pith. sign in

arxiv: 2510.14240 · v5 · submitted 2025-10-16 · 💻 cs.AI

LiveResearchBench: A Live Benchmark for User-Centric Deep Research in the Wild

Jiayu Wang , Yifei Ming , Riya Dulepet , Qinglin Chen , Austin Xu , Zixuan Ke , Frederic Sala , Aws Albarghouthi
show 2 more authors
Caiming Xiong Shafiq Joty
This is my paper

Pith reviewed 2026-05-18 06:58 UTC · model grok-4.3

classification 💻 cs.AI
keywords benchmarkdeep researchAI agentsweb searchcitation evaluationlong-form generationmulti-agent systemsuser-centric tasks
0
0 comments X

The pith

A benchmark of 100 expert-curated tasks enables systematic evaluation of AI systems that synthesize citation-grounded reports from live web sources.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces LiveResearchBench to address gaps in testing deep research agents, which must search and synthesize information across hundreds of dynamic web pages into comprehensive reports. It defines four principles for good tasks: they should reflect real user needs, demand current information, allow consistent interpretation, and require multi-faceted analysis. The authors built the benchmark through extensive human curation and paired it with DeepEval, a suite of protocols that scores reports on coverage, citations, consistency, and depth while aiming for high agreement with human judges. This combination lets them compare seventeen frontier systems and surface common failure patterns along with promising design elements. A reader would care because better evaluation tools can guide the creation of agents that reliably handle complex, real-world information tasks instead of relying on narrow or outdated tests.

Core claim

LiveResearchBench consists of 100 tasks spanning daily life, enterprise, and academic domains, each constructed to require extensive real-time web search and synthesis into citation-grounded long-form reports. DeepEval supplies four complementary evaluation protocols that jointly measure content quality, report-level attributes such as citation accuracy and association, and overall analytical depth, with the explicit goal of producing stable scores that align with human judgment. Together these resources support fair comparison across single-agent search systems, deep research agents, and multi-agent setups, revealing both recurring limitations and the system components that correlate with a

What carries the argument

LiveResearchBench is the set of 100 expert-curated, dynamic tasks that enforce user-centric, unambiguous, and search-intensive requirements; DeepEval is the accompanying evaluation suite that applies content-level and report-level metrics through four protocols to produce stable, human-aligned scores on citation-grounded outputs.

If this is right

  • Current systems exhibit recurring weaknesses in citation accuracy, source association, and depth of analysis that limit report reliability.
  • Design choices such as multi-step planning and explicit verification steps correlate with better performance on multi-faceted tasks.
  • The benchmark enables repeated evaluation as web content evolves, supporting longitudinal tracking of agent progress.
  • Single-agent versus multi-agent architectures show measurable differences in handling the search-intensive and synthesis demands of the tasks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Widespread adoption could shift research focus from narrow question-answering metrics toward end-to-end report quality in agent development.
  • The task design principles could extend to evaluating agents on private or domain-specific data sources beyond public web search.
  • Periodic refresh of the task set would keep the benchmark aligned with changing real-world information needs.

Load-bearing premise

Expert curation over 1,500 hours produces tasks that stay unambiguous and genuinely demand multi-source live search for every user without introducing interpretation differences or hidden biases that would make comparisons across systems unfair.

What would settle it

Observe whether independent users or different research systems produce reports that satisfy the task requirements yet differ substantially in scope, sources, or conclusions, or whether many tasks can be answered adequately from model parameters alone without live search.

Figures

Figures reproduced from arXiv: 2510.14240 by Austin Xu, Aws Albarghouthi, Caiming Xiong, Frederic Sala, Jiayu Wang, Qinglin Chen, Riya Dulepet, Shafiq Joty, Yifei Ming, Zixuan Ke.

Figure 1
Figure 1. Figure 1: Recent efforts such as DeepResearchGym (Coelho et al., 2025) adopts short and static [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Domain distribution and task coverage of LiveResearchBench. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Six-stage data generation pipeline for LiveResearchBench. We begin with user interviews [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Five-stage data verification pipeline. Research questions and checklists are first [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of rubric tree–based evaluation of citation accuracy. Compared to evaluating [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Distribution of report lengths across systems and tasks. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Analysis depth (win rate over Open Deep Research). Most systems collect and organize information but struggle to synthesize deeper insights. Obs.❺ Multi-agent systems lead presentation, but surface polish does not imply grounded quality. Multi-agent systems achieve the highest presentation scores (77.7 avg.), with Open Deep Research (81.0) and Deerflow+ (78.8) producing the most polished outputs. Perplexit… view at source ↗
Figure 8
Figure 8. Figure 8: Instruction for evaluating presentation and organization quality of research reports. Placeholders such as {user_prompt}, {query}, {checklist_section} and {report_- content} are filled at runtime. A.2 FACTUAL & LOGICAL CONSISTENCY Grading prompts and rubrics. The grading prompt for factual and logical inconsistency is provided in [PITH_FULL_IMAGE:figures/full_fig_p017_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Instruction for evaluating research reports on logical and factual inconsistency. The rubrics [PITH_FULL_IMAGE:figures/full_fig_p019_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Instruction for evaluating Coverage & Comprehensiveness of the generated reports. [PITH_FULL_IMAGE:figures/full_fig_p020_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Instruction for evaluating analysis depth of the generated reports. Placeholders [PITH_FULL_IMAGE:figures/full_fig_p021_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Instruction for evaluating research reports on citation association. The rubrics table is [PITH_FULL_IMAGE:figures/full_fig_p022_12.png] view at source ↗
Figure 22
Figure 22. Figure 22: Overall, these fixes substantially reduce early termination and bring [PITH_FULL_IMAGE:figures/full_fig_p024_22.png] view at source ↗
Figure 13
Figure 13. Figure 13: Qualitative outputs for the same prompt in Figure 1 from [PITH_FULL_IMAGE:figures/full_fig_p033_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Error pattern: mismatched in-text citations and references. [PITH_FULL_IMAGE:figures/full_fig_p036_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Error pattern: missing links (URLs). 37 [PITH_FULL_IMAGE:figures/full_fig_p037_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Error pattern: inconsistent citation format. [PITH_FULL_IMAGE:figures/full_fig_p038_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Error pattern: references not mentioned in the report. [PITH_FULL_IMAGE:figures/full_fig_p039_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Error pattern: broken table format. 40 [PITH_FULL_IMAGE:figures/full_fig_p040_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Error pattern: out-of-order references. 41 [PITH_FULL_IMAGE:figures/full_fig_p041_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Error pattern: embedded citations breaking text flow. [PITH_FULL_IMAGE:figures/full_fig_p042_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Error pattern: hallucinated information. [PITH_FULL_IMAGE:figures/full_fig_p042_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Fine-grained pass rates (P1–P10) for the [PITH_FULL_IMAGE:figures/full_fig_p046_22.png] view at source ↗
read the original abstract

Deep research -- producing comprehensive, citation-grounded reports by searching and synthesizing information from hundreds of live web sources -- marks an important frontier for agentic systems. To rigorously evaluate this ability, four principles are essential: tasks should be (1) user-centric, reflecting realistic information needs, (2) dynamic, requiring up-to-date information beyond parametric knowledge, (3) unambiguous, ensuring consistent interpretation across users, and (4) multi-faceted and search-intensive, requiring search over numerous web sources and in-depth analysis. Existing benchmarks fall short of these principles, often focusing on narrow domains or posing ambiguous questions that hinder fair comparison. Guided by these principles, we introduce LiveResearchBench, a benchmark of 100 expert-curated tasks spanning daily life, enterprise, and academia, each requiring extensive, dynamic, real-time web search and synthesis. Built with over 1,500 hours of human labor, LiveResearchBench provides a rigorous basis for systematic evaluation. To evaluate citation-grounded long-form reports, we introduce DeepEval, a comprehensive suite covering both content- and report-level quality, including coverage, presentation, citation accuracy and association, consistency and depth of analysis. DeepEval integrates four complementary evaluation protocols, each designed to ensure stable assessment and high agreement with human judgments. Using LiveResearchBench and DeepEval, we conduct a comprehensive evaluation of 17 frontier deep research systems, including single-agent web search, single-agent deep research, and multi-agent systems. Our analysis reveals current strengths, recurring failure modes, and key system components needed to advance reliable, insightful deep research. Our code is available at: https://github.com/SalesforceAIResearch/LiveResearchBench.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces LiveResearchBench, a benchmark of 100 expert-curated tasks spanning daily life, enterprise, and academia, designed to evaluate deep research systems that produce comprehensive, citation-grounded reports from live web sources. Guided by four principles (user-centric, dynamic, unambiguous, and multi-faceted/search-intensive), the tasks were built with over 1,500 hours of human labor. The authors also present DeepEval, a suite with four complementary protocols for assessing content- and report-level quality (coverage, presentation, citation accuracy, consistency, depth), and use it to evaluate 17 frontier systems (single-agent search, deep research, and multi-agent), revealing strengths, failure modes, and needed components. Code is released at the provided GitHub link.

Significance. If the curation successfully yields tasks that are verifiably unambiguous and genuinely require extensive multi-source search without shortcuts or interpretation variance, LiveResearchBench would fill an important gap by providing a live, user-centric benchmark for an emerging frontier in agentic AI. The introduction of DeepEval with multiple protocols designed for high human agreement is a constructive contribution, as is the broad evaluation across 17 systems and the public code release. These elements could support more systematic progress tracking if the core task properties are empirically demonstrated.

major comments (2)
  1. [Abstract / Task Curation] Abstract and task creation description: the central claim that LiveResearchBench supplies a 'rigorous basis' for evaluation rests on the assertion that 1,500 hours of expert curation produces tasks that are unambiguously interpretable and require genuine multi-source search. No inter-annotator agreement metrics, pilot studies with multiple user cohorts, or empirical source-count distributions are reported to validate these properties, leaving the four principles as unverified assertions rather than demonstrated features. This directly affects the reliability of the 17-system comparison.
  2. [DeepEval and System Evaluation] Evaluation section (implied by the 17-system study): while DeepEval is described as integrating four protocols for stable assessment, the manuscript provides no quantitative results on agreement with human judgments or ablation of the individual protocols, making it difficult to assess whether the reported strengths and failure modes are robust.
minor comments (2)
  1. [Abstract] The abstract states tasks span 'daily life, enterprise, and academia' but does not break down the 100 tasks by category or provide example task statements; adding a table or appendix with representative examples would improve clarity.
  2. [Conclusion / Code Availability] The GitHub link is provided, but the manuscript does not specify which artifacts (task definitions, DeepEval implementation, or evaluation scripts) are included; explicit enumeration would aid reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and for acknowledging the potential contributions of LiveResearchBench and DeepEval. We address each major comment below and indicate where we will revise the manuscript to incorporate additional evidence.

read point-by-point responses

  1. Referee: [Abstract / Task Curation] Abstract and task creation description: the central claim that LiveResearchBench supplies a 'rigorous basis' for evaluation rests on the assertion that 1,500 hours of expert curation produces tasks that are unambiguously interpretable and require genuine multi-source search. No inter-annotator agreement metrics, pilot studies with multiple user cohorts, or empirical source-count distributions are reported to validate these properties, leaving the four principles as unverified assertions rather than demonstrated features. This directly affects the reliability of the 17-system comparison.

    Authors: We agree that quantitative validation would strengthen the central claims. The 1,500 hours of curation involved iterative expert review to enforce the four principles, but the manuscript does not report inter-annotator agreement, pilot-study results, or source-count distributions. In the revised version we will add these metrics, including curator agreement scores, descriptions of pilot cohorts, and empirical distributions of sources per task, to empirically support the task properties and the reliability of the system comparisons. revision: yes

  2. Referee: [DeepEval and System Evaluation] Evaluation section (implied by the 17-system study): while DeepEval is described as integrating four protocols for stable assessment, the manuscript provides no quantitative results on agreement with human judgments or ablation of the individual protocols, making it difficult to assess whether the reported strengths and failure modes are robust.

    Authors: We appreciate the observation. The four protocols were designed for stability and high human agreement, yet the current manuscript does not include the corresponding quantitative agreement statistics or protocol ablations. We will revise the evaluation section to report correlation with human judgments and ablation results for each protocol, thereby providing direct evidence for the robustness of the strengths and failure modes observed across the 17 systems. revision: yes

Circularity Check

0 steps flagged

No circularity: benchmark and evaluation suite introduced as independent contributions

full rationale

The paper's core contribution is the expert-curated LiveResearchBench dataset of 100 tasks and the accompanying DeepEval evaluation suite, both presented as novel artifacts built from 1,500 hours of human labor. No mathematical derivations, equations, fitted parameters, or predictions appear in the provided text. The four guiding principles are stated as design criteria rather than derived results, and no self-citations or prior author work are invoked to justify uniqueness or force the benchmark's properties. The claims rest on the described curation process and evaluation protocols, which are self-contained and do not reduce to equivalent inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work rests on standard benchmarking assumptions plus the domain claim that expert curation can reliably satisfy the four stated principles.

axioms (1)
  • domain assumption Expert human curation can produce tasks that are simultaneously user-centric, dynamic, unambiguous, and multi-faceted.
    This premise is invoked to justify the benchmark construction and is required for the claim of rigorous, fair evaluation.

pith-pipeline@v0.9.0 · 5875 in / 1231 out tokens · 29975 ms · 2026-05-18T06:58:43.265659+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search Systems

    cs.CL 2026-05 unverdicted novelty 7.0

    BRIGHT-Pro and RTriever-Synth advance reasoning-intensive retrieval by adding multi-aspect evidence evaluation and aspect-decomposed synthetic training, with the fine-tuned RTriever-4B showing gains over its base model.

  2. OpenJarvis: Personal AI, On Personal Devices

    cs.LG 2026-05 unverdicted novelty 6.0

    OpenJarvis decomposes personal AI into Intelligence, Engine, Agents, Tools & Memory, and Learning primitives and applies LLM-guided spec search to produce on-device configurations that reach within 3.2 pp of cloud bas...

  3. MARCA: A Checklist-Based Benchmark for Multilingual Web Search

    cs.CL 2026-04 accept novelty 6.0

    MARCA is a bilingual benchmark using 52 questions and validated checklists to evaluate LLM web-search completeness and correctness in English and Portuguese.

  4. AutoResearch AI: Towards AI-Powered Research Automation for Scientific Discovery

    cs.AI 2026-05 unverdicted novelty 4.0

    A survey organizing AI-powered research automation into five workflow stages, defining AutoResearch and Vibe Research, and proposing five evaluation dimensions while noting domain-conditioned limits on autonomy.

Reference graph

Works this paper leans on

66 extracted references · 66 canonical work pages · cited by 4 Pith papers

  1. [1]

    References

    URLhttps://arxiv.org/abs/2409.16191. Haochen Tan, Zhijiang Guo, Zhan Shi, Lu Xu, Zhili Liu, Yunlong Feng, Xiaoguang Li, Yasheng Wang, Lifeng Shang, Qun Liu, and Linqi Song. Proxyqa: An alternative framework for evaluating long-form text generation with large language models, 2024. URL https://arxiv.org/ abs/2401.15042. Peiyi Wang, Lei Li, Liang Chen, Zefa...

    work page doi:10.18653/v1/2025.acl-long.470 2024

  2. [2]

    Factual accuracy (e.g., whether the report is free of factual errors) is a separate criterion and should NOT be considered here

    work page

  3. [3]

    In section X

    The same source can be used to cite multiple claims. This does NOT constitute a contradiction. OUTPUT FORMAT: Respond in JSON with the following fields: -specific_issues: A list of specific problems, with exact quotes or locations in the text (e.g., “In section X...”, “The claim that ‘...’ is unsupported”). -total_issues: Total number of issues (must matc...

    work page

  4. [4]

    Compare both reportsonly on depth, using the rubric above

    work page

  5. [5]

    Give each report a0–25 depth score

    work page

  6. [6]

    winner":

    Decide the outcome: - If one report’s score is more than 1 point higher→it wins. - If the difference is≤1point→call it a tie. Exclude Entirely - Coverage (breadth of topics). - Factual correctness or consistency. - Presentation, formatting, style, grammar. - Citation traceability. - Length alone (verbosity̸=depth). Output Format (JSON only) { "winner": "A...

    work page 2024

  7. [7]

    A URL may not always be attached immediately after a claim; in some cases, it appears at the end of a paragraph and is intended to support the preceding claims within that paragraph

    work page

  8. [8]

    In section X

    Factual accuracy (e.g., whether the report is free of factual errors) is a separate criterion and should NOT be considered here. OUTPUT FORMAT: Respond in JSON with the following fields: -specific_issues: A list of specific problems, with exact quotes or locations in the text (e.g.,“In section X...”, “The claim that ‘...’ is unsupported”). -total_issues: ...

    work page arXiv 1900

  9. [9]

    emerged within state and literati-dominated systems of production and taste [17][18]. Literati painting emphasized brushwork, monochrome ink, and poetic inscription, while academies cultivated refined mineral colors and precise draftsmanship; these aesthetics were shaped further under foreign dynasties, including the Mongol Yuan [18]. - Representative wor...

    work page

  10. [10]

    | | EGYPT_MAMLUK_OTTOMAN | Spice-route shifts; Ottoman integration and patronage reorientation [20] | | WCSUDAN_ISLAM | Islam's role; monumental adobe architecture; scholarship; coinage/trade

    work page

  11. [11]

    | | OTTOMAN_SILKS | Bursa workshops; nakkashane codification; palace demand and export; sericulture shifts [25] | | JAPAN_UKIYOE_POLY | Collaborative production; kento registration; urban subjects [26][27] | | EUROPE_19C_PRINT | Industrial proliferation; etching revival; Japonisme [28] | --- ### 6. Debates, Institutions, and Markets - Figuration and image...

    work page 2016

  12. [12]

    routledge.com/Studies-In-Iconology-Humanistic-Themes-In-The-Art-Of-The-Renaissance/ Panofsky-Panofsky/p/book/9780064300254

    Studies in Iconology: Humanistic Themes in the Art of the Renaissance: https://www. routledge.com/Studies-In-Iconology-Humanistic-Themes-In-The-Art-Of-The-Renaissance/ Panofsky-Panofsky/p/book/9780064300254

    work page arXiv

  13. [13]

    Painting and Experience in Fifteenth-Century Italy (Internet Archive item): https:// archive.org/details/paintingexperien00baxa

    work page

  14. [14]

    The Painting of Modern Life (Princeton University Press): https://press.princeton.edu/ books/paperback/9780691009032/the-painting-of-modern-life

    work page arXiv

  15. [15]

    Why Have There Been No Great Women Artists? (Thames & Hudson product page): https://www .thamesandhudson.com/products/why-have-there-been-no-great-women-artists

    work page

  16. [16]

    Orientalism (Goodreads editions page): https://www.goodreads.com/book/show/23300798- orientalism

    work page arXiv

  17. [17]

    The Predicament of Culture (Oxfam Books listing): https://www.oxfambookshop.au/products /581532_clifford_predicamentofculture

    work page

  18. [18]

    Global Art History (De Gruyter/Brill): https://www.degruyter.com/document/cover/isbn /9783839440612/product_pages

    work page

  19. [19]

    routledge.com/African-Indigenous-Knowledges-in-a-Postcolonial-World-A-Tribute-to-Toyin /Afolayan-Babalola-Ibrahim/p/book/9780367516833

    African Indigenous Knowledges in a Postcolonial World (Routledge): https://www. routledge.com/African-Indigenous-Knowledges-in-a-Postcolonial-World-A-Tribute-to-Toyin /Afolayan-Babalola-Ibrahim/p/book/9780367516833

    work page arXiv

  20. [20]

    Artforum (letters image associated with McEvilley's critique): https://www.artforum.com /wp-content/uploads/2016/09/article00_large-359.jpg

    work page 2016

  21. [21]

    Artforum (additional page image referencing the critique): https://www.artforum.com/wp -content/uploads/2017/01/article_large-261.jpg

    work page 2017

  22. [22]

    The Met Heilbrunn Timeline of Art History: https://www.metmuseum.org/toah

    work page

  23. [23]

    Oxford Art Online (Grove Art Online): https://digital.library.wisc.edu/1711.web/ oxfordart

    work page

  24. [24]

    Smarthistory (About/Essays): https://smarthistory.org/

    work page

  25. [25]

    JSTOR Archive Journals (about access): https://eifl.net/e-resources/jstor-archive- journals

    work page

  26. [26]

    Project MUSE (About): https://about.muse.jhu.edu/

    work page

  27. [27]

    Unit 7: Trade and Artistic Exchange (Met educator curriculum, Islamic world): https:// www.metmuseum.org/learn/educators/curriculum-resources/~/media/Files/Learn/For%20 Educators/Publications%20for%20Educators/Islamic%20Teacher%20Resource/Unit7.pdf

    work page

  28. [28]

    | Heilbrunn Timeline of Art History: https://www.metmuseum.org/ toah/ht/06/eac.html

    China, 500-1000 A.D. | Heilbrunn Timeline of Art History: https://www.metmuseum.org/ toah/ht/06/eac.html

    work page

  29. [29]

    China, 1000-1400 A.D.: https://www.metmuseum.org/toah/ht/07/eac.html 32

    work page

  30. [30]

    Eastern and Southern Africa, 1400-1600 A.D.: https://www.metmuseum.org/toah/ht/08/afa. html

    work page

  31. [31]

    Egypt, 1400-1600 A.D.: https://www.metmuseum.org/toah/ht/08/afe.html

    work page

  32. [32]

    Guinea Coast, 1400-1600 A.D.: https://www.metmuseum.org/toah/ht/08/afg.html

    work page

  33. [33]

    Western and Central Sudan, 1000-1400 A.D.: https://www.metmuseum.org/toah/ht/07/afu. html

    work page

  34. [34]

    Age of Transition: Byzantine Culture in the Islamic World (Met symposia volume): https ://resources.metmuseum.org/resources/metpublications/pdf/ Age_of_Transition_Byzantine_Culture_in_the_Islamic_World.pdf

    work page

  35. [35]

    Tile from a Squinch (muqarnas element; Met 20.120.189): https://www.metmuseum.org/art/ collection/search/447256

    work page

  36. [36]

    Silks from Ottoman Turkey (Bursa sericulture; nakkashane; saz/chintamani): https://www .metmuseum.org/essays/silks-from-ottoman-turkey

    work page

  37. [37]

    metmuseum.org/essays/woodblock-prints-in-the-ukiyo-e-style

    Woodblock Prints in the Ukiyo-e Style (technique; kento; collaboration): https://www. metmuseum.org/essays/woodblock-prints-in-the-ukiyo-e-style

    work page

  38. [38]

    Art of the Pleasure Quarters and the Ukiyo-e Style (urban culture; subjects; markets): https://www.metmuseum.org/essays/art-of-the-pleasure-quarters-and-the-ukiyo-e-style

    work page

  39. [39]

    good for society

    The Print in the Nineteenth Century (industrial prints; etching revival; Japonisme): https://www.metmuseum.org/essays/the-print-in-the-nineteenth-century Notes on sources and images: All images embedded above are drawn exclusively from the provided resources. Assertions are restricted to the content of the cited materials. Gaps are acknowledged explicitly...

    work page 2023

  40. [40]

    mental health and well-being (e.g., anxiety, depression, self-esteem, and loneliness),

    work page

  41. [41]

    In the report, I would like to have 6 pro arguments and 6 con arguments

    educational and developmental impacts (e.g., learning, attention span, and digital literacy), 3) social and civil outcomes (e.g., friendships, community engagement, political awareness, and misinformation exposure), and 4) safety and risk factors (e.g., cyberbullying, privacy, exploitation, and screen-time effects). In the report, I would like to have 6 p...

    work page 2020

  42. [42]

    car g or x

    h ttps:/ / www . car g or x. c om/blog/ elec tric -trucks-hit -the-tipping-poin t -is-2025-the- y ear

    work page 2025

  43. [43]

    c om/indu str y -an aly sis / nor th-america-elec tric -truck -m ark e t

    h ttps:/ / www .gminsigh ts. c om/indu str y -an aly sis / nor th-america-elec tric -truck -m ark e t

    work page

  44. [44]

    or g/ r epor ts / global-e v -outlook -2025/ tr ends-in-hea vy -duty -elec tric - v ehicles

    h ttps:/ / www .iea. or g/ r epor ts / global-e v -outlook -2025/ tr ends-in-hea vy -duty -elec tric - v ehicles

    work page 2025

  45. [45]

    c om/ ne ws / elec tric -trucks-m ark e t -t o-r each-124-billion-b y -2030/

    h ttps:/ / www .ac t -ne ws. c om/ ne ws / elec tric -trucks-m ark e t -t o-r each-124-billion-b y -2030/ ... C u st omer Segmen ts and A dop tion Driv ers

    work page 2030

  46. [46]

    tr anspor ta tion.g o v / rur al/ e v / t oolkit / e v -bene fits-and-ch alleng es / ch alleng es-and-e v olving- solutions

    h ttps:/ / www . tr anspor ta tion.g o v / rur al/ e v / t oolkit / e v -bene fits-and-ch alleng es / ch alleng es-and-e v olving- solutions

    work page

  47. [47]

    c om/ scienc e / ar ticle /pii/ S1361920924004723

    h ttps:/ / www .scienc edir ec t. c om/ scienc e / ar ticle /pii/ S1361920924004723

    work page

  48. [48]

    c om/blog/ elec tric -truck -ch ar ging-the-in fr astruc tur e-ch alleng es

    h ttps:/ / www .helio x -ener g y . c om/blog/ elec tric -truck -ch ar ging-the-in fr astruc tur e-ch alleng es

    work page

  49. [49]

    h ttps:/ / www .nr el.g o v / ne ws / fea tur e /2025/ the-dawn-o f-elec tric -trucking-calls-for -high-po w er -ch ar ging

    work page 2025

  50. [50]

    bolt out o f the blue

    h ttps:/ / www . cb tne ws. c om/ elec tric - v ehicle-ch ar ging-in fr astruc tur e-ch alleng es-and-solutions / ... Mismat ched in- t e xt citations and r ef er ences An aly sis: The r epor t r eu ses cita tion n umbers acr oss m ultiple r e fer enc e lists, ca u sing ambiguity in m apping in- t e x t cita tions t o sour c es. Figure 14: Error pattern: ...

    work page

  51. [51]

    Ex t ensor carpi uln aris injuries in t ennis pla y ers: a study o f 28 cases

    Mon talv an B, P arier J, Br asseur JL, L e V ie t D , Dr ape JL. Ex t ensor carpi uln aris injuries in t ennis pla y ers: a study o f 28 cases. *Br J Spor ts Med * . 2006;40(5):424–429. (I den tifies ECU t endinopa th y , instability , rup tur e; outlines immobiliza tion and sur gical pr o t oc ols.)

    work page 2006

  52. [52]

    Ph y sio sour c es on ECU t endinopa th y —o v eru se mech anism, s ymp t oms, c ommon in r ack e t spor ts [2015+ g ener al kno wledg e]

    work page 2015

  53. [53]

    less-th an-sa tisfac t or y

    L ogli AL e t al. Uln ar wrist pain in a t ennis pla y er: case r e vie w and disc u ssion. * JB JS R e vie ws * . 2022. (Disc u sses TF C C, UT ligamen t injur y and ar thr osc opic r epair .) ... Mis sing Link s (URLs) An aly sis: In-t e x t cita tions (e .g ., [2], [9]) either lack a pr oper r e fer enc e en tr y or appear in the r e fer enc e list wit...

    work page 2022

  54. [54]

    (2024, A ugu st 21)

    F or d Br oadens Elec trifica tion Str a t eg y t o R each Mor e ... (2024, A ugu st 21). F r om ther oad. for d. c om. h ttps:/ / www . fr om ther oad. for d. c om/ u s / en/ ar ticles /2024/ for d-br oadens-elec trifica tion-str a t eg y -t o-r each- mor e-c u st omers-

    work page 2024

  55. [55]

    F or d ' s R ole in Sh aping the 2024 Elec tric V ehicle Mark e t. (n. d.). Willisfor d. c om. h ttps:/ / www . willisfor d. c om/ for d-elec tric - v ehicle-m ark e t -2024-outlook.h tm

    work page 2024

  56. [56]

    (2024, April 22)

    In t egr a t ed Su stain ability and F in ancial R epor t 2024. (2024, April 22). C orpor a t e . for d. c om. h ttps:/ / c orpor a t e . for d. c om/ c on t en t / dam/ c orpor a t e / u s / en-u s / doc umen ts / r epor ts /2024-in t egr a t ed-su stain ability - and-fin ancial-r epor t.pdf

    work page 2024

  57. [57]

    (2025, Sep t ember 6)

    Mu stang Suv 2024. (2025, Sep t ember 6). C er t -t est -ne w .itlab .stan for d. edu. h ttps:/ / c er t -t est - ne w .itlab .stan for d. edu/ m u stang-suv -2024 R eport fr om Manu s: Figure 17: Error pattern: references not mentioned in the report. 39 R eport fr om Gemini-2.5 Flash: C ompr ehensiv e R esear ch R eport: Global ESG R eporting Str a t egi...

    work page 2024

  58. [58]

    My an aly sis is based on the * most r ec en t v ersions a v ailable as o f Sep t ember 2025* , c onsidering r epor t ed pr oposals and sta tu ses

    Multi-C olumn C omparison T able | Dimension | SEC Clim a t e Disclosur e R ule (as o f 2025, c urr en tly sta y ed)| EU C orpor a t e Su stain ability R epor ting Dir ec tiv e (C SRD) (as o f 2025)| California SB-253 & SB-261 (as o f 2025) I t' s impor tan t t o no t e th a t the r egula t or y landscape for ESG r epor ting is dyn amic, and some o f the ...

    work page 2025

  59. [59]

    Sleep and A cademic Ex c ellenc e: A Deeper L ook. (2024). * Stan for d L ong e vity * . [h ttps:/ / long e vity .stan for d. edu/lifestyle /2024/01/10/ sleep-and-academic -e x c ellenc e-a-deeper -look /](h ttps:/ / long e vity .stan for d. edu/lifestyle /2024/01/10/ sleep-and-academic -e x c ellenc e-a-deeper -look /)

    work page 2024

  60. [60]

    P olyph asic Sleep: P r os and C ons o f Shor t Sleep Schedule . (2018). *V er y w ell Health * . [h ttps:/ / www . v er y w ellhealth. c om/pr os-and-c ons-o f-a-polyph asic -sleep-schedule-4165843](h ttps:/ / www . v er y w ellhealth. c om/pr os-and-c ons-o f-a-polyph asic -sleep-schedule-4165843)

    work page 2018

  61. [61]

    Sleep h y giene: Simple pr ac tic es for be tt er r est. (2025). *Har v ar d Health P ublishing * . [h ttps:/ / www .health.h ar v ar d. edu/ sta ying-health y / sleep-h y giene-simple-pr ac tic es-for -be tt er -r est](h ttps:/ / www .health.h ar v ar d. edu/ sta ying-health y / sleep-h y giene-simple-pr ac tic es-for -be tt er -r est) ... R e fer enc es

    work page 2025

  62. [62]

    The impac t o f e x t ended sleep on da y time aler tness, vigilanc e and mood. (2025). *R esear chGa t e * . [h ttps:/ / www .r esear chga t e .ne t / publica tion/8370477_ The _impac t_ o f _ e x t ended_ sleep _ on_ da y time _ aler tness _ vigilanc e _ and_mood] (h ttps:/ / www .r esear chga t e .ne t / publica tion/8370477_ The _impac t_ o f _ e x t ...

    work page arXiv 2025

  63. [63]

    A[2]no ther ,

    Cir cadian Rh y thms | Na tion al Institut e o f Gener al Medical Scienc es. (2025). *NI GMS * . [h ttps:/ / www .nigms.nih.g o v / educa tion/ fac t -shee ts /P ag es / cir cadian-rh y thms](h ttps:/ / www .nigms.nih.g o v / educa tion/ fac t -shee ts /P ag es / cir cadian-rh y thms) ... Out -of -or der R ef er ences An aly sis: The r e fer enc e list is...

    work page 2025

  64. [64]

    ** Sa uc on y E ndorphin Elit e 2** * R unner ' s W orld R a ting Ca t eg or y: Best Ligh tw eigh t R ac er

    work page

  65. [65]

    **Br ooks Calder a 8** * R unner ' s W orld R a ting Ca t eg or y: Best Ov er all T r ail * C omposit e Sc or e: 0.53

    work page

  66. [66]

    LLM technology

    **L ululemon Wildfeel ** * R unner ' s W orld R a ting Ca t eg or y: Best Hybrid * C omposit e Sc or e: 0.33 ... * W eigh t: 0.0 o z * Dr op: 0.0 mm * W eigh t: 4.5 o z * Dr op: 0.0 mm * W eigh t: 1.0 o z * Dr op: 7.0 mm * C omposit e Sc or e: 1.11 Hallucinat ed inf or mation An aly sis: The r epor t fabrica t es da ta and pr oduc es a logically inc onsis...

    work page 2025