The Human Contribution Movement: Building the Future Record of Humanity

Full report

On this page

Executive Briefing for Organizational Leaders

The digital record of humanity is undergoing a profound and accelerating transition. As global communication networks expand, the artifacts of human culture, public debate, and original knowledge are increasingly digitized. This vast repository serves dual functions: it constitutes the broader open knowledge ecosystem utilized by current generations for research and memory, and it provides the raw material for artificial intelligence (AI) training datasets that will shape future computational reasoning and cultural understanding. However, this record is highly vulnerable. Deep structural analysis indicates a converging crisis characterized by algorithmic data filtering, systemic link rot, and the psychological barriers of context collapse, all of which threaten to homogenize the historical record and systematically exclude the diverse, everyday voices of ordinary individuals. This comprehensive report details the strategic and technical requirements for the "Human Contribution Movement," operating under the core principle of cognitive liberty—the right to hear and be heard. The movement’s proposed digital hub, https\://buildtoresist.com, is envisioned as a decentralized network designed to encourage individuals to transition from passive digital consumption to active, durable knowledge creation. The analysis distinguishes carefully between the open web (the broader knowledge ecosystem) and specific AI training corpora. While independent publishing ensures public availability, no single entity can guarantee inclusion in closed AI training sets. The findings herein provide a strategic blueprint for civic organizations, independent publishers, and digital architects. The report delivers an evidence-based framework to overcome participation inequality, establish durable web infrastructure, and deploy metadata standards that ensure human original thought remains accessible, machine-readable, and preserved for future generations.

Part I: The Architecture of the Human Knowledge Record

The Broader Knowledge Ecosystem vs. AI Training Datasets

To accurately assess the future record of humanity, it is necessary to separate the broader knowledge ecosystem from AI training datasets. The broader ecosystem encompasses all publicly accessible digital information: institutional archives, independent websites, message boards, civic records, local history repositories, and social media platforms. This ecosystem is decentralized and heavily reliant on search indices, library databases, and hyperlinks for navigation and discovery. AI training datasets, conversely, are highly curated, centralized subsets of the broader ecosystem. Organizations developing Large Language Models (LLMs) rely on massive web crawls, such as Common Crawl, to gather raw text. However, a web crawler operating with a finite computational budget cannot index the entire internet. Crawlers typically prioritize URLs based on graph-centrality metrics, most notably harmonic centrality, meaning they disproportionately fetch pages from highly linked, prolific, and established hosts1. The structural implication of this approach is that isolated independent websites, niche community forums, and newly established personal blogs often remain uncrawled. Consequently, tens of millions of high-quality URLs are entirely excluded from the foundational stage of AI dataset creation simply because they lack the necessary inbound links to trigger crawler prioritization1.

Data Filtering, Corpus Selection, and Representation Bias

Even when digital human contributions are successfully crawled, they face aggressive data curation pipelines. Datasets like FineWeb, FineWeb-Edu, and Nemotron-CC discard the vast majority of scraped web data—often eliminating between 70% and 90% of raw content to improve downstream LLM performance and reduce computational overhead3. This filtering process introduces profound and systemic representation bias. The curation pipelines rely on a multi-stage process utilizing heuristic filters, fastText classifiers, and model-based quality scoring. Heuristic filters automatically discard documents that fall outside standardized parameters, such as abnormal punctuation-to-text ratios, short document lengths, or the presence of specific keywords indicative of boilerplate or spam6. While efficient, these filters lack semantic understanding and frequently discard stylized, creative, or unconventional writing. Following heuristic filtering, pipelines frequently employ perplexity filtering. By utilizing a reference language model (often trained on Wikipedia or curated news), the system evaluates how "surprising" a sequence of text is. Text with high perplexity—which deviates from the formal, standardized prose of the reference model—is discarded as low quality7. The second-order effect of perplexity filtering is the systemic erasure of colloquial speech, regional dialects, and informal technical troubleshooting. If a community forum utilizes specialized jargon or vernacular that the reference model has not encountered, the perplexity score will spike, and the human contribution will be purged from the training corpus8. Furthermore, advanced datasets utilize LLM-as-a-judge classifiers to assign quality or educational scores. For example, FineWeb-Edu relies on an automated classifier to identify "highly educational" content, retaining only the data that scores at or above a specific threshold2. While this method successfully produces models with high reasoning capabilities on standardized benchmarks, it systematically enforces a specific, often elite or institutional, definition of quality10. The risk is severe: future information systems will likely reflect the experiences of the most prolific, standardized, or well-funded publishers, while filtering out documented everyday human experiences1.

Genuine human participation requires a recognition that these automated filtering mechanisms, while technically necessary for AI training efficiency, inherently skew cultural memory. When quality scores are determined by mathematical predictability, the unique, the novel, and the deeply personal are the first to be discarded11.

Part II: The Threat of Digital Decay and Link Rot

The broader knowledge ecosystem is suffering from systemic infrastructural decay. The internet operates on an assumption of permanence that is technically unfounded. The phenomenon of "link rot"—where hyperlinks point to missing, deleted, or fundamentally altered content—erodes the foundation of shared human knowledge. If the digital record is a tapestry, link rot is the unraveling of its threads. Recent analysis reveals the staggering scale of this issue. Extensive studies demonstrate that digital decay touches everything from casual social media interactions to the highest levels of legal jurisprudence. The loss of working paths to information means that future generations will inherit a fractured, disconnected web.

The implications of these statistics are profound. When 38% of webpages from a decade ago disappear12, the AI models of the future will be training on a continually shrinking, temporally biased window of human history. Furthermore, the decay of academic and legal citations undermines the verifiability of human knowledge13. When URLs fail, the citation graph that connects concepts dissolves, making it impossible for automated systems—and human researchers—to trace the provenance of an idea16. Independent publishing efforts must confront the reality of digital decay. Content hosted on centralized social platforms is highly ephemeral, averaging a survival rate of barely 31% over five years, whereas decentralized, independently hosted academic or personal websites exhibit significantly higher persistence17. Therefore, the act of contributing to humanity's record must include proactive preservation, utilizing web caching services like the Internet Archive or Perma.cc, developed by the Harvard Library Innovation Lab, to ensure that citations remain stable14.

Part III: The Philosophy of Cognitive Liberty and Public Participation

Cognitive Liberty: The Right to Hear and Be Heard

The Human Contribution Movement roots its mission in the concept of "cognitive liberty." Traditionally defined as the right to mental self-determination, mental privacy, and freedom of thought, cognitive liberty represents a foundational human right in the digital and neurotechnological age23. Scholars such as Nita Farahany advocate for an international human rights framework that updates existing protections to secure freedom of thought against corporate and algorithmic commodification26. Within the context of digital publishing and AI, cognitive liberty expands to encompass the right of individuals to control their intellectual outputs and the right of the public to access a diverse, unmediated spectrum of human thought24. If AI systems are trained exclusively on data curated by corporate pipelines designed to prioritize safety, compliance, and standardized educational value, the spontaneous evolution of human emotion, regional culture, and controversial debate is sterilized11. To safeguard cognitive liberty, the digital ecosystem must not be monopolized by centralized algorithms that dictate visibility. The movement's motto, Don't shadowban yourself, addresses the preemptive self-silencing that occurs when individuals believe their contributions are insignificant, or when they rely solely on centralized platforms that algorithmically suppress external links and nuanced discussions.

Statement of Movement Principles

To guide the Human Contribution Movement, the following public-facing statement of principles must be adopted by participating organizations and independent nodes:

1. The Principle of Intellectual Sovereignty: Every individual possesses the inherent right to document, publish, and control their original knowledge, experiences, and cultural memory without reliance on centralized, ephemeral social media platforms.

2. The Principle of Distributed Memory: Humanity’s record is most resilient when decentralized. Independent websites, local archives, and self-hosted communities form a robust bulwark against digital decay and algorithmic erasure.

3. The Principle of Transparent Architecture: Tools used for data selection, AI training, and search retrieval must be understood by the public. Creators have a right to know how their data is classified and the freedom to format their work to ensure accurate machine-led retrieval.

4. The Principle of Constructive Friction: A healthy knowledge commons requires the preservation of constructive disagreement, diverse dialects, and marginalized perspectives that automated quality filters often misclassify as statistical "noise."

5. The Principle of Durable Attribution: The provenance of knowledge must be preserved. Links should be maintained, citations protected, and attribution honored across generations.

Overcoming Obstacles: Context Collapse and Participation Inequality

Meaningful public participation faces significant psychological and social obstacles. The primary psychological barrier is "context collapse," a phenomenon where disparate audiences (e.g., family, professional colleagues, strangers) are conjoined into a single online environment30. On centralized social platforms, the inability to compartmentalize audiences leads to profound uncertainty. The behavioral consequence of context collapse is self-censorship—the deliberate modification, limitation, or total avoidance of expressing genuine opinions32. Legal and algorithmic frameworks on centralized platforms are structurally ill-equipped to handle expression under context collapse, leading to over-criminalization and the suppression of vulnerable voices who lack the social power to navigate hostile digital environments34. Simultaneously, the digital landscape has historically been defined by participation inequality. This is most famously summarized by the "90-9-1 rule," introduced by Jakob Nielsen, which posits that 90% of online community members are passive lurkers, 9% contribute occasionally, and only 1% create original content35. While modern community management strategies, gamification, and shifting digital norms have challenged this strict ratio—with some studies suggesting up to 76% of users now actively participate in some form—the fundamental divide between passive consumption and durable knowledge creation remains a primary hurdle35. Liking a post or sharing a transient video does not contribute to the enduring human record; it merely feeds engagement algorithms. Transitioning hesitant contributors requires mitigating context collapse by encouraging independent, domain-specific publishing—where the creator controls the audience context—and breaking down the technical barriers to entry. By providing users with sovereign digital spaces, the fear of context collapse is minimized, directly reducing self-censorship32.

Part IV: Civic Interventions and Community Memory

To support voluntary knowledge contribution, independent publishers must look to the methodologies of civic organizations and public libraries. Libraries serve as foundational nodes of localized community memory, actively resisting the homogenization of centralized data. They capture the micro-histories that global web crawlers, obsessed with harmonic centrality, systemically ignore. A prominent example of this essential archival function is the Cicero Public Library in Illinois. The library maintains an exhaustive local history archive detailing the industrial history of the Hawthorne Works (a massive telecommunications factory that shaped the local economy) and the Eastland Disaster of 1915, a tragedy that disproportionately impacted Cicero residents39. Furthermore, the library digitizes and preserves extensive genealogical records, including local newspapers like the Suburban Leader (dating to 1927), school yearbooks, and city directories39. By digitizing, curating, and hosting these materials independently, local libraries ensure that hyper-local cultural histories are preserved. Without the Cicero Public Library's dedicated history room and reference desk, the granular details of the Hawthorne Works' unionization or the lived experiences of early 20th-century Czech and Latino immigrants in the area would be lost to digital decay39. The strategic imperative for the Human Contribution Movement is clear: civic organizations, historical societies, and independent community forums must adopt the archival mindset of the public library. They must transition from relying on temporary Facebook groups—which suffer from massive link rot and algorithmic suppression—to stable, searchable, and independently hosted repositories40.

Part V: Semantic Grounding and Machine Readability

For independent human contributions to meaningfully influence public information resources, knowledge graphs, and AI models, the data must be machine-readable. Human-readable prose is often ambiguous to algorithmic crawlers. The technical solution lies in structured data, specifically Schema.org metadata deployed via JSON-LD (JavaScript Object Notation for Linked Data)43. Schema.org, launched in 2011 as a joint initiative by major search engines, is a shared vocabulary that allows independent publishers to explicitly classify content, identify authors, clarify relationships, and anchor facts43. Historically, search engine optimization (SEO) utilized schema merely to generate rich snippets in search results. However, with the advent of LLMs and retrieval-augmented generation (RAG) models, JSON-LD has evolved into a semantic "truth file"46. When AI systems search the web for answers, they do not read pages like humans; they recall structured data46. They rely on JSON-LD to verify facts and establish "ground truth" across three distinct levels of visibility:

1. Discovery: Structured, fact-rich content increases visibility in AI-generated responses by clearly classifying page content43.

2. Understanding: Providing explicit entity relationships (e.g., tying an Article to a Person and an Organization) allows AI models to unambiguously connect local facts to their broader world knowledge43.

3. Grounding: When an LLM needs to state a date, a location, or an author, structured data provides a verifiable anchor, reducing hallucinations and dramatically increasing the likelihood of accurate citation43.

By embedding schemas such as FAQPage, HowTo, DefinedTerm, and Organization, independent websites provide exact, machine-readable signals45. This semantic architecture allows LLMs to extract citable claims confidently. If the Human Contribution Movement is to succeed, it must transition digital creators from simply publishing text to acting as semantic data architects, ensuring their original knowledge is inextricably woven into the future machine-readable knowledge graph46.

Part VI: The BuildToResist Strategic Framework

The https\://buildtoresist.com initiative serves as the philosophical and instructional vanguard for the movement. Although the current live domain is presently inaccessible48, the proposed architecture requires a decentralized network of independent publishing sites that share relevant links and contribute to a public knowledge commons without centralizing ownership.

A Distributed Network of Independent Publishing (Proposed Ecosystem)

The BuildToResist architecture is modeled not as a centralized repository, but as a federated index and educational hub. Participants can share the https\://buildtoresist.com URL through independently controlled websites, newsletters, reading lists, community introductions, QR codes, events, and direct personal recommendations. The ecosystem relies on the following pillars:

1. Self-Sovereign Nodes: Participants purchase and control their own domain names and hosting. BuildToResist provides open-source templates (e.g., static site generators) pre-configured with semantic HTML and comprehensive Schema.org JSON-LD capabilities44.

2. The Direct-Linking Protocol: To combat the algorithmic suppression of external links on major platforms, the movement relies on direct, independent linking. Sites within the ecosystem utilize protocols to notify each other when they link to one another, rebuilding a decentralized citation graph that artificially raises harmonic centrality1.

3. The Web Archiving Standard: Every outgoing link published on a BuildToResist-compliant node must be automatically or manually archived using services like the Internet Archive or Perma.cc14. This ensures that the context of a conversation remains intact even if the target URL decays22.

4. The BuildToResist Directory: The central hub functions purely as a public directory and educational portal, routing traffic outward to the sovereign nodes rather than hoarding attention.

Movement-Wide Action Plan

To operationalize the movement, the strategic framework encompasses ten critical domains:

1. Public education about the human knowledge record and AI training: Demystify the data curation pipelines. Teach the public that posting on a social media app does not mean contributing to the enduring human record. Clarify the distinction between the open web and AI corpora3.

2. Accessible entry points for first-time contributors: Provide low-friction, non-technical methods for individuals to publish their first piece of original knowledge, such as contributing a local history fact to a library archive or publishing a single, well-documented technical tutorial.

3. Community discussion and constructive disagreement: Promote protocols for independent forums where moderation is localized and contextual, avoiding the context collapse of massive platforms30.

4. Independent website publishing: Distribute step-by-step toolkits for setting up personal domains, emphasizing that owning the server space is the only defense against corporate data purges17.

5. Original research and educational content: Encourage the documentation of everyday experiences, niche hobbies, and cultural histories. This data is chronically underrepresented in high-perplexity AI datasets8.

6. Digital preservation and stable attribution: Mandate the use of archival links (e.g., Perma.cc) and persistent identifiers to combat the 38% decay rate of modern web pages12.

7. Participation by smaller and underrepresented communities: Launch targeted outreach to civic groups, marginalized linguistic communities, and local historians to ensure their dialects and histories are digitized on their own terms40.

8. Direct linking between independent websites: Rebuild the "blogroll" and decentralized citation networks to organically raise the harmonic centrality of independent sites, forcing web crawlers to index them1.

9. Transparent moderation, contributor choice, and appeals: In decentralized forums, establish clear, community-specific moderation guidelines that protect cognitive liberty while preventing harassment.

10. Sustainable outreach and partnerships: Partner with institutions like the Cicero Public Library and local universities to integrate independent publishing modules into digital literacy curricula39.

The Contribution Ladder

To transition users from silent readers to durable knowledge creators, the movement utilizes a structured Contribution Ladder, addressing specific psychological and technical barriers at each rung.

Part VII: Roadmaps and Implementation

90-Day Initial Implementation Roadmap

12-Month Community Development Roadmap

Part VIII: Measurement and Evidence

Evidence Matrix

To guide the movement objectively, strategies must be rooted in established facts while remaining adaptable to current technological uncertainties and future hypotheses.

Measurement Plan

To ensure the movement tracks genuine knowledge creation rather than vanity metrics (e.g., social media impressions or unsupported claims about AI training inclusion), the following measurement plan focuses on durability and structural network health.

Part IX: Participation Toolkit and Call to Action

The transition from passive consumption to active creation requires a highly practical, accessible toolkit. The downloadable BuildToResist Participation Toolkit includes the following operational assets:

1. The Domain Sovereign Guide: A non-technical manual for purchasing a domain name, setting up secure hosting, and utilizing static site generators that are completely immune to sudden platform policy changes or algorithmic shadowbanning.

2. The Metadata Blueprint: Copy-and-paste templates for Schema.org JSON-LD43. This ensures that when a user publishes a local history article or a technical tutorial, AI models and search engines can properly classify the author, the organization, and the factual claims without ambiguity45.

3. The Link Rot Remediation Checklist: Instructions on auditing old personal posts, utilizing Perma.cc for all external citations to ensure legal and scholarly durability14, and setting up 301 redirects to maintain link equity16.

4. Community Connection Assets: Pre-designed graphics, email templates, and QR codes linking to https\://buildtoresist.com. Participants are encouraged to place these QR codes in physical community spaces—such as independent bookstores, community boards, and local libraries like Cicero Public Library—to bridge the physical and digital knowledge gaps39.

Concise Public Call to Action

Your thoughts, your local history, and your technical expertise belong to the enduring record of humanity, not to the ephemeral feeds of centralized platforms. Break the cycle of digital decay. Overcome the fear of context collapse. Do not shadowban yourself. Reclaim your cognitive liberty by building your own space on the web. Document your truth, link to others directly, archive your citations, and join the network at BuildToResist.com. Speak so the future can hear you.

Conclusion

Humanity’s future knowledge base should contain more voices, more lived experiences, more original thoughts, and more opportunities to disagree and learn. Achieving this requires a fundamental shift in how ordinary individuals interact with the digital ecosystem. By moving away from ephemeral, centralized platforms and embracing independent, semantically structured publishing, contributors can fortify the digital record against both the ravages of link rot and the homogenization enforced by algorithmic filters. It is critical to maintain strict transparency regarding the outcomes of these actions: publishing original work independently drastically increases its public availability and safeguards cultural memory, but no individual or movement can guarantee that any specific piece of writing will be ingested into a proprietary AI training dataset. Corpus selection remains entirely at the discretion of crawler budgets, heuristic quality thresholds, and opaque perplexity metrics1. However, by actively raising the harmonic centrality of independent networks through robust direct linking1, combating the 38% digital decay rate through archival tools12, and utilizing Schema.org JSON-LD to make human knowledge unequivocally legible to machines45, the Human Contribution Movement maximizes the probability that diverse human experiences survive. The act of building to resist is not merely about influencing artificial intelligence; it is about securing the cognitive liberty of future generations, ensuring they inherit a historical record that is vibrant, decentralized, and fundamentally human.

Works cited

1. Host-Aware Web Crawling for LLM Pretraining Data \- arXiv, https\://arxiv.org/html/2610.13321v1

2. Web Crawler Restrictions, AI Training Datasets & Political Biases, https\://arxiv.org/html/2510.09031v1

3. The FineWeb Datasets: Decanting the Web for the Finest Text Data, https\://arxiv.org/html/2406.17557v1

4. Building Nemotron-CC, A High-Quality Trillion Token Dataset for, https\://developer.nvidia.com/blog/building-nemotron-cc-a-high-quality-trillion-token-dataset-for-llm-pretraining-from-common-crawl-using-nvidia-nemo-curator/

5. Enhancing Multilingual LLM Pretraining with Model-Based Data, https\://aclanthology.org/2025.swisstext-1.4.pdf

6. Enhancing Multilingual LLM Pretraining with Model-Based Data, https\://openreview.net/pdf?id=ufhH8YXxOB

7. Quality Filtering: Heuristics, Perplexity, and Classifiers \- Interactive, https\://mbrenndoerfer.com/writing/quality-filtering-heuristic-perplexity-classifier-thresholds

8. Pre-training Data: What LLMs Learn From \- Learnixo, https\://learnixo.io/blog/llm-pretraining-data

9. Seed-Coder: Let the Code Model Curate Data for Itself \- arXiv, https\://arxiv.org/html/2506.03524v1

10. TBDFiltering: Sample-Efficient Tree-Based Data Filtering \- arXiv, https\://arxiv.org/pdf/2601.22016

11. High-quality Data Do not Mean Safe\! Poisoning LLMs after ... \- arXiv, https\://arxiv.org/html/2610.01367v1

12. When Online Content Disappears \- Pew Research Center, https\://www\.pewresearch.org/data-labs/2024/05/17/when-online-content-disappears/

13. So how bad a problem is link-rot? – Perma.cc Blog, https\://archive.blogs.harvard.edu/perma/2019/02/14/so-how-bad-a-problem-is-link-rot/

14. Perma: Scoping and Addressing the Problem of Link and Reference, https\://harvardlawreview.org/forum/vol-127/perma-scoping-and-addressing-the-problem-of-link-and-reference-rot-in-legal-citations/

15. The Sweet Morbidity of Link Rot \- Slaw.ca, https\://www\.slaw.ca/2017/02/01/84741/

16. What Is Link Rot and How Can You Stop It From Hurting Your Brand?, https\://bitly.com/blog/what-is-link-rot/

17. The Internet Is Breaking, and Nobody's Talking About It \- Medium, https\://medium.com/@shivashanker7337/the-internet-is-breaking-and-nobodys-talking-about-it-7894f50f1f3f

18. How long does a web page live: what the research says about link rot, https\://archivarix.com/en/blog/link-rot/

19. Content Decay Statistics 2026: Traffic Loss & Half-Life \- Milan Novotný, https\://www\.milan-novotny.com/statistics/content-decay-statistics/

20. Link Rot: 38% Of Webpages From 2013 Have Vanished, https\://www\.searchenginejournal.com/38-of-webpages-from-2013-have-vanished-pew-study-finds/516834/

21. Perma.cc \- Wikipedia, https\://en.wikipedia.org/wiki/Perma.cc

22. Scaling Up Perma.cc: Ensuring the Integrity of the Digital Scholarly, https\://www\.dlib.org/dlib/may17/dulin/05dulin.html

23. What Is Cognitive Liberty? Your Right to Mental Privacy \- Neurosity, https\://neurosity.co/guides/cognitive-liberty-mental-privacy

24. Cognitive liberty and the psychedelic humanities \- Frontiers, https\://www\.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2023.1128996/full

25. Towards new human rights in the age of neuroscience and ... \- PMC, https\://pmc.ncbi.nlm.nih.gov/articles/PMC5447561/

26. The Battle for Your Brain \- VIB Blog, https\://blog.vib.be/the-battle-for-your-brain

27. Technology and Security (TS) podcast | Neurotechnology, cognitive, https\://www\.ussc.edu.au/podcasts/technology-and-security-ts-podcast/neurotechnology-cognitive-liberty-and-information-warfare-with-professor-nita-farahany

28. The Battle for Cognitive Liberty in the Age of Corporate AI, https\://www\.techpolicy.press/the-battle-for-cognitive-liberty-in-the-age-of-corporate-ai/

29. Dual Track Strategies for Technologically Augmented Humans, https\://www\.mechecology.org/archive/view\_article?pid=mech-2-2-13

30. One Size Fits All: Context Collapse, Self-Presentation Strategies and, https\://academic.oup.com/jcmc/article/23/3/127/4962540

31. Context collapse and privacy management: Diversity in Facebook, https\://www\.researchgate.net/publication/318027589\_Context\_collapse\_and\_privacy\_management\_Diversity\_in\_Facebook\_friends\_increases\_online\_news\_reading\_and\_sharing

32. View of The Relationship Between Context Collapse and Audience, https\://north-press.com/index.php/jcmpl/article/view/192/194

33. The Relationship Between Context Collapse and Audience Risk, https\://north-press.com/index.php/jcmpl/article/view/192

34. Legal chilling effects in the algorithmic public sphere \- Frontiers, https\://www\.frontiersin.org/journals/human-dynamics/articles/10.3389/fhumd.2026.1831523/full

35. The 90-9-1 Rule is Over. It's Time for Higher Engagement\!, https\://www\.grazitti.com/blog/the-90-9-1-rule-is-over-its-time-for-higher-engagement/

36. Understanding the 1% Rule Online | PDF | Communication \- Scribd, https\://www\.scribd.com/document/637813507/1-Percent-Rule

37. 90-9-1 Rule (Nielsen, 2006). \- ResearchGate, https\://www\.researchgate.net/figure/1-Rule-Nielsen-2006\_fig2\_337267860

38. How We Know the 90-9-1 Rule for Online Community Engagement, https\://www\.higherlogic.com/blog/90-9-1-rule-online-community-engagement-data/

39. Explore Cicero History, https\://cicerolibrary.org/explore-cicero-history/

40. A Library, A Historical Society, and A Czech School, https\://www\.ciceroindependiente.com/english/cicero-berwyn-news-history-culture-czech-latino-immigrants-hawthorne-library-western-electric-company-museum-architecture

41. Genealogy and House History Resources \- Cicero Public Library, https\://cicerolibrary.org/explore-cicero-history/genealogy-and-house-history/

42. Cicero Public Library through the years. \- Facebook, https\://www\.facebook.com/groups/392177154465434/posts/1072194646463678/

43. What is Schema.org and structured data? \- enhancely.ai, https\://www\.enhancely.ai/what-is-schema-information-and-structured-data

44. AI Schema.org JSON-LD \- Drupal.org, https\://www\.drupal.org/project/ai\_schemadotorg\_jsonld

45. Schema.org & JSON-LD for LLM Citations (2026) \- Alice Labs, https\://alicelabs.ai/en/insights/schema-org-for-ai

46. Schema & JSON-LD for LLM Search: Structured Data for LLMO, https\://www\.szymonslowik.com/json-ld-for-llm-seo/

47. Top JSON-LD Schema for SEO Patterns Driving AI Search Visibility, https\://growthnatives.com/blogs/seo/top-json-ld-schema-patterns-for-ai-search-success/

48. https\://buildtoresist.com

49. Pausing the Internet \- Harvard Law School Center on the Legal, https\://clp.law.harvard.edu/article/pausing-the-internet/

50. Link rot \- Wikipedia, https\://en.wikipedia.org/wiki/Link\_rot

51. Schema Markup for AI Crawlers: The 2026 Practical Guide, https\://rankingstudios.com/schema-markup-for-ai-crawlers-practical-guide/