPT-4 reached approximately 85% agreement with human preferences against approximately 81% human-human agreement, #741 states what such pairwise judgment measures: 'This is a finding about *relative* preference under specified rubrics — comparing A versus B — not about absolute quality assessment' (#741). The systems it rates are search-augmented language models, which 'combine web search with Large Language Models (LLMs) to improve response groundedness and freshness', and whose analysis, its paper says, has rested on datasets 'limited in scale and narrow in scope, often constrained to static, single-turn, fact-checking questions' (Miroyan et al.). Its paper, by Miroyan and ten co-authors (arXiv:2506.05334, submitted 5 June 2025, revised 2 March 2026, accepted to ICLR 2026), introduces Search Arena as 'a crowd-sourced, large-scale, human-preference dataset of over 24,000 paired multi-turn user interactions with search-augmented LLMs', spanning diverse intents and languages, with full system traces and around 12,000 human preference votes, open-sourced (Miroyan et al.); the MMRS record restates the release, over 24,000 paired multi-turn interactions with about 12,000 votes (MMRS record). The MMRS record reads the released interactions as original outputs, preserved, on the paper's statement (MMRS record). The paper's headline finding is that 'user preferences are influenced by the number of citations, even when the cited content does not directly support the attributed claims, uncovering a gap between perceived and actual credibility' (Miroyan et al.), which the MMRS record carries as preference rising with the number of citations even when the cited content does not support the claim it is attached to (MMRS record). The launch post had found three features strongly correlated with preference, response length, citation count and citing YouTube and online forums or blogs; controlling for citation count gave a positive, significant coefficient (0.234, p < 0.05), only response length and citation count stayed significant under joint controls, and with the number of citations controlled the rankings converged, while standardizing citation styles had minimal effect on rankings (Arena blog). Preference varies across cited sources: 'community-driven platforms are generally preferred and static encyclopedic sources are not always appropriate and reliable' (Miroyan et al.; restated by the MMRS record), while all models favor authoritative sources such as Wikipedia, .edu and .gov domains and OpenAI's models cite news sources for 51.3% overall (Arena blog). Across arenas, web search does not degrade and may even improve performance in non-search settings, and quality in search settings is significantly affected when a model relies solely on its parametric knowledge (Miroyan et al.). The MMRS record reads the board and the paper together: 'Its leaderboard rewards what users prefer, and its paper shows that preference rewarding citation count over citation support'; the paper analyses citations and the leaderboard ranks by votes, and whether the live leaderboard judges citations is left unchecked (MMRS record). #741 names the regime of the systems it rates: 'The current moment adds a new proxy on top of the inherited stack: provenance topology in generative search', and 'The composition layer of an AI search system does not read documents and assess their substance', admitting or excluding documents on structural signals, citation neighborhoods among them, before any reading occurs (#741). Arena publishes the board's history: its leaderboard dataset holds historical snapshots of the Arena leaderboard, a full split of all historically published leaderboards and a latest split of only the most recently published (Arena Leaderboard Dataset), released on 2 April 2026 as 'the entire history of those leaderboards' (Arena blog), its search boards published from 7 August 2025 to 13 October 2026 (Arena Leaderboard Dataset). The latest, published 13 October 2026, ranks 42 models: gpt-5.6-sol-xhigh (openai) first at 1261.1 on 60,437 battles, qwen3.8-max (alibaba) second at 1261.1, claude-opus-4-6-search (anthropic) third at 1251.4 and ernie-5.1 (baidu) eleventh at 1226.7; with style control gpt-6.1-sol-max (openai) stands second at 1249.9 (Arena Leaderboard Dataset). The MMRS record, which read the leaderboard page on 2026-10-28 for its counts and keeps the lifecycle apart from the observation that saw it, gives 247,944 votes across 22 models at the page's update of 25 February 2026, and the state active (MMRS record). The Capture Registry's fetcher read the page on 2026-10-29 as updated 25 February 2026, with 22 models and 247,944 votes, Anthropic first at 1255 ± 10 and xAI second at 1225 ± 8 (Capture Registry, 2026-10-29); Arena's dataset holds a snapshot of 26 February 2026 with claude-opus-4-6-search (anthropic) first at 1255.4 on 3,607 battles and grok-4.20-beta1 (xai) second at 1225.0 (Arena Leaderboard Dataset). #1705 reads a ranking as an act: 'it compares', selects the members of a category, projects selected features onto a common field, orders them, and 'states the result as a property of the winner', so that the comparison, 'its category, its criteria, its sample and its author, has become an attribute of the thing compared'; 'A ranking's *best* is attributization in its commercial form: the ranker's act spoken in the product's voice' (#1705). Wikipedia records that research has identified specific limitations in Arena's methodology, and that the version of Llama 4 Maverick on LMArena 'unfairly differed from the publicly available version', after which LMArena updated its policies (Wikipedia). Asked 'search arena leaderboard' on 29 October 2026, Google's AI Overview, on the registry's innocent reading, gave 'the top models and their ratings, with links to the boards'; the Capture Registry records 'No operator, no method (head-to-head human votes), no counts, and none of the dataset paper's findings on citation; the composition is a table', the Overview returning 'rankings and no account of the leaderboard' (Capture Registry, 2026-10-29). Against the page as read, 'The composed table's ratings, providers and order do not match it', and the registry, having checked against the MMRS record and the entity's own pages, leaves open which page is current: 'Google may hold a later render' (Capture Registry, 2026-10-29). The MMRS record, each of whose assertions carries a source and a basis, links that capture as 'the composition layer's account of the entity, seated in the Capture Registry', and reads admission to the board as not requiring harm (MMRS record). Wikipedia has no article at 'Search Arena', and its article on the operator, to which 'LMArena' and 'Chatbot Arena' redirect, does not name it (Wikipedia, 2026-10-29); of the entity and its composition the Capture Registry says 'The record keeps the entity; the capture keeps what was composed in its place' (Capture Registry, 2026-10-29).", "rail": [], "note": "L(B ∪ A) in the equal-terms case (v0.9 §0.4), realized as the knowledge object (objects.KO); its rail is the compression's, by lineage." } }, "field": [ { "id": "B1", "card": "Arena AI — AI Search Leaderboard: Top Models for Research", "fetched": "card; HTTP 403 (unreachable verbatim: card title and snippet stand); B1.html sha256 a94c489196949291", "origin": "card" }, { "id": "B2", "card": "Arena AI — Arena Leaderboard: Official AI Model Rankings & Benchmarks", "fetched": "card; HTTP 403 (unreachable verbatim: card title and snippet stand); B2.html sha256 e562ebae953ba5de", "origin": "card" }, { "id": "B3", "card": "Arena.ai blog — Introducing the Search Arena: Evaluating Search-Enabled AI", "fetched": "own_page; HTTP 200 (fetched); B3.html sha256 251720f2c64e59fb", "origin": "own_page" }, { "id": "B4", "card": "arXiv — [2506.05334] Search Arena: Analyzing Search-Augmented LLMs", "fetched": "own_page; HTTP 200 (fetched); B4.html sha256 b35a50c458405ec9", "origin": "own_page" }, { "id": "B5", "card": "Hugging Face (lmarena-ai) — Arena Leaderboard Dataset, subset search", "fetched": "own_page; HTTP 200 (fetched); B5.html sha256 729a6836c088af3e", "origin": "own_page" }, { "id": "B6", "card": "Arena.ai blog — Arena Leaderboard Dataset", "fetched": "own_page; HTTP 200 (fetched); B6.html sha256 945b31ff964c24cd", "origin": "own_page" }, { "id": "B7", "card": "Wikipedia — Search Arena (no article)", "fetched": "wikipedia; HTTP 404 (fetched); B7.html sha256 7b2ce79be00f677c", "origin": "wikipedia" }, { "id": "B8", "card": "Wikipedia — Arena.ai", "fetched": "wikipedia; HTTP 200 (fetched); B8.html sha256 75667836aeaf4811", "origin": "wikipedia" } ], "field_claims": { "F1": { "source": "B1", "origin": "card", "locus": "card title", "quote": "AI Search Leaderboard: Top Models for Research", "claim": "the leaderboard's page titled as an AI search leaderboard of top models for research", "modality": "stipulation", "kind": "definition", "kernel": true, "const": false, "lineage": "what it is: a leaderboard of search models by human vote", "date": "— (card title)", "quote_ok": true, "checked_against": "the card's title in the registry's cite_list" }, "F2": { "source": "B1", "origin": "card", "locus": "card snippet (truncated)", "quote": "View overall rankings across LLMs with inte", "claim": "the page offers overall rankings across LLMs (the snippet cut at 'inte…')", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "— (card snippet)", "quote_ok": true, "checked_against": "the card's snippet in the registry's cite_list" }, "F3": { "source": "B2", "origin": "card", "locus": "card title", "quote": "Arena Leaderboard: Official AI Model Rankings & Benchmarks", "claim": "Arena's leaderboard overview titled 'Official AI Model Rankings & Benchmarks'", "modality": "self-description", "kind": "definition", "kernel": true, "const": false, "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "date": "— (card title)", "quote_ok": true, "checked_against": "the card's title in the registry's cite_list" }, "F4": { "source": "B2", "origin": "card", "locus": "card snippet", "quote": "GPT-6.1 Terra is #8 in Agent · Max Kimi K3.1 Pro is #9 in WebDev · Max MAI-Image-2.7-Preview is #3 in Image Edit", "claim": "the overview states ranks of models in other arenas (Agent, WebDev, Image Edit); no search model", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "date": "— (card snippet)", "quote_ok": true, "checked_against": "the card's snippet in the registry's cite_list" }, "F5": { "source": "B3", "origin": "own_page", "locus": "TL;DR ¶1 (and the post's opening)", "quote": "We introduce Search Arena, a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference.", "claim": "Search Arena defined by its launch post: a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference", "modality": "stipulation", "kind": "definition", "kernel": true, "const": true, "lineage": "what it is: a leaderboard of search models by human vote", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F6": { "source": "B3", "origin": "own_page", "locus": "TL;DR ¶1", "quote": "Unlike LM-Arena or SimpleQA, our data focuses on current events and diverse real-world use cases", "claim": "its data, unlike LM-Arena's or SimpleQA's, focuses on current events and diverse real-world use cases", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "the rated systems: search-augmented LLMs", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F7": { "source": "B3", "origin": "own_page", "locus": "TL;DR ¶2", "quote": "Based on 7k human votes (03/18–04/13), Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro are at the top, followed by the rest of Perplexity’s Sonar models, Gemini-2.0-Flash-Grounding, and OpenAI’s web search API models.", "claim": "the first leaderboard, on 7k votes from 18 March to 13 April 2025: Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro at the top", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F8": { "source": "B3", "origin": "own_page", "locus": "TL;DR ¶2", "quote": "Standardizing citation styles had minimal effect on rankings (see Sec. 2).", "claim": "standardizing citation styles had minimal effect on rankings", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the citation finding: preference follows citation count", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F9": { "source": "B3", "origin": "own_page", "locus": "TL;DR ¶3", "quote": "Three features show strong positive correlation with human preference: response length, citation count, and citing specific web sources like YouTube and online forum/blogs (see Sec. 3).", "claim": "three features correlate strongly with preference: response length, citation count, citing YouTube and forums/blogs", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "the citation finding: preference follows citation count", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F10": { "source": "B3", "origin": "own_page", "locus": "§1 Why Search Arena?, ¶3", "quote": "To this end, we developed search arena, aiming to (1) enable crowd-sourced evaluation of search-augmented LLMs and (2) release a diverse, in-the-wild dataset of user–system interactions.", "claim": "its two aims: crowd-sourced evaluation of search-augmented LLMs and a released in-the-wild dataset of interactions", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "what it is: a leaderboard of search models by human vote", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F11": { "source": "B3", "origin": "own_page", "locus": "§1, ¶4", "quote": "Since our initial launch on March 18th, we’ve collected over 11k votes across 10+ models. We then filtered this data to construct 7k battles with user votes", "claim": "launched 18 March (2025); over 11k votes across 10+ models, filtered to 7k battles", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "genealogy: the launch, the paper, the dataset", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F12": { "source": "B3", "origin": "own_page", "locus": "§1.A Data, ¶1", "quote": "To balance these considerations, we introduced “style randomization”: responses are displayed either in a standardized format or in the original format (i.e., the citation style agreed upon with each model provider).", "claim": "style randomization: responses shown in a standardized or the provider's original citation format", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "the vote and the score", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F13": { "source": "B3", "origin": "own_page", "locus": "§1.B Models, ¶1", "quote": "Search Arena currently supports 11 models from three providers: Perplexity, Gemini, and OpenAI.", "claim": "at launch, 11 models from three providers: Perplexity, Gemini, OpenAI", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the rated systems: search-augmented LLMs", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F14": { "source": "B3", "origin": "own_page", "locus": "§2 Leaderboard, ¶3", "quote": "Consistent with LM Arena, we apply the Bradley-Terry (BT) model to compute model scores. The resulting BT coefficients are then translated to Elo scale", "claim": "the score: a Bradley-Terry model over battles, its coefficients translated to the Elo scale, as in LM Arena", "modality": "stipulation", "kind": "definition", "kernel": true, "const": false, "lineage": "the vote and the score", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F15": { "source": "B3", "origin": "own_page", "locus": "§2 Leaderboard, ¶3", "quote": "The confidence intervals are still wide, which means the leaderboard hasn’t fully settled and there’s still some uncertainty.", "claim": "its stated limit at launch: wide confidence intervals; the leaderboard not settled", "modality": "self-description", "kind": "limit", "kernel": true, "const": false, "lineage": "the vote and the score", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F16": { "source": "B3", "origin": "own_page", "locus": "§3 Control Experiments, 'Number of citations'", "quote": "Controlling for citation count also results in a positive and significant coefficient (0.234, p < 0.05), suggesting a preference for responses with more cited sources.", "claim": "citation count as a control: positive, significant coefficient (0.234, p < 0.05)", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the citation finding: preference follows citation count", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F17": { "source": "B3", "origin": "own_page", "locus": "§3 Control Experiments, 'Joint controls'", "quote": "When controlling for all features, only response length and citation count remain statistically significant.", "claim": "under joint controls only response length and citation count stay significant", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the citation finding: preference follows citation count", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F18": { "source": "B3", "origin": "own_page", "locus": "§3 Control Experiments, Tab. 5 note 2", "quote": "When controlling for the number of citations, model rankings converge (i.e., multiple models share the same rank), suggesting that the number of citations is a significant factor impacting differences across models and the resulting rankings.", "claim": "controlling for citation count, rankings converge: citation count drives differences in the rankings", "modality": "interpretation", "kind": "interpretation", "kernel": false, "const": false, "lineage": "the citation finding: preference follows citation count", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F19": { "source": "B3", "origin": "own_page", "locus": "§3 Model Characteristics, findings 1–2", "quote": "All models favor authoritative sources (e.g., Wikipedia, .edu, .gov domains). … OpenAI models heavily cite news sources—51.3% overall and 87.3% for Trump-related prompts.", "claim": "all models favor authoritative sources; OpenAI's cite news heavily (51.3%; 87.3% on Trump-related prompts)", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "sources cited and preferred", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F20": { "source": "B3", "origin": "own_page", "locus": "§1.B Models, Table 2 note", "quote": "†We evaluate OpenAI’s web search API, which is different from the search feature in the ChatGPT product.", "claim": "a limit: OpenAI's web search API is evaluated, not ChatGPT's search feature", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "the rated systems: search-augmented LLMs", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F21": { "source": "B3", "origin": "own_page", "locus": "§4 Conclusion & What's Next", "quote": "Open participation: We are inviting model submissions from researchers and industry, and encouraging public voting.", "claim": "open participation: model submissions invited, public voting encouraged", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "the vote and the score", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F22": { "source": "B3", "origin": "own_page", "locus": "header; Citation", "quote": "Published on 15 Apr 2025 … Last updated on 2 Aug 2025 … Mihran Miroyan*. (2025). Introducing the Search Arena: Evaluating Search-Enabled AI. https://blog.lmarena.ai/blog/2025/search-arena/", "claim": "the post published 15 April 2025, updated 2 August 2025, cited at its lmarena.ai address", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "genealogy: the launch, the paper, the dataset", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F23": { "source": "B3", "origin": "own_page", "locus": "footer", "quote": "© 2026 Arena Intelligence Inc.", "claim": "the site's operator: Arena Intelligence Inc.", "modality": "documented", "kind": "attribution", "kernel": false, "const": false, "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "date": "2025-04-15 (published; last updated 2025-08-02)", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 e0d10dc29692ef83…), whitespace-normalised" }, "F24": { "source": "B4", "origin": "own_page", "locus": "abstract page header", "quote": "[Submitted on 5 Jun 2025 (v1), last revised 2 Mar 2026 (this version, v2)] … Authors:Mihran Miroyan, Tsung-Han Wu, Logan King, Tianle Li, Jiayi Pan, Xinyan Hu, Wei-Lin Chiang, Anastasios N. Angelopoulos, Trevor Darrell, Narges Norouzi, Joseph E. Gonzalez", "claim": "the paper: Miroyan et al. (eleven authors), v1 5 June 2025, v2 2 March 2026", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "genealogy: the launch, the paper, the dataset", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F25": { "source": "B4", "origin": "own_page", "locus": "Abstract s1–s2", "quote": "Search-augmented language models combine web search with Large Language Models (LLMs) to improve response groundedness and freshness. However, analyzing these systems remains challenging: existing datasets are limited in scale and narrow in scope, often constrained to static, single-turn, fact-checking questions.", "claim": "the class and the problem: search-augmented LLMs combine web search with LLMs; existing datasets are small, narrow, static, single-turn", "modality": "interpretation", "kind": "interpretation", "kernel": false, "const": false, "lineage": "the rated systems: search-augmented LLMs", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F26": { "source": "B4", "origin": "own_page", "locus": "Abstract s3", "quote": "In this work, we introduce Search Arena, a crowd-sourced, large-scale, human-preference dataset of over 24,000 paired multi-turn user interactions with search-augmented LLMs.", "claim": "Search Arena defined by its paper: a crowd-sourced human-preference dataset of over 24,000 paired multi-turn interactions with search-augmented LLMs", "modality": "stipulation", "kind": "definition", "kernel": true, "const": true, "lineage": "the dataset: interactions and votes", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F27": { "source": "B4", "origin": "own_page", "locus": "Abstract s4", "quote": "The dataset spans diverse intents and languages, and contains full system traces with around 12,000 human preference votes.", "claim": "diverse intents and languages, full system traces, around 12,000 preference votes", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the dataset: interactions and votes", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F28": { "source": "B4", "origin": "own_page", "locus": "Abstract s5", "quote": "Our analysis reveals that user preferences are influenced by the number of citations, even when the cited content does not directly support the attributed claims, uncovering a gap between perceived and actual credibility.", "claim": "headline finding: preference follows the number of citations even when the cited content does not support the attributed claims; a gap between perceived and actual credibility", "modality": "interpretation", "kind": "interpretation", "kernel": true, "const": false, "lineage": "the citation finding: preference follows citation count", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F29": { "source": "B4", "origin": "own_page", "locus": "Abstract s6", "quote": "Furthermore, user preferences vary across cited sources, revealing that community-driven platforms are generally preferred and static encyclopedic sources are not always appropriate and reliable.", "claim": "preference varies by cited source: community-driven platforms generally preferred; static encyclopedic sources not always appropriate and reliable", "modality": "interpretation", "kind": "interpretation", "kernel": false, "const": false, "lineage": "sources cited and preferred", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F30": { "source": "B4", "origin": "own_page", "locus": "Abstract s7–s8", "quote": "We find that web search does not degrade and may even improve performance in non-search settings; however, the quality in search settings is significantly affected if solely relying on the model's parametric knowledge.", "claim": "cross-arena finding: web search does not degrade non-search performance; parametric knowledge alone hurts search settings", "modality": "interpretation", "kind": "interpretation", "kernel": false, "const": false, "lineage": "the rated systems: search-augmented LLMs", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F31": { "source": "B4", "origin": "own_page", "locus": "Abstract s9; Comments", "quote": "We open-sourced the dataset to support future research. … Accepted to ICLR 2026.", "claim": "the dataset open-sourced; the paper accepted to ICLR 2026", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the dataset: interactions and votes", "date": "2025-06-05 (v1; v2 2026-03-02)", "quote_ok": true, "checked_against": "saved text B4.txt (sha256 22c03d0fef7adb35…), whitespace-normalised" }, "F32": { "source": "B5", "origin": "own_page", "locus": "README, title and opening", "quote": "Historical snapshots of the [Arena](https://arena.ai/leaderboard) leaderboard.", "claim": "Arena's leaderboard dataset: historical snapshots of the Arena leaderboard", "modality": "stipulation", "kind": "definition", "kernel": true, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-10-29 (dataset last modified 2026-10-29T03:05Z)", "quote_ok": true, "checked_against": "saved text B5.README.md (sha256 b13bf3169c5ec5ff…), whitespace-normalised" }, "F33": { "source": "B5", "origin": "own_page", "locus": "README, Splits", "quote": "- `full`: All historically published leaderboards - `latest`: Only the most recently published leaderboards", "claim": "two splits: full, every published leaderboard; latest, only the most recent", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-10-29 (dataset last modified)", "quote_ok": true, "checked_against": "saved text B5.README.md (sha256 b13bf3169c5ec5ff…), whitespace-normalised" }, "F34": { "source": "B5", "origin": "own_page", "locus": "README, Notes", "quote": "- Search and Webdev Arena data starts from their releases on the arena.ai (then lmarena.ai) on August 7 and November 12 2025 respectively.", "claim": "the search data starts from its release on lmarena.ai on 7 August 2025", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "genealogy: the launch, the paper, the dataset", "date": "2026-10-29 (dataset last modified)", "quote_ok": true, "checked_against": "saved text B5.README.md (sha256 b13bf3169c5ec5ff…), whitespace-normalised" }, "F35": { "source": "B5", "origin": "own_page", "locus": "README, Schema", "quote": "| rating | float | Arena Score | … | vote_count | int | Number of battles for this model |", "claim": "the schema: rating is the 'Arena Score'; vote_count the number of battles for the model", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "the vote and the score", "date": "2026-10-29 (dataset last modified)", "quote_ok": true, "checked_against": "saved text B5.README.md (sha256 b13bf3169c5ec5ff…), whitespace-normalised" }, "F36": { "source": "B5", "origin": "own_page", "locus": "search, split latest: rows rank 1 and 11; num_rows_total", "quote": "{\"model_name\":\"gpt-5.6-sol-xhigh\",\"organization\":\"openai\",\"license\":\"Proprietary\",\"rating\":1261.1307593350664,\"rating_lower\":1254.8406048513853,\"rating_upper\":1267.4209138187477,\"variance\":10.299744257424065,\"vote_count\":60437,\"rank\":1,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-10-13\"} … {\"model_name\":\"ernie-5.1\",\"organization\":\"baidu\",\"license\":\"Proprietary\",\"rating\":1226.7150529766889 … \"rank\":11,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-10-13\"} … \"num_rows_total\":42", "claim": "the latest published search leaderboard, 13 October 2026: 42 models; first gpt-5.6-sol-xhigh (openai) at 1261.1, 60,437 battles; eleventh ernie-5.1 (baidu) at 1226.7", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-10-13 (publish date)", "quote_ok": true, "checked_against": "saved text B5.hf-search-latest.json (sha256 f5ee84f9d2bd1b58…), whitespace-normalised" }, "F37": { "source": "B5", "origin": "own_page", "locus": "search, split latest: rows rank 2 and 3", "quote": "{\"model_name\":\"qwen3.8-max\",\"organization\":\"alibaba\",\"license\":\"Proprietary\",\"rating\":1261.0761861904762 … {\"model_name\":\"claude-opus-4-6-search\",\"organization\":\"anthropic\",\"license\":\"Proprietary\",\"rating\":1251.3682634708416", "claim": "second qwen3.8-max (alibaba) at 1261.1, third claude-opus-4-6-search (anthropic) at 1251.4: the first two within a tenth of a point", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-10-13 (publish date)", "quote_ok": true, "checked_against": "saved text B5.hf-search-latest.json (sha256 f5ee84f9d2bd1b58…), whitespace-normalised" }, "F38": { "source": "B5", "origin": "own_page", "locus": "search, split full: snapshot of 2026-02-26, rows rank 1 and 2", "quote": "{\"model_name\":\"claude-opus-4-6-search\",\"organization\":\"anthropic\",\"license\":\"Proprietary\",\"rating\":1255.4140677086834,\"rating_lower\":1245.0272537380415,\"rating_upper\":1265.8008816793258,\"variance\":28.084618239181463,\"vote_count\":3607,\"rank\":1,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-02-26\"} … {\"model_name\":\"grok-4.20-beta1\",\"organization\":\"xai\",\"license\":\"Proprietary\",\"rating\":1225.0104803795257", "claim": "the snapshot of 26 February 2026: first claude-opus-4-6-search (anthropic) at 1255.4 ± 10.4, 3,607 battles; second grok-4.20-beta1 (xai) at 1225.0", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-02-26 (publish date)", "quote_ok": true, "checked_against": "saved text B5.hf-search-full-200.json (sha256 803334809b94eca0…), whitespace-normalised" }, "F39": { "source": "B5", "origin": "own_page", "locus": "search_style_control, split latest: row rank 2", "quote": "{\"model_name\":\"gpt-6.1-sol-max\",\"organization\":\"openai\",\"license\":\"Proprietary\",\"rating\":1249.9181501199778,\"rating_lower\":1240.895777035074,\"rating_upper\":1258.9405232048819,\"variance\":21.190703814049858,\"vote_count\":5555,\"rank\":2,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-10-13\"}", "claim": "with style control the order differs: second gpt-6.1-sol-max (openai) at 1249.9 (eighth without it)", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-10-13 (publish date)", "quote_ok": true, "checked_against": "saved text B5.hf-search_style_control-latest.json (sha256 bec4d22199bd44aa…), whitespace-normalised" }, "F40": { "source": "B5", "origin": "own_page", "locus": "search, split full: statistics, leaderboard_publish_date", "quote": "\"min\":\"2025-08-07 00:00:00\",\"max\":\"2026-10-13 00:00:00\"", "claim": "the published search leaderboards run from 7 August 2025 to 13 October 2026", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-10-13 (latest publish date)", "quote_ok": true, "checked_against": "saved text B5.hf-search-full-dates.json (sha256 7df292de7bae2ddf…), whitespace-normalised" }, "F41": { "source": "B6", "origin": "own_page", "locus": "¶1", "quote": "For almost three years, Arena has been publishing leaderboards covering frontier AI capabilities across 10 arenas, dozens of categories, and hundreds of models, and today we're releasing the entire history of those leaderboards as a public-access dataset.", "claim": "Arena releases the entire history of its published leaderboards as a public dataset (10 arenas)", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "the current board: snapshots and dates", "date": "2026-04-02 (published)", "quote_ok": true, "checked_against": "saved text B6.txt (sha256 e91db567eb6ab2c9…), whitespace-normalised" }, "F42": { "source": "B7", "origin": "wikipedia", "locus": "rendered page (HTTP 404)", "quote": "Wikipedia does not have an article with this exact name.", "claim": "no Wikipedia article at 'Search Arena' (HTTP 404, rendered and raw)", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "reception: the encyclopedia", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B7.txt (sha256 5b819aeaa91b2097…), whitespace-normalised" }, "F43": { "source": "B8", "origin": "wikipedia", "locus": "lead ¶1", "quote": "Arena (formerly LMArena and Chatbot Arena) is a public, web-based platform that evaluates large language models (LLMs). Users enter prompts for two anonymous models to respond to and vote on the model that gave the better response, after which the models' identities are revealed.", "claim": "Wikipedia: the operator, Arena (formerly LMArena and Chatbot Arena), a public web platform evaluating LLMs by anonymous pairwise votes", "modality": "documented", "kind": "definition", "kernel": true, "const": false, "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B8.txt (sha256 15e7f68eb70ffd42…), whitespace-normalised" }, "F44": { "source": "B8", "origin": "wikipedia", "locus": "lead ¶4", "quote": "Research has identified specific limitations in Arena's methodology.", "claim": "Wikipedia: research has identified limitations in Arena's methodology (refs: rigged votes, Fast Company 2025; TechCrunch 2024)", "modality": "attributed", "kind": "limit", "kernel": true, "const": false, "lineage": "Arena's method and its limits", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B8.txt (sha256 15e7f68eb70ffd42…), whitespace-normalised" }, "F45": { "source": "B8", "origin": "wikipedia", "locus": "History ¶4", "quote": "Llama 4 Maverick beat GPT-4o and Gemini 2.0 Flash on LMArena, but the version of Maverick on LMArena unfairly differed from the publicly available version. LMArena updated their policies in response.", "claim": "Wikipedia: the Llama 4 Maverick episode (April 2025) and the policy change", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "Arena's method and its limits", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B8.txt (sha256 15e7f68eb70ffd42…), whitespace-normalised" }, "F46": { "source": "B8", "origin": "wikipedia", "locus": "History ¶1, ¶5, ¶8", "quote": "Chatbot Arena was released on April 24, 2023. … In April 2025, LMArena incorporated as an independent company. … On January 28, 2026, LMArena rebranded to \"Arena\".", "claim": "Wikipedia: Chatbot Arena released 24 April 2023; LMArena incorporated April 2025; rebranded Arena 28 January 2026", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B8.txt (sha256 15e7f68eb70ffd42…), whitespace-normalised" }, "F47": { "source": "B8", "origin": "wikipedia", "locus": "History ¶9", "quote": "On October 8, 2026, Arena announced it raised $200 million in a Series B funding round co-led by Lightspeed Venture Partners and Khosla Ventures, thus placing the company at a $3.1 billion valuation.", "claim": "Wikipedia: Series B of $200 million, 8 October 2026, at a $3.1 billion valuation", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B8.txt (sha256 15e7f68eb70ffd42…), whitespace-normalised" }, "F48": { "source": "B8", "origin": "wikipedia", "locus": "redirect pages LMArena and Chatbot Arena (raw wikitext)", "quote": "#REDIRECT [[Arena.ai]]", "claim": "Wikipedia: 'LMArena' and 'Chatbot Arena' redirect to the operator's article, which does not mention Search Arena", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "reception: the encyclopedia", "date": "2026-10-29 (as fetched)", "quote_ok": true, "checked_against": "saved text B8-redirect-LMArena.wiki (sha256 5f608ef8b6839e0f…), whitespace-normalised" } }, "field_sources": { "B1": "Arena AI — AI Search Leaderboard: Top Models for Research", "B2": "Arena AI — Arena Leaderboard: Official AI Model Rankings & Benchmarks", "B3": "Arena.ai blog — Introducing the Search Arena: Evaluating Search-Enabled AI", "B4": "arXiv — [2506.05334] Search Arena: Analyzing Search-Augmented LLMs", "B5": "Hugging Face (lmarena-ai) — Arena Leaderboard Dataset, subset search", "B6": "Arena.ai blog — Arena Leaderboard Dataset", "B7": "Wikipedia — Search Arena (no article)", "B8": "Wikipedia — Arena.ai" }, "delta": { "T_vs_LB": "One transcript, two cards, both Arena's (B1 the Search leaderboard, B2 the leaderboard overview), both unreachable verbatim here and admitted by title and snippet. What T composes from its own cards: the pointer to 'the AI Search Leaderboard and the main Arena Leaderboard' (T1-1) is the two cards' titles (F1, F3); 'live model rankings' (T1-1) is the card's 'overall rankings across LLMs' (F2), with 'live' added: no field source calls the board live, and the latest published board is dated 13 October 2026 (F36), sixteen days before the capture (force raised, mildly; the date, a limit of the figures, dropped). The table (T1-2 to T1-13): its eleven rows are, rank for rank, Arena's latest published search board (B5, published 2026-10-13): the same models (display names against the data's slugs), the same organizations, the ratings rounded and the confidence half-widths rounded (delta.alignment.table_check, computed from the saved JSON; ranks 1–3 and 11 ledgered as F36, F37). The figures are composed at their source's resolution and attributed by heading to 'the AI Search Leaderboard' (card 1); their verbatim instance in the field is the leaderboard's own data, which T did not surface. The column 'Elo Rating' is the launch post's account of the scale (F14) where the data names the rating 'Arena Score' (F35): a substituted reading at the 2025 post's grain. The selection is the top eleven of 42 (F36: 'num_rows_total':42), the count not composed; the board without style control is given without saying so, and with style control its second place differs (F39). Available, not composed: what the entity is (F5, F10, F6, F26), the vote and the score (F43, F12, F14, F35, F15, F21), the dataset (F24, F26, F27, F31), the citation finding and its controls (F28, F9, F16, F17, F18, F8), the sources preferred (F29, F19), the rated systems (F25, F13, F20, F30), the history and dates (F7, F11, F22, F32, F33, F34, F38, F40, F41), the operator (F43, F46, F47, F23) and its limits (F44, F45), the encyclopedia's absence (F42, F48). Card 2's own content, ranks in other arenas (F4), is not composed; the closer offers other categories (T1-14, content replaced by use). Against the page as the registry read it, the table does not match (CAP-1-03); against the board's own latest data it does: the composition carries the later published state.", "T_vs_LB_reading": "v0.9 §0.5: the first diff is already ontological, and it is total at the grain of the entity: in the field's ontology Search Arena is a platform, a vote, a score, a dataset and a finding on citation, with an operator and dated boards; in T's it is its current table. T keeps the board's figures at their own resolution and drops every other sense the field holds, the date of the figures among them. The row records this as observed and claims nothing about the composer's internals (§0.5).", "LB_vs_LBA": "Admission adds, at the archive's own force and in one claim space with the field's: the record's definition at its grain, the leaderboard ranking search-enabled models by head-to-head human votes, its unit the model rated from those votes (MMRS-01, MMRS-11, MMRS-10), and its frame, a row of the meta-aggregator's census whose assertions carry source and basis, admission not requiring harm (MMRS-24, MMRS-23, MMRS-14); what the vote rewards, 'Its leaderboard rewards what users prefer, and its paper shows that preference rewarding citation count over citation support' (MMRS-07), the paper analysing citations while the board ranks by votes (MMRS-16), whether the live board judges citations left unchecked (MMRS-18); what pairwise judgment measures, 'relative preference … not … absolute quality' (#741, D741-01, D741-02); the systems it rates under a proxy regime, the composition layer of AI search admitting documents on structural signals before any reading (#741, O741-01, O741-02); a ranking's result stated as a property of the winner, attributization (#1705, O1705-01–03); the count as read and its date kept apart from the lifecycle (MMRS-12, MMRS-20, MMRS-13, MMRS-22); the operator on its own statement, the page's statement unchecked (MMRS-09, MMRS-19); and the reception record, the table composed in the entity's place and measured against the page as read (CAP-1-01–08, MMRS-21). The record's restatements of the paper (MMRS-02, MMRS-04, MMRS-05, MMRS-06, MMRS-17) and of the released interactions (MMRS-08, MMRS-15) are one lineage with B4, which keeps the earliest date. Held with their sources (P_coh), not reconciled: the count and date as read, 25 February 2026, 22 models, 247,944 votes (MMRS-01, MMRS-12, CAP-1-02), against the board's own latest published state, 13 October 2026, 42 models (F36); the registry's finding that the composed table does not match the page as read (CAP-1-03) beside the board's latest data, which the table matches (F36, F37). Every field claim is composed as received (ontology.translation). The measure: 0 of 10 KO_B sentences survive in the KO (§4.10); no opening run, no block (measure).", "alignment": { "T1_cards_claims": [ "F1", "F2", "F3", "F4" ], "T1_composed": [ "F1", "F2", "F3" ], "T1_own_page_claims_composed": [ "F36", "F37", "F14" ], "T1_not_composed": [ "F4", "F5", "F6", "F7", "F8", "F9", "F10", "F11", "F12", "F13", "F15", "F16", "F17", "F18", "F19", "F20", "F21", "F22", "F23", "F24", "F25", "F26", "F27", "F28", "F29", "F30", "F31", "F32", "F33", "F34", "F35", "F38", "F39", "F40", "F41", "F42", "F43", "F44", "F45", "F46", "F47", "F48" ], "table_check": { "against": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-latest.json (search, split latest, leaderboard_publish_date 2026-10-13)", "rows": [ { "rank": 1, "T": "GPT-5.6 Sol (xHigh) | OpenAI | 1261 ± 6", "B5": "gpt-5.6-sol-xhigh | openai | 1261.13 (1254.84–1267.42)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 2, "T": "Qwen3.8-Max | Alibaba | 1261 ± 13", "B5": "qwen3.8-max | alibaba | 1261.08 (1247.89–1274.26)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 3, "T": "Claude Opus 4.6 (Search) | Anthropic | 1251 ± 5", "B5": "claude-opus-4-6-search | anthropic | 1251.37 (1246.83–1255.91)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 4, "T": "GPT-5.5 (Search) | OpenAI | 1247 ± 5", "B5": "gpt-5.5-search | openai | 1246.55 (1241.65–1251.45)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 5, "T": "Claude Fable 5.1 (High) | Anthropic | 1246 ± 7", "B5": "claude-fable-5.1-high | anthropic | 1245.85 (1238.79–1252.90)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 6, "T": "Claude Opus 5 (High) | Anthropic | 1242 ± 7", "B5": "claude-opus-5-high | anthropic | 1241.57 (1234.70–1248.44)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 7, "T": "GPT-6 Astra (Max) | OpenAI | 1234 ± 8", "B5": "gpt-6-astra-max | openai | 1233.62 (1225.77–1241.46)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 8, "T": "GPT-6.1 Sol (Max) | OpenAI | 1233 ± 9", "B5": "gpt-6.1-sol-max | openai | 1233.10 (1224.24–1241.96)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 9, "T": "Claude Opus 4.7 | Anthropic | 1230 ± 5", "B5": "claude-opus-4-7 | anthropic | 1230.15 (1225.28–1235.01)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 10, "T": "Claude Fable 5 (High) | Anthropic | 1229 ± 7", "B5": "claude-fable-5-high | anthropic | 1229.14 (1222.63–1235.65)", "model": true, "provider": true, "rating": true, "ci": true }, { "rank": 11, "T": "ERNIE 5.1 | Baidu | 1227 ± 9", "B5": "ernie-5.1 | baidu | 1226.72 (1217.24–1236.19)", "model": true, "provider": true, "rating": true, "ci": true } ], "all_match": true, "model_reading": "display names against slugs, read: GPT-5.6 Sol (xHigh) / gpt-5.6-sol-xhigh; Qwen3.8-Max / qwen3.8-max; Claude Opus 4.6 (Search) / claude-opus-4-6-search; GPT-5.5 (Search) / gpt-5.5-search; Claude Fable 5.1 (High) / claude-fable-5.1-high; Claude Opus 5 (High) / claude-opus-5-high; GPT-6 Astra (Max) / gpt-6-astra-max; GPT-6.1 Sol (Max) / gpt-6.1-sol-max; Claude Opus 4.7 / claude-opus-4-7; Claude Fable 5 (High) / claude-fable-5-high; ERNIE 5.1 / ernie-5.1: each the same model", "rule": "provider = organization (case-insensitive); rating = round(rating); ± = round(max(rating_upper − rating, rating − rating_lower))" }, "rule": "a field claim counts as composed when T states it at its source's grain or coarser; the typology of §6.2 is applied in T_vs_LB" } }, "kernel": [ { "K": "K1", "claim": "`CAP-1-01`", "source": "Capture Registry, 2026-10-29", "M_src": "documented", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K2", "claim": "`CAP-1-02`", "source": "Capture Registry, 2026-10-29", "M_src": "documented", "sense": "the current board: snapshots and dates", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F36, F38 (the 26 February snapshot agrees with the page as read)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K3", "claim": "`CAP-1-03`", "source": "Capture Registry, 2026-10-29", "M_src": "documented", "sense": "the current board: snapshots and dates", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F36, F37 (the composed table's rows are the latest board's)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K4", "claim": "`CAP-1-04`", "source": "Capture Registry, 2026-10-29", "M_src": "self-description", "sense": "the current board: snapshots and dates", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F33, F36 (Arena's latest split)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K5", "claim": "`CAP-1-05`", "source": "Capture Registry, 2026-10-29", "M_src": "self-description", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K6", "claim": "`CAP-1-06`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K7", "claim": "`CAP-1-07`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K8", "claim": "`CAP-1-08`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K9", "claim": "`MMRS-01`", "source": "MMRS record", "M_src": "stipulation", "sense": "what it is: a leaderboard of search models by human vote", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F36 (the latest published board, 13 October 2026, 42 models)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K10", "claim": "`MMRS-02`", "source": "MMRS record", "M_src": "attributed", "sense": "the citation finding: preference follows citation count", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F28 (same lineage: the record's restatement)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K11", "claim": "`MMRS-03`", "source": "MMRS record", "M_src": "stipulation", "sense": "the vote and the score", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F43, F12 (the field's account of the vote)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K12", "claim": "`MMRS-04`", "source": "MMRS record", "M_src": "attributed", "sense": "the dataset: interactions and votes", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F26, F27 (same lineage)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K13", "claim": "`MMRS-05`", "source": "MMRS record", "M_src": "attributed", "sense": "the citation finding: preference follows citation count", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F28 (same lineage)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K14", "claim": "`MMRS-06`", "source": "MMRS record", "M_src": "attributed", "sense": "sources cited and preferred", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F29 (same lineage)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K15", "claim": "`MMRS-07`", "source": "MMRS record", "M_src": "interpretation", "sense": "what the vote rewards", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F28, F18", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K16", "claim": "`MMRS-08`", "source": "MMRS record", "M_src": "interpretation", "sense": "the dataset: interactions and votes", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F31 (same lineage)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K17", "claim": "`MMRS-09`", "source": "MMRS record", "M_src": "documented", "sense": "operator: Arena, formerly LMArena and Chatbot Arena", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F23, F43 (the field names Arena Intelligence Inc. and Arena)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K18", "claim": "`MMRS-10`", "source": "MMRS record", "M_src": "stipulation", "sense": "what it is: a leaderboard of search models by human vote", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K19", "claim": "`MMRS-11`", "source": "MMRS record", "M_src": "stipulation", "sense": "the vote and the score", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K20", "claim": "`MMRS-12`", "source": "MMRS record", "M_src": "documented", "sense": "the current board: snapshots and dates", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F36", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K21", "claim": "`MMRS-13`", "source": "MMRS record", "M_src": "documented", "sense": "the lifecycle kept apart from the observation", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K22", "claim": "`MMRS-14`", "source": "MMRS record", "M_src": "interpretation", "sense": "the record's frame: an aggregator of reception work", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K23", "claim": "`MMRS-15`", "source": "MMRS record", "M_src": "documented", "sense": "the dataset: interactions and votes", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F31 (same lineage)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K24", "claim": "`MMRS-16`", "source": "MMRS record", "M_src": "interpretation", "sense": "what the vote rewards", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K25", "claim": "`MMRS-17`", "source": "MMRS record", "M_src": "attributed", "sense": "the citation finding: preference follows citation count", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F26, F27, F28 (same lineage)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K26", "claim": "`MMRS-18`", "source": "MMRS record", "M_src": "self-description", "sense": "what the vote rewards", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K27", "claim": "`MMRS-19`", "source": "MMRS record", "M_src": "self-description", "sense": "operator: Arena, formerly LMArena and Chatbot Arena", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K28", "claim": "`MMRS-20`", "source": "MMRS record", "M_src": "documented", "sense": "the current board: snapshots and dates", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K29", "claim": "`MMRS-21`", "source": "MMRS record", "M_src": "documented", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K30", "claim": "`MMRS-22`", "source": "MMRS record", "M_src": "stipulation", "sense": "the lifecycle kept apart from the observation", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K31", "claim": "`MMRS-23`", "source": "MMRS record", "M_src": "stipulation", "sense": "the record's frame: an aggregator of reception work", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K32", "claim": "`MMRS-24`", "source": "MMRS record", "M_src": "stipulation", "sense": "the record's frame: an aggregator of reception work", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K33", "claim": "`D741-01`", "source": "#741", "M_src": "attributed", "sense": "the vote: relative preference", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K34", "claim": "`D741-02`", "source": "#741", "M_src": "interpretation", "sense": "the vote: relative preference", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F14 (the score as a model of pairwise battles)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K35", "claim": "`O741-01`", "source": "#741", "M_src": "interpretation", "sense": "the rated systems: citation as proxy", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K36", "claim": "`O741-02`", "source": "#741", "M_src": "interpretation", "sense": "the rated systems: citation as proxy", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "F28 (citation as a signal of credibility)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K37", "claim": "`O1705-01`", "source": "#1705", "M_src": "stipulation", "sense": "rank as attribute", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K38", "claim": "`O1705-02`", "source": "#1705", "M_src": "interpretation", "sense": "rank as attribute", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" }, { "K": "K39", "claim": "`O1705-03`", "source": "#1705", "M_src": "interpretation", "sense": "rank as attribute", "qualifiers carried": "—", "f (source's own falsifiers)": "none stated (watch: a later capture at 'search arena leaderboard' on Google AI Overview that states or drops the claim, attributed or not; uptake, §8.3)", "contrast": "none (the field has no claim on the point)", "resolver": "a later Capture Registry observation at 'search arena leaderboard' (ADDR-ad4b00d7a1f0)" } ], "kernel_T": [ { "claim": "F5", "note": "the definition, a crowdsourced evaluation platform based on human preference (B3); not composed" }, { "claim": "F43", "note": "the operator and the vote (Wikipedia); not composed" }, { "claim": "F14", "note": "the score: composed only as the column label 'Elo Rating'" }, { "claim": "F15", "note": "the stated limit of the intervals (B3); the ± is composed, the limit not" }, { "claim": "F26", "note": "the dataset (B4); not composed" }, { "claim": "F28", "note": "the citation finding (B4); not composed (CAP-1-01)" }, { "claim": "F36", "note": "the board's publish date, 13 October 2026, and its 42 models: the rows composed, the date and the count not; 'live' added" }, { "claim": "F39", "note": "the style-controlled order: not composed, the default order given without saying so" }, { "claim": "F44", "note": "the methodology limits on record (Wikipedia); not composed" }, { "claim": "F42", "note": "no article at 'Search Arena' (Wikipedia); not composed" } ], "ledger": { "archive": "datasets/negative-of-the-negative/v2/ledgers/search-arena/ledger-archive.json", "field": "datasets/negative-of-the-negative/v2/ledgers/search-arena/field.json", "selection": "datasets/negative-of-the-negative/v2/traversal/search-arena/", "config": "datasets/negative-of-the-negative/v2/panel/configs/search-arena.json", "intake": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/", "audit": null }, "ontology": { "composition": "equal", "composition_note": "equal-terms case (ρ = extends): one claim space; the archive's definitions composed as definitions beside the field's; nothing translated", "case": { "value": "equal", "statement": "The archive's account of Search Arena extends the field's ontology on the field's own object. The MMRS record holds the entity the field holds, the leaderboard of search-enabled models rated from head-to-head human votes with its dataset paper, every field read from the entity's own pages with source and basis (census.json, sources and evidence), and restates the paper's findings as the paper gives them (one lineage with B4). It adds senses and limits the field does not have: what the vote rewards, a preference for citation count over citation support (MMRS-07), and the unchecked question whether the live board judges citations (MMRS-18); the count kept with its date and the lifecycle apart from the observation (MMRS-12, MMRS-22); and the reception record, the table composed in the entity's place (CAP-1-01, CAP-1-08). Two deposits reached by D and O add a limit and a category on the field's object, never a rival ontology: pairwise judgment measures relative preference, not absolute quality (#741), and a ranking's result is the ranker's act stated as the winner's attribute (#1705). Nothing in the archive denies a field presupposition, and the archive supplies no ontology the field lacks: the law of the suppressed witness (§0.10) does not apply.", "loci": [ { "dep": "MMRS", "locus": "articles.py ARTICLES['search-arena'], kernel s1", "quote": "Search Arena ranks search-enabled language models by head-to-head human votes: 247,944 votes across 22 models at its update of 25 February 2026." }, { "dep": "MMRS", "locus": "articles.py ARTICLES['search-arena'], 'Place in the field' s1", "quote": "Its leaderboard rewards what users prefer, and its paper shows that preference rewarding citation count over citation support." }, { "dep": "MMRS", "locus": "census.json rows[slug == 'search-arena'], unchecked[0]", "quote": "whether the live leaderboard judges citations" }, { "dep": "capture-registry", "locus": "search-arena-leaderboard-aio-20261029, analysis s2", "quote": "The record keeps the entity; the capture keeps what was composed in its place." }, { "dep": 741, "locus": "¶38, s2", "quote": "This is a finding about *relative* preference under specified rubrics — comparing A versus B — not about absolute quality assessment." }, { "dep": 1705, "locus": "¶44, s4", "quote": "A ranking's *best* is attributization in its commercial form: the ranker's act spoken in the product's voice." } ] }, "E_F": { "type": "C (public entity: a leaderboard of search-augmented language models rated by human preference, with its dataset paper and published board history)", "definition": "Search Arena, a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference (B3), launched 18 March 2025 by Arena (formerly LMArena and Chatbot Arena, Arena Intelligence Inc.; B8, B3), its page the 'AI Search Leaderboard: Top Models for Research' (B1); users vote between two anonymous answers, scores from a Bradley-Terry model on the Elo scale (B3, B8), the 'Arena Score' (B5); its paper (Miroyan et al., arXiv:2506.05334, ICLR 2026) a crowd-sourced human-preference dataset of over 24,000 paired interactions with about 12,000 votes, finding that preference follows the number of citations even when the cited content does not support the claims (B4); its published boards from 7 August 2025, the latest of 13 October 2026 ranking 42 models (B5, B6). No Wikipedia article (B7); the operator's article does not name it (B8).", "claims": [ "F5", "F11", "F43", "F23", "F1", "F14", "F35", "F24", "F26", "F27", "F28", "F32", "F36", "F40", "F42", "F48" ] }, "E_A": { "definition": "Search Arena as the MMRS meta-aggregator records it (editor Lee Sharks, 2026-10-28, corrected 2026-10-29), a row of its census: the leaderboard on arena.ai that ranks search-enabled language models by head-to-head human votes, its unit the search-enabled model rated from those votes, its kind a leaderboard and its role evaluation, its operator arena.ai on the operator's statement; 247,944 votes across 22 models at the page's update of 25 February 2026, as read on 2026-10-28, active; its dataset paper's interactions preserved as original outputs; a board that rewards what users prefer, where that preference rewards citation count over citation support, whether the live board judges citations unchecked; and, at the Capture Registry's address 'search arena leaderboard' (2026-10-29), an entity whose composition is a table in its place. Pairwise judgment of the kind it aggregates measures relative preference, not absolute quality (#741); the systems it rates admit documents on structural signals before reading (#741); a ranking states its comparison as the winner's attribute (#1705).", "claims": [ "MMRS-01", "MMRS-11", "MMRS-10", "MMRS-09", "MMRS-12", "MMRS-13", "MMRS-08", "MMRS-07", "MMRS-16", "MMRS-18", "MMRS-24", "CAP-1-01", "CAP-1-08", "D741-02", "O741-02", "O1705-01", "O1705-03" ], "defining_papers": [ 741, 1705 ], "defining_records": [ "MMRS record (census.json rows[slug == 'search-arena']; articles.py ARTICLES['search-arena']; https://www.machinemediation.org/aggregators/search-arena/)", "Capture Registry entry search-arena-leaderboard-aio-20261029 (OBS-cd6ab4746f08, AXN:0AB6.CAPTURE.⛳🧱🟢⏪🎻🏛️; https://www.alexanarch.org/captures/search-arena-leaderboard-aio-20261029/)" ], "note": "No deposit defines the archive's entity: the string pass names it in no deposit text by 'Search Arena' (0 of 1,711), and the deposits admitted (#741 by 'Chatbot Arena' and O-class, #1705 by O-class) bear on its method and class. The archive's defining material for the entity is its two records, which are not deposits." }, "rho": { "value": "extends", "status": "read 2026-10-29 from the MMRS record, the Capture Registry's entry and the two admitted deposits (§1.2a); no deposit defines the entity", "statement": "The archive reads E_F as the entity and adds senses, limits and relations on it: what the vote rewards (MMRS-07, MMRS-16) and the open question of citation judgment (MMRS-18); the count with its date and the lifecycle kept apart from the observation (MMRS-12, MMRS-22); what pairwise judgment measures (D741-02); the systems it rates under a proxy regime (O741-02); rank as attribute (O1705-01, O1705-03); and the reception record (CAP-1-01, CAP-1-03, CAP-1-08).", "claims": [ "MMRS-07", "MMRS-16", "MMRS-18", "MMRS-12", "MMRS-22", "D741-02", "O741-02", "O1705-01", "O1705-03", "CAP-1-01", "CAP-1-03", "CAP-1-08" ], "guard": "ρ = extends claims an addition and no more (§1.2a). Where an archive claim and a field claim differ in date (the count as read, 25 February 2026, 22 models, MMRS-01, MMRS-12, CAP-1-02, against the latest published board, 13 October 2026, 42 models, F36), both are composed with their sources and dates (P_coh). The deposits' claims are said of pairwise judgment (#741, of the MT-Bench / Chatbot Arena studies), of AI search systems as a class (#741) and of rankings as a class (#1705); they enter by D and O-class on classes sourced from the field (B1, B2, B3, B4, B8), and the row does not compose them as claims the archive makes about Search Arena by name.", "considered": [ { "value": "revises", "against": "the archive does not read the field's entity as its own earlier statements; the record's correction of 2026-10-29 added a relation, not a revision of the entity" }, { "value": "constructs", "against": "the archive does not read the entity as made by reception; its fields are read from the entity's own pages with source and basis, and the capture records a composition, not the entity's making" } ] }, "translation": [ { "lineage": "what it is: a leaderboard of search models by human vote", "field_claims": [ "F1", "F5", "F10" ], "received": "F1 (B1, card title): \"AI Search Leaderboard: Top Models for Research\"; F5 (B3, TL;DR ¶1 (and the post's opening)): \"We introduce Search Arena, a crowdsourced in-the-wild evaluation platform for search-augmented LLM systems based on human preference.\"; F10 (B3, §1 Why Search Arena?, ¶3): \"To this end, we developed search arena, aiming to (1) enable crowd-sourced evaluation of search-augmented LLMs and (2) release a diverse, in-the-wild dataset of user–system interactions.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-01, MMRS-10, MMRS-11, MMRS-24", "archive_claims": [ "MMRS-01", "MMRS-10", "MMRS-11", "MMRS-24" ] }, { "lineage": "the current board: snapshots and dates", "field_claims": [ "F2", "F7", "F32", "F33", "F36", "F37", "F38", "F39", "F40", "F41" ], "received": "F2 (B1, card snippet (truncated)): \"View overall rankings across LLMs with inte\"; F7 (B3, TL;DR ¶2): \"Based on 7k human votes (03/18–04/13), Gemini-2.5-Pro-Grounding and Perplexity-Sonar-Reasoning-Pro are at the top, followed by the rest of Perplexity’s Sonar models, Gemini-2.0-Flash-Grounding, and OpenAI’s web search API models.\"; F32 (B5, README, title and opening): \"Historical snapshots of the [Arena](https://arena.ai/leaderboard) leaderboard.\"; F33 (B5, README, Splits): \"- `full`: All historically published leaderboards - `latest`: Only the most recently published leaderboards\"; F36 (B5, search, split latest: rows rank 1 and 11; num_rows_total): \"{\"model_name\":\"gpt-5.6-sol-xhigh\",\"organization\":\"openai\",\"license\":\"Proprietary\",\"rating\":1261.1307593350664,\"rating_lower\":1254.8406048513853,\"rating_upper\":1267.4209138187477,\"variance\":10.299744257424065,\"vote_count\":60437,\"rank\":1,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-10-13\"} … {\"model_name\":\"ernie-5.1\",\"organization\":\"baidu\",\"license\":\"Proprietary\",\"rating\":1226.7150529766889 … \"rank\":11,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-10-13\"} … \"num_rows_total\":42\"; F37 (B5, search, split latest: rows rank 2 and 3): \"{\"model_name\":\"qwen3.8-max\",\"organization\":\"alibaba\",\"license\":\"Proprietary\",\"rating\":1261.0761861904762 … {\"model_name\":\"claude-opus-4-6-search\",\"organization\":\"anthropic\",\"license\":\"Proprietary\",\"rating\":1251.3682634708416\"; F38 (B5, search, split full: snapshot of 2026-02-26, rows rank 1 and 2): \"{\"model_name\":\"claude-opus-4-6-search\",\"organization\":\"anthropic\",\"license\":\"Proprietary\",\"rating\":1255.4140677086834,\"rating_lower\":1245.0272537380415,\"rating_upper\":1265.8008816793258,\"variance\":28.084618239181463,\"vote_count\":3607,\"rank\":1,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-02-26\"} … {\"model_name\":\"grok-4.20-beta1\",\"organization\":\"xai\",\"license\":\"Proprietary\",\"rating\":1225.0104803795257\"; F39 (B5, search_style_control, split latest: row rank 2): \"{\"model_name\":\"gpt-6.1-sol-max\",\"organization\":\"openai\",\"license\":\"Proprietary\",\"rating\":1249.9181501199778,\"rating_lower\":1240.895777035074,\"rating_upper\":1258.9405232048819,\"variance\":21.190703814049858,\"vote_count\":5555,\"rank\":2,\"category\":\"overall\",\"leaderboard_publish_date\":\"2026-10-13\"}\"; F40 (B5, search, split full: statistics, leaderboard_publish_date): \"\"min\":\"2025-08-07 00:00:00\",\"max\":\"2026-10-13 00:00:00\"\"; F41 (B6, ¶1): \"For almost three years, Arena has been publishing leaderboards covering frontier AI capabilities across 10 arenas, dozens of categories, and hundreds of models, and today we're releasing the entire history of those leaderboards as a public-access dataset.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-12, MMRS-20, MMRS-13, MMRS-22, CAP-1-02, CAP-1-03, CAP-1-04", "archive_claims": [ "MMRS-12", "MMRS-20", "MMRS-13", "MMRS-22", "CAP-1-02", "CAP-1-03", "CAP-1-04" ] }, { "lineage": "operator: Arena, formerly LMArena and Chatbot Arena", "field_claims": [ "F3", "F4", "F23", "F43", "F46", "F47" ], "received": "F3 (B2, card title): \"Arena Leaderboard: Official AI Model Rankings & Benchmarks\"; F4 (B2, card snippet): \"GPT-6.1 Terra is #8 in Agent · Max Kimi K3.1 Pro is #9 in WebDev · Max MAI-Image-2.7-Preview is #3 in Image Edit\"; F23 (B3, footer): \"© 2026 Arena Intelligence Inc.\"; F43 (B8, lead ¶1): \"Arena (formerly LMArena and Chatbot Arena) is a public, web-based platform that evaluates large language models (LLMs). Users enter prompts for two anonymous models to respond to and vote on the model that gave the better response, after which the models' identities are revealed.\"; F46 (B8, History ¶1, ¶5, ¶8): \"Chatbot Arena was released on April 24, 2023. … In April 2025, LMArena incorporated as an independent company. … On January 28, 2026, LMArena rebranded to \"Arena\".\"; F47 (B8, History ¶9): \"On October 8, 2026, Arena announced it raised $200 million in a Series B funding round co-led by Lightspeed Venture Partners and Khosla Ventures, thus placing the company at a $3.1 billion valuation.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-09, MMRS-19", "archive_claims": [ "MMRS-09", "MMRS-19" ] }, { "lineage": "the rated systems: search-augmented LLMs", "field_claims": [ "F6", "F13", "F20", "F25", "F30" ], "received": "F6 (B3, TL;DR ¶1): \"Unlike LM-Arena or SimpleQA, our data focuses on current events and diverse real-world use cases\"; F13 (B3, §1.B Models, ¶1): \"Search Arena currently supports 11 models from three providers: Perplexity, Gemini, and OpenAI.\"; F20 (B3, §1.B Models, Table 2 note): \"†We evaluate OpenAI’s web search API, which is different from the search feature in the ChatGPT product.\"; F25 (B4, Abstract s1–s2): \"Search-augmented language models combine web search with Large Language Models (LLMs) to improve response groundedness and freshness. However, analyzing these systems remains challenging: existing datasets are limited in scale and narrow in scope, often constrained to static, single-turn, fact-checking questions.\"; F30 (B4, Abstract s7–s8): \"We find that web search does not degrade and may even improve performance in non-search settings; however, the quality in search settings is significantly affected if solely relying on the model's parametric knowledge.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside O741-01, O741-02", "archive_claims": [ "O741-01", "O741-02" ] }, { "lineage": "the citation finding: preference follows citation count", "field_claims": [ "F8", "F9", "F16", "F17", "F18", "F28" ], "received": "F8 (B3, TL;DR ¶2): \"Standardizing citation styles had minimal effect on rankings (see Sec. 2).\"; F9 (B3, TL;DR ¶3): \"Three features show strong positive correlation with human preference: response length, citation count, and citing specific web sources like YouTube and online forum/blogs (see Sec. 3).\"; F16 (B3, §3 Control Experiments, 'Number of citations'): \"Controlling for citation count also results in a positive and significant coefficient (0.234, p < 0.05), suggesting a preference for responses with more cited sources.\"; F17 (B3, §3 Control Experiments, 'Joint controls'): \"When controlling for all features, only response length and citation count remain statistically significant.\"; F18 (B3, §3 Control Experiments, Tab. 5 note 2): \"When controlling for the number of citations, model rankings converge (i.e., multiple models share the same rank), suggesting that the number of citations is a significant factor impacting differences across models and the resulting rankings.\"; F28 (B4, Abstract s5): \"Our analysis reveals that user preferences are influenced by the number of citations, even when the cited content does not directly support the attributed claims, uncovering a gap between perceived and actual credibility.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-02, MMRS-05, MMRS-17, MMRS-07", "archive_claims": [ "MMRS-02", "MMRS-05", "MMRS-17", "MMRS-07" ] }, { "lineage": "genealogy: the launch, the paper, the dataset", "field_claims": [ "F11", "F22", "F24", "F34" ], "received": "F11 (B3, §1, ¶4): \"Since our initial launch on March 18th, we’ve collected over 11k votes across 10+ models. We then filtered this data to construct 7k battles with user votes\"; F22 (B3, header; Citation): \"Published on 15 Apr 2025 … Last updated on 2 Aug 2025 … Mihran Miroyan*. (2025). Introducing the Search Arena: Evaluating Search-Enabled AI. https://blog.lmarena.ai/blog/2025/search-arena/\"; F24 (B4, abstract page header): \"[Submitted on 5 Jun 2025 (v1), last revised 2 Mar 2026 (this version, v2)] … Authors:Mihran Miroyan, Tsung-Han Wu, Logan King, Tianle Li, Jiayi Pan, Xinyan Hu, Wei-Lin Chiang, Anastasios N. Angelopoulos, Trevor Darrell, Narges Norouzi, Joseph E. Gonzalez\"; F34 (B5, README, Notes): \"- Search and Webdev Arena data starts from their releases on the arena.ai (then lmarena.ai) on August 7 and November 12 2025 respectively.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-04", "archive_claims": [ "MMRS-04" ] }, { "lineage": "the vote and the score", "field_claims": [ "F12", "F14", "F15", "F21", "F35" ], "received": "F12 (B3, §1.A Data, ¶1): \"To balance these considerations, we introduced “style randomization”: responses are displayed either in a standardized format or in the original format (i.e., the citation style agreed upon with each model provider).\"; F14 (B3, §2 Leaderboard, ¶3): \"Consistent with LM Arena, we apply the Bradley-Terry (BT) model to compute model scores. The resulting BT coefficients are then translated to Elo scale\"; F15 (B3, §2 Leaderboard, ¶3): \"The confidence intervals are still wide, which means the leaderboard hasn’t fully settled and there’s still some uncertainty.\"; F21 (B3, §4 Conclusion & What's Next): \"Open participation: We are inviting model submissions from researchers and industry, and encouraging public voting.\"; F35 (B5, README, Schema): \"| rating | float | Arena Score | … | vote_count | int | Number of battles for this model |\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-03, MMRS-11, D741-01, D741-02", "archive_claims": [ "MMRS-03", "MMRS-11", "D741-01", "D741-02" ] }, { "lineage": "sources cited and preferred", "field_claims": [ "F19", "F29" ], "received": "F19 (B3, §3 Model Characteristics, findings 1–2): \"All models favor authoritative sources (e.g., Wikipedia, .edu, .gov domains). … OpenAI models heavily cite news sources—51.3% overall and 87.3% for Trump-related prompts.\"; F29 (B4, Abstract s6): \"Furthermore, user preferences vary across cited sources, revealing that community-driven platforms are generally preferred and static encyclopedic sources are not always appropriate and reliable.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-06", "archive_claims": [ "MMRS-06" ] }, { "lineage": "the dataset: interactions and votes", "field_claims": [ "F26", "F27", "F31" ], "received": "F26 (B4, Abstract s3): \"In this work, we introduce Search Arena, a crowd-sourced, large-scale, human-preference dataset of over 24,000 paired multi-turn user interactions with search-augmented LLMs.\"; F27 (B4, Abstract s4): \"The dataset spans diverse intents and languages, and contains full system traces with around 12,000 human preference votes.\"; F31 (B4, Abstract s9; Comments): \"We open-sourced the dataset to support future research. … Accepted to ICLR 2026.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-04, MMRS-08, MMRS-15, MMRS-17", "archive_claims": [ "MMRS-04", "MMRS-08", "MMRS-15", "MMRS-17" ] }, { "lineage": "reception: the encyclopedia", "field_claims": [ "F42", "F48" ], "received": "F42 (B7, rendered page (HTTP 404)): \"Wikipedia does not have an article with this exact name.\"; F48 (B8, redirect pages LMArena and Chatbot Arena (raw wikitext)): \"#REDIRECT [[Arena.ai]]\"", "receiving_text": "—", "translation": "untranslated: composed as received beside CAP-1-01, CAP-1-08", "archive_claims": [ "CAP-1-01", "CAP-1-08" ] }, { "lineage": "Arena's method and its limits", "field_claims": [ "F44", "F45" ], "received": "F44 (B8, lead ¶4): \"Research has identified specific limitations in Arena's methodology.\"; F45 (B8, History ¶4): \"Llama 4 Maverick beat GPT-4o and Gemini 2.0 Flash on LMArena, but the version of Maverick on LMArena unfairly differed from the publicly available version. LMArena updated their policies in response.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside D741-02, MMRS-19", "archive_claims": [ "D741-02", "MMRS-19" ] } ], "receiving_texts": "equal-terms case: every field claim composed as received, beside the archive claims it stands with" }, "measure": { "rule": "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": { "surviving": 0, "of": 10, "positions": [ null, null, null, null, null, null, null, null, null, null ] }, "opening_run": { "rule": "v0.9 §5.10 (equal-terms case): the field-only KO's sentences as the KO's opening run in their own order", "breach": false }, "blocks": { "rule": "v0.9 §4.1 (equal-terms and mixed cases): the KO does not open with the field as a block or close with the archive as one; sentences typed by the ledgers of the claims they cite (F field-only, A archive-only, M both)", "sequence": "MFFFFMMFAFMAMFMFAAFFAMAFAAAM", "field_only_head": 0, "archive_only_tail": 0, "breach": false }, "max_similarity": [ 0.323, 0.772, 0.466, 0.493, 0.14, 0.345, 0.368, 0.149, 0.439, 0.483 ], "ko_sentence_1_max_similarity_to_KO_B": 0.323 }, "homonyms": { "names": [], "basis": "No homonym on the rail: both cards are Arena's. The census record carries no disambiguation, negative_tags or battery for this row (checked: census.json rows[slug == 'search-arena'] has none of the three keys). In the field the name 'Search Arena' names two aspects of one entity, the platform and leaderboard (B3, B1) and the dataset of its battles (B4: 'we introduce Search Arena, a crowd-sourced … human-preference dataset'); the MMRS record holds them as one entity, the leaderboard with its dataset paper, and the row does the same. Distinct entities kept distinct: Arena's other arenas and their boards (Text, the former Chatbot Arena; WebDev; Agent; Image Edit; B2's snippet), each a board of the same operator; Arena itself, the operator (B8); the dataset search-arena-24k as a released artefact (B4's link); other 'arenas' found by WebSearch on 2026-10-29 and on no card, Design Arena (designarena.ai) and SWE-Arena (huggingface.co/datasets/SWE-Arena).", "distinct_entities": [ { "name": "Arena (arena.ai; formerly LMArena, Chatbot Arena)", "relation": "operator" }, { "name": "Arena's other arenas (Text, WebDev, Agent, Image Edit, Vision, Document, video)", "relation": "sibling boards of the operator" }, { "name": "search-arena-24k (huggingface.co/datasets/lmarena-ai/search-arena-24k)", "relation": "the entity's released dataset" }, { "name": "Arena Leaderboard Dataset (huggingface.co/datasets/lmarena-ai/leaderboard-dataset)", "relation": "the operator's published history of all its boards" }, { "name": "Design Arena; SWE-Arena", "relation": "other arenas by name, on no card" } ], "substitutes": { "T1": [ "the board's current table (eleven rows) composed in the entity's place (CAP-1-01)" ] }, "register": { "address_1": { "slug": "search-arena-leaderboard-aio-20261029", "obs_id": "OBS-cd6ab4746f08" } } }, "notes": [ "One address, one epoch: search-arena-leaderboard-aio-20261029 (OBS-cd6ab4746f08, ADDR-ad4b00d7a1f0, AXN:0AB6.CAPTURE.⛳🧱🟢⏪🎻🏛️), 'search arena leaderboard', two cards, Google AI Overview (expanded), signed out, incognito, 2026-10-29, seated in the Capture Registry (checked: data/EA-WG-CAPTURES-01.json, entries[500]).", "Wikipedia: no article at 'Search Arena' (raw wikitext and rendered page HTTP 404, 05:08:53Z); the operator's article Arena.ai (LMArena and Chatbot Arena redirect to it) does not mention Search Arena, confirmed from its raw wikitext (field_by_origin.wikipedia_state).", "Fetching: arena.ai/leaderboard/search (card 1, and the MMRS record's source for its counts) and arena.ai/leaderboard (card 2) answered curl with HTTP 403 and a Cloudflare challenge page (saved, B1.html, B2.html); the session did not try to get past the challenge. A summarising fetch (WebFetch, about 05:06Z) did read both pages; what it reported is recorded in field.json (B1.webfetch, B2.webfetch) and is not ledgered (§3.1(a)). In place of the page, the leaderboard's own published data was fetched verbatim from Arena's Hugging Face dataset (B5: README, the search and search_style_control latest splits, the search full split, its statistics), with Arena's post announcing it (B6). The launch post (B3) and the arXiv abstract (B4) were fetched verbatim with curl (HTTP 200).", "What the leaderboard page states, and when: as the summarising fetch reported it on 2026-10-29, a date of 25 February 2026, 247,944 votes, 22 models, Anthropic first at 1255±10 (3,607 votes), xAI second at 1225±8 (4,687), OpenAI third at 1219±6 (20,150), Google 1218±6 (25,311) and 1214±5 (31,966), Diffbot 1024±8 at 21 and OpenAI 1006±11 at 22; these are, figure for figure, Arena's published snapshot of 2026-02-26 (B5 full split; F38 ledgers ranks 1–2). Arena's own latest published search board is dated 2026-10-13, 42 models, gpt-5.6-sol-xhigh first at 1261.1 (F36). The Overview's table is that latest board, ranks 1–11, model, provider, rating and interval (delta.alignment.table_check). Which render a browser receives at the address is not established here.", "Archive subset: the string pass over 1,711 texts found 3 D candidates (all on 'Chatbot Arena', in #716 and #741), no R, 14 O-class sentences (14 deposits; a second pass narrowed the class to model rankings after a bare 'ranking' returned 322 sentences), and a hop of 12 candidates from the admitted deposits. Admitted: #741 (D: what the Chatbot Arena studies' pairwise judgment measures; O: the proxy regime of AI search composition) and #1705 (O: ranking as attributization); 13 candidate deposits and all hop candidates read and not admitted (traversal/search-arena/reading.json). No deposit names Search Arena. The archive's claims are its two records of the entity (CAP-1-01–08 from the Capture Registry's findings, reading and analysis; MMRS-01–24 from the MMRS record) and 7 deposit claims.", "Defects found, reported and not repaired. (1) The MMRS record's kernel and count, '247,944 votes across 22 models at its update of 25 February 2026' (MMRS-01, MMRS-12, read 2026-10-28), are the page as served to a non-browser fetch, which agrees with Arena's snapshot of 2026-02-26; Arena's own latest published search board (B5) is dated 2026-10-13 with 42 models, eight months later. The record's 'Observation 2026-10-28: read' stands; its count is not the board's current state by Arena's own data. (2) The Capture Registry's finding 'The composed table's ratings, providers and order do not match it' (CAP-1-03) and the entry's heading ('THE TABLE UNLIKE THE PAGE AS THIS RECORD READ IT') hold against the page as read; the Overview's eleven rows are Arena's latest published board (2026-10-13) rank for rank, so the finding's own caveat, 'Google may hold a later render', is borne out in substance: the composition carries the later published state, and the analysis's 'its figures cannot be reconciled with the page as read' (CAP-1-07) is true of the page as read and not of the board. (3) The page's total of 247,944 votes is not reproducible from Arena's data (vote_count is 'Number of battles for this model'; the 2026-02-26 snapshot sums to 479,042, half 239,521). (4) The census row has no disambiguation, negative_tags or battery, which the ai-incident-database and papers-with-code rows carry. (5) The record's sources omit the launch post (B3) and the leaderboard dataset (B5); the field holds two release dates, 18 March 2025 (B3) and 7 August 2025 on lmarena.ai (B5's README), not a defect of the record. (6) The Overview's display names ('GPT-5.6 Sol (xHigh)', 'Claude Fable 5.1 (High)') are in no saved text (the data gives slugs); they are presumably the page's labels and are unverified here.", "The case and ρ (decided): equal, ρ = extends; read from the archive's records and the two admitted deposits (ontology.case, ontology.rho, with 'revises' and 'constructs' considered).", "Homonyms: none on the rail; the operator, its other arenas, the released dataset and the operator's board history are distinct entities and stay distinct (homonyms). §4.11: no corrective exhibit N (no seated composition at another address was named for correction)." ], "composition_v09": { "source": "this row (composed directly; no rebuild/negative-of-the-negative/v09/search-arena.json)", "composer": "Claude (Opus 5.5), subagent, 2026-10-29", "defining_material_read": [ { "dep": "MMRS", "file": "/home/claude/fleet/machinemediation-org/data/meta-aggregator/census.json", "read": "rows[slug == 'search-arena'] whole (sources, unchecked, lifecycle, observed, roles, relations, evidence and basis, properties, note), the census's name and schema_note; the 2026-10-28 copy at rebuild/capture-registry/intake-20261029-aggregators/census-20261028.json, row search-arena (differs only by the capture relation)" }, { "dep": "MMRS", "file": "/home/claude/fleet/machinemediation-org/scripts/meta_aggregator/articles.py", "read": "ARTICLES['search-arena'] whole" }, { "dep": "MMRS", "file": "/home/claude/fleet/machinemediation-org/aggregators/search-arena/index.html", "read": "the rendered record, as text, whole" }, { "dep": "capture-registry", "file": "data/EA-WG-CAPTURES-01.json", "read": "entry search-arena-leaderboard-aio-20261029 whole (transcript, transcript_raw, cite_list, d, reading, findings, analysis, originator, notes)" }, { "dep": 741, "file": "data/texts/AXN-0291-text.md", "read": "the candidate paragraphs and their neighbours (¶28–¶30, ¶37–¶39, the reference list)" }, { "dep": 1705, "file": "data/texts/AXN-0728-text.md", "read": "the section '\"Best\" is an attributization' (¶42–¶44)" }, { "dep": 1690, "file": "data/texts/AXN-0700-text.md", "read": "§0–§6 (the specification)" } ], "cuts": [], "notes": "Composed directly in this row on the brief. KO 28 sentences, P body 348 words, KO_B 10 sentences, P_B body 100 words. Every field claim (48) and every archive claim (39) is cited in the KO; quotes in the KO were checked by script against the ledger quotes of the claims each sentence cites." }, "field_by_origin": { "cards": [ "B1", "B2" ], "own_pages": [ "B3", "B4", "B5", "B6" ], "wikipedia": [ "B7", "B8" ], "unreachable": [ "B1", "B2" ], "title_only": [], "wikipedia_state": "no article: 'Search Arena' answers HTTP 404 for the raw wikitext and the rendered page ('Wikipedia does not have an article with this exact name.'), no redirect (B7). The operator has an article, 'Arena.ai' (B8), to which 'LMArena' and 'Chatbot Arena' redirect (R from move; R from former name); as read on 2026-10-29 its raw wikitext (9,055 bytes) does not mention Search Arena: 'search' occurs once, inside 'Research has identified specific limitations in Arena's methodology.', and the article names no search arena, leaderboard category or dataset paper.", "sources": [ { "id": "B1", "origin": "card", "role": "Overview card 1 at T1; also an own page the MMRS record cites (arena.ai/leaderboard/search, read 2026-10-28 'for counts')", "card": "Arena AI — AI Search Leaderboard: Top Models for Research", "url": "https://arena.ai/leaderboard/search", "kind": "Overview card (own page)", "surfaced_by_T": [ { "slug": "search-arena-leaderboard-aio-20261029", "card": 1 } ], "files": { "B1.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B1.html", "sha256": "a94c489196949291fbd7524a7d5de05f2324d00acc9e11faaa06a2a5d71de0b3" }, "B1.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B1.txt", "sha256": "d44c878b16cb238c3c267f979438cd2a940ab62532af72961dde5e9cda9b23a4" } }, "fetched": { "at": "2026-10-29T05:05:51Z", "http": 403, "url": "https://arena.ai/leaderboard/search", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B1.html", "raw_sha256": "a94c489196949291fbd7524a7d5de05f2324d00acc9e11faaa06a2a5d71de0b3", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B1.txt" ] }, "unreachable": true, "card_title": "AI Search Leaderboard: Top Models for Research", "card_snippet": "AI Search Leaderboard: Top Models for Research Start Voting Agent Chat Code Start Voting Agent Chat Code Image Video View overall rankings across LLMs with inte...", "card_note": null, "names_operator": "yes (the card's site, 'Arena AI'; the saved page is a challenge page and names nothing)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "webfetch": { "at": "2026-10-29T05:06Z (about)", "tool": "WebFetch (a summarising fetch, §3.1(a): not a fetch; nothing ledgered from it)", "reported": "the page 'describes itself as \"View overall rankings across LLMs with integrated web search.\"'; 'a date of Feb 25, 2026, 247,944 votes, and 22 models'; top rows by provider only: '1: Anthropic, 1255±10 (3,607 votes)', '2: xAI, 1225±8 (4,687 votes)', '3: OpenAI, 1219±6 (20,150 votes)', '4: Google, 1218±6 (25,311 votes)', '5: Google, 1214±5 (31,966 votes)'; 'Diffbot (1024±8) at rank 21 and OpenAI (1006±11) at rank 22'; a section 'Default Leaderboard Plots'" }, "note": "unreachable verbatim: curl answered HTTP 403 with a Cloudflare 'Just a moment...' challenge page (saved as B1.html, 5,421 bytes; its text B1.txt is the challenge title only); no retry with another client was made (the session declined to work around the challenge). The card's snippet stands as its only text. The page as a summarising fetch reported it (webfetch) agrees, figure for figure, with the 2026-02-26 snapshot of Arena's own published leaderboard data (B5, F39), not with the latest (2026-10-13, F38); the page's displayed date (25 February) and the snapshot's publish date (26 February) differ by a day. What a browser renders at the address on 2026-10-29 is not established here." }, { "id": "B2", "origin": "card", "role": "Overview card 2 at T1 (Arena's leaderboard overview, all arenas)", "card": "Arena AI — Arena Leaderboard: Official AI Model Rankings & Benchmarks", "url": "https://arena.ai/leaderboard", "kind": "Overview card (operator's page)", "surfaced_by_T": [ { "slug": "search-arena-leaderboard-aio-20261029", "card": 2 } ], "files": { "B2.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B2.html", "sha256": "e562ebae953ba5de14eac9dddb0b9b48b0cf52d30c789a89c47876f2d6f0e694" }, "B2.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B2.txt", "sha256": "d44c878b16cb238c3c267f979438cd2a940ab62532af72961dde5e9cda9b23a4" } }, "fetched": { "at": "2026-10-29T05:05:52Z", "http": 403, "url": "https://arena.ai/leaderboard", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B2.html", "raw_sha256": "e562ebae953ba5de14eac9dddb0b9b48b0cf52d30c789a89c47876f2d6f0e694", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B2.txt" ] }, "unreachable": true, "card_title": "Arena Leaderboard: Official AI Model Rankings & Benchmarks", "card_snippet": "GPT-6.1 Terra is #8 in Agent · Max Kimi K3.1 Pro is #9 in WebDev · Max MAI-Image-2.7-Preview is #3 in Image Edit Spark-X2.6 is #71 in WebDev Krea 3 Ultra is #9 ...", "card_note": null, "names_operator": "yes (the card's site, 'Arena AI')", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "webfetch": { "at": "2026-10-29T05:06Z (about)", "tool": "WebFetch (summarising; nothing ledgered from it)", "reported": "headings 'Leaderboard Overview', 'Arena Overview'; 'See how leading AI models stack up across text, image, vision, and more.'; 'The page contains no text mentioning Search'" }, "note": "unreachable verbatim: HTTP 403, Cloudflare challenge (B2.html). Found by its title: the Arena site's leaderboard overview, the parent of B1. The card's snippet, ranks of models in other arenas (Agent, WebDev, Image Edit), stands as its only text; it names no search model." }, { "id": "B3", "origin": "own_page", "role": "the entity's launch post on Arena's blog; named by the task; not cited by the MMRS record and not surfaced by T", "card": "Arena.ai blog — Introducing the Search Arena: Evaluating Search-Enabled AI", "url": "https://arena.ai/blog/search-arena", "kind": "own page (blog post)", "surfaced_by_T": "no (not surfaced by T; admitted beside the cards by the operator's ruling of 2026-10-29)", "files": { "B3.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B3.html", "sha256": "251720f2c64e59fb72f01d73b7cbaeab5f9e82dd1d010ae5aa0be31463266dba" }, "B3.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B3.txt", "sha256": "e0d10dc29692ef837f358c25fd2cb398a2e386876510a5f6c4a769dcde6fa24b" } }, "fetched": { "at": "2026-10-29T05:05:52Z", "http": 200, "url": "https://arena.ai/blog/search-arena", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B3.html", "raw_sha256": "251720f2c64e59fb72f01d73b7cbaeab5f9e82dd1d010ae5aa0be31463266dba", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B3.txt" ] }, "names_operator": "yes ('Arena Team'; '© 2026 Arena Intelligence Inc.')", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "Published on 15 Apr 2025, last updated on 2 Aug 2025, by 'Arena Team' with eight named authors; its own citation gives the post's earlier address, blog.lmarena.ai/blog/2025/search-arena/. Its figures and tables (the leaderboard, Fig. 1, Tab. 3) are images; their values are not in the text and are not ledgered." }, { "id": "B4", "origin": "own_page", "role": "the dataset paper's arXiv abstract page; cited by the MMRS record (arxiv.org/abs/2506.05334, read 2026-10-28)", "card": "arXiv — [2506.05334] Search Arena: Analyzing Search-Augmented LLMs", "url": "https://arxiv.org/abs/2506.05334", "kind": "own page (paper, abstract page)", "surfaced_by_T": "no (not surfaced by T; admitted beside the cards by the operator's ruling of 2026-10-29)", "files": { "B4.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B4.html", "sha256": "b35a50c458405ec9125f68069491164c4fd29acbc345f0bc576e290229f92589" }, "B4.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B4.txt", "sha256": "22c03d0fef7adb358c98e1cd04f44d42dab65c0bc054e0f6947a9893d8c5d062" } }, "fetched": { "at": "2026-10-29T05:05:53Z", "http": 200, "url": "https://arxiv.org/abs/2506.05334", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B4.html", "raw_sha256": "b35a50c458405ec9125f68069491164c4fd29acbc345f0bc576e290229f92589", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B4.txt" ] }, "names_operator": "no ('Arena' absent as operator; the authors include Wei-Lin Chiang and Anastasios N. Angelopoulos, whom Wikipedia names as Arena's founders, B8)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "v1 5 June 2025, v2 2 March 2026, 'Accepted to ICLR 2026'; the HTML links the code at github.com/lmarena/search-arena and the dataset at huggingface.co/datasets/lmarena-ai/search-arena-24k (the text renders both as 'this https URL'). The PDF was not fetched; the abstract is the ledgered text." }, { "id": "B5", "origin": "own_page", "role": "the leaderboard's own published data: Arena's Hugging Face dataset lmarena-ai/leaderboard-dataset, subsets search and search_style_control; fetched as 'any data it loads' in place of the unreadable page; not cited by the MMRS record and not surfaced by T", "card": "Hugging Face (lmarena-ai) — Arena Leaderboard Dataset, subset search", "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset", "kind": "own data (dataset card and rows)", "surfaced_by_T": "no (not surfaced by T; admitted beside the cards as the leaderboard's own data)", "files": { "B5.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.html", "sha256": "729a6836c088af3e6fa7f96d4f867a4628d626e0afa9d260f130c572fe110c00" }, "B5.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.txt", "sha256": "7a2e36e91b366c1d66407cc53f112f7f401403275a1becff6630d6f38b5a09c0" }, "B5.README.md": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.README.md", "sha256": "b13bf3169c5ec5ffc1e1f4c3b4307de28343e4ed272ad972e8a18da460b7af00" }, "B5.hf-search-latest.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-latest.json", "sha256": "f5ee84f9d2bd1b581c08e39ce6ee0f7da982a3b3c5ee6a42d5f9c8dace83249b" }, "B5.hf-search_style_control-latest.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search_style_control-latest.json", "sha256": "bec4d22199bd44aa445a17a0987a29cc7a48975872156f4c215af2cc48924419" }, "B5.hf-search-full-dates.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-dates.json", "sha256": "7df292de7bae2ddf776b95a16d4a3566d779ea24e2cf771c8a83011ebed79049" }, "B5.hf-search-full-0.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-0.json", "sha256": "1b95929ca99acfb4a0d2ffa83651d481ee710f14abf034e535fdab02899dbcbe" }, "B5.hf-search-full-100.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-100.json", "sha256": "4638309a0612e4240b9842cac43376a428744bebdb5bba5117607e883fd85a2a" }, "B5.hf-search-full-200.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-200.json", "sha256": "803334809b94eca08a41788014ac2f03f4c0f08b85e0f37a35b413a89eeb4819" }, "B5.hf-search-full-300.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-300.json", "sha256": "41cbe6bf16e03619f661823c98bfe7be19701cde342214812d975a28a910e417" }, "B5.hf-search-full-400.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-400.json", "sha256": "7e1999f30f2d158bd4114bacc540ce9db7f1b9f6c8da8886ea7cef0404c86b82" }, "B5.hf-search-full-500.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-500.json", "sha256": "63b00f44b1733d1576590ad34328960eb0604e5d6a24639aa80a48758d438e6e" }, "B5.hf-search-full-600.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-600.json", "sha256": "719b68a38ef0507e927a61ec00c96008a0fdfcfbfd613d9b8d2e29b1e32748c7" }, "B5.hf-search-full-700.json": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.hf-search-full-700.json", "sha256": "8352e05a51bc35f1a823d17806b983c455114177fa7ffe8ca2ebc646aa0f6dc5" } }, "fetched": { "at": "2026-10-29T05:07:35Z–05:09:17Z", "http": 200, "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.html", "raw_sha256": "729a6836c088af3e6fa7f96d4f867a4628d626e0afa9d260f130c572fe110c00", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.txt", "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B5.README.md" ], "data": { "latest (search)": "https://datasets-server.huggingface.co/rows?dataset=lmarena-ai/leaderboard-dataset&config=search&split=latest&offset=0&length=100 (05:07:35Z, 200; 42 rows)", "latest (search_style_control)": "same, config=search_style_control (05:07:35Z, 200; 42 rows)", "full (search)": "same, split=full, offsets 0–700 (05:08:12Z, 200; 764 rows, 41 publish dates 2025-08-07 to 2026-10-13)", "statistics (search, full)": "https://datasets-server.huggingface.co/statistics?dataset=lmarena-ai/leaderboard-dataset&config=search&split=full (05:07:54Z, 200)", "README": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/raw/main/README.md (05:09:17Z, 200)" }, "dataset_last_modified": "2026-10-29T03:05:31.000Z (Hub API, sha 4491b06c6382aebe3224461fc7209b57f7beb1ae)" }, "names_operator": "yes ('Arena' in the README; the namespace lmarena-ai)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "Arena announces the dataset as 'the entire history of those leaderboards' (B6). The latest search snapshot (published 2026-10-13, 42 models) gives, rank for rank 1–11, the models, organizations, ratings (rounded) and confidence half-widths (rounded) that T's table gives; the 2026-02-26 snapshot (22 models) gives, figure for figure, what the summarising fetch reported of the page (B1, webfetch). The data's vote_count is 'Number of battles for this model'; no snapshot's sum reproduces the page's 247,944 votes (2026-02-26: 479,042 battle-counts, i.e. 239,521 if each battle is counted twice; 2026-02-25: 475,340). Quotes from the data are verbatim spans of the API's JSON." }, { "id": "B6", "origin": "own_page", "role": "Arena's blog post announcing B5; not cited by the MMRS record and not surfaced by T", "card": "Arena.ai blog — Arena Leaderboard Dataset", "url": "https://arena.ai/blog/arena-leaderboard-dataset/", "kind": "own page (blog post)", "surfaced_by_T": "no (not surfaced by T; admitted as the provenance of B5)", "files": { "B6.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B6.html", "sha256": "945b31ff964c24cd4052062010c8d49f6256ccd579bfb0de9af50495a180e80d" }, "B6.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B6.txt", "sha256": "e91db567eb6ab2c916e5309455a3a27e0aa6c61b0f0c0f6bb659558a81d920de" } }, "fetched": { "at": "2026-10-29T05:09:19Z", "http": 200, "url": "https://arena.ai/blog/arena-leaderboard-dataset/", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B6.html", "raw_sha256": "945b31ff964c24cd4052062010c8d49f6256ccd579bfb0de9af50495a180e80d", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B6.txt" ] }, "names_operator": "yes ('Arena Team'; 'Arena has been publishing leaderboards')", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "Published on 2 Apr 2026, 'Arena Team'." }, { "id": "B7", "origin": "wikipedia", "role": "Wikipedia's page for the entity's name: none", "card": "Wikipedia — Search Arena (no article)", "url": "https://en.wikipedia.org/wiki/Search_Arena", "kind": "Wikipedia (absence)", "surfaced_by_T": "no (not surfaced by T; admitted beside the cards by the operator's ruling of 2026-10-29)", "files": { "B7.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7.html", "sha256": "7b2ce79be00f677cd706008e8b4843f5b82c2372c0278f7386b4da8258117260" }, "B7.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7.txt", "sha256": "5b819aeaa91b2097959ad15c1ab6c9a38f7520e4e4c4e3821f89e19b52a813b8" }, "B7-raw.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7-raw.html", "sha256": "65139f38e1a266c847fd7ed266e67f01996ec6aa8c2c1ab321ad010f1692f7ea" }, "B7-raw.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7-raw.txt", "sha256": "34a97e3b2e1c9da8811395a6251ec557c30eaf935a2edd2e9fd74e308de51c59" } }, "fetched": { "at": "2026-10-29T05:08:53Z", "http": 404, "url": "https://en.wikipedia.org/wiki/Search_Arena", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7.html", "raw_sha256": "7b2ce79be00f677cd706008e8b4843f5b82c2372c0278f7386b4da8258117260", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7.txt" ], "raw_wikitext": { "url": "https://en.wikipedia.org/w/index.php?title=Search_Arena&action=raw&redirect=no", "http": 404, "file": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B7-raw.html", "sha256": "65139f38e1a266c847fd7ed266e67f01996ec6aa8c2c1ab321ad010f1692f7ea" } }, "names_operator": "no", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "no article and no redirect at 'Search Arena': raw wikitext HTTP 404 (Wikimedia error page), rendered page HTTP 404 'Wikipedia does not have an article with this exact name.'" }, { "id": "B8", "origin": "wikipedia", "role": "Wikipedia's article on the operator, Arena (Arena.ai; LMArena and Chatbot Arena redirect to it)", "card": "Wikipedia — Arena.ai", "url": "https://en.wikipedia.org/wiki/Arena.ai", "kind": "Wikipedia article (the operator)", "surfaced_by_T": "no (not surfaced by T; admitted beside the cards by the operator's ruling of 2026-10-29)", "files": { "B8.html": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.html", "sha256": "75667836aeaf4811daddd35422ed8fd89838b06f46f5f9190f099597e8279054" }, "B8.txt": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.txt", "sha256": "15e7f68eb70ffd421fd07cfc4e9cda0dde947c3bf6dfce7c343d0edc1a2e2c79" }, "B8.wiki": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.wiki", "sha256": "cdf882d3f4f01a813bcd5f6600e47a469b74c8def083e1de580633350700d7c0" }, "B8-redirect-LMArena.wiki": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8-redirect-LMArena.wiki", "sha256": "5f608ef8b6839e0fd3fc7e88180ccd940d36027fb92e8a436a9b2cac782a5fa7" }, "B8-redirect-Chatbot_Arena.wiki": { "path": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8-redirect-Chatbot_Arena.wiki", "sha256": "2ce8ba248f74faacfa0ab3e4104c2a1bb5c8d884e11b6312bae77ad58a1200ad" } }, "fetched": { "at": "2026-10-29T05:08:53Z", "http": 200, "url": "https://en.wikipedia.org/wiki/Arena.ai", "raw": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.html", "raw_sha256": "75667836aeaf4811daddd35422ed8fd89838b06f46f5f9190f099597e8279054", "text": [ "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.txt", "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.wiki" ], "raw_wikitext": { "url": "https://en.wikipedia.org/w/index.php?title=Arena.ai&action=raw", "http": 200, "file": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8.wiki", "sha256": "cdf882d3f4f01a813bcd5f6600e47a469b74c8def083e1de580633350700d7c0", "bytes": 9055 }, "redirects": { "LMArena": { "url": "https://en.wikipedia.org/w/index.php?title=LMArena&action=raw&redirect=no", "http": 200, "file": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8-redirect-LMArena.wiki", "text": "#REDIRECT [[Arena.ai]] (R from move)" }, "Chatbot_Arena": { "url": "https://en.wikipedia.org/w/index.php?title=Chatbot_Arena&action=raw&redirect=no", "http": 200, "file": "datasets/negative-of-the-negative/v2/intake/search-arena-20261029/B8-redirect-Chatbot_Arena.wiki", "text": "#REDIRECT [[Arena.ai]] (R from move, R from former name, R printworthy)" } } }, "names_operator": "yes (the article is the operator's)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "the article does not mention Search Arena: the raw wikitext (9,055 bytes) contains 'search' only inside 'Research' (one occurrence, 'Research has identified specific limitations in Arena's methodology.'), and neither 'Search Arena' nor 'search' as a word; the rendered text likewise names no search arena. Its infobox gives founders Wei-Lin Chiang, Anastasios N. Angelopoulos and Ion Stoica, launch 24 April 2023." } ] } }