Rebuild CognitiveLiberties.com Around Intellectual Diversity, Not Secrecy

Current-state audit, mission and IA redesign, public charter, GEO/AEO/AG-UI specification and 365-day roadmap · 11 October 2026

Executive decision

Reposition; do not erase. CognitiveLiberties.com already provides substantial material on freedom of inquiry, source provenance, algorithmic discoverability and independent networks. However, its homepage, charter, network framing and existing 30/90/365-day plan still foreground private thought formation, surveillance resistance, confidentiality and defensive resilience. The new editorial hierarchy should make thinking, searching, sharing, intellectual disagreement, public discoverability and AI representation the defining public mission. Privacy remains a protected choice, no longer the lead organizing story. Existing historical work should be retained with versioned editorial notes rather than silently replaced.

Audit scope and limits: This is a desk audit of publicly visible pages accessed on 11 October 2026: homepage, About, Charter, Research Library, Network, Self-Audit, Evidence Ledger and Worldwide Action Plan [1-8]. It is not an authenticated CMS audit, full 48-guide content review, verification of all 74 underlying research reports, Search Console evaluation, crawl of every route, accessibility conformance test, server inspection, analytics study or stakeholder interview. A change in the live site after this date could alter findings. Quoted passages below are brief excerpts from the observed pages; counts are the site's own published descriptions, not independently audited archive totals.

1. Current-state site and editorial audit

The current homepage also already says that ranking and AI source selection influence which publishers become invisible; the project should amplify this material, not mischaracterize its existing work as wholly privacy-only [1]. The site says its research program has 74 exact reports, but that number is a publisher-reported scope. The proposed ten new reports should be listed with a new distinct series identifier rather than being inserted into historical counts without reconciliation.

2. Rewritten mission, vision and principles

Mission: CognitiveLiberties.com advances the freedom to think independently, explore lawful ideas without prejudgment, share knowledge voluntarily, encounter disagreement and remain meaningfully discoverable. We research how search engines, AI assistants and information institutions represent human viewpoints, and we develop practical, evidence-respecting ways to hold them accountable.

Vision: A world where no single company, government or information system determines the boundaries of humanity's discoverable ideas; where people can inspect original evidence, disagree honestly, contribute their perspectives and make informed choices about technology and disclosure.

Core principles: Curiosity is not guilt; participation is voluntary; publication and discovery matter; original sources deserve accessible links; unusual perspectives deserve faithful description but not automatic factual validation; source selection requires measurable scrutiny; concentration requires contestability; contributors retain clear attribution and rights; safety targets concrete harm; corrections are visible; governance is plural; individual privacy choices remain intact.

3. New information architecture

Introduce tags for claim status (established / contested / personal / speculative / historical) and information function (original research / primary record / contrary interpretation / synthesis). Every substantive article should have 'Go to originals', 'Best criticism', 'What remains unknown' and 'Update/correction' modules.

4. Redesigned homepage concept

Hero: 'Think freely. Search widely. Share knowledge. Keep ideas discoverable.' Supporting text: 'Cognitive liberty is the freedom to explore lawful questions, contribute original perspectives, meet disagreement and test what powerful information systems leave out.' Two prominent primary links: Explore viewpoints and Share original work. Privacy guidance appears as a clearly accessible secondary link, Choose how much to disclose.

First viewport: three equal paths: 'Discover an idea,' 'Follow the original evidence,' and 'Contribute a perspective.' Next: a live, human-reviewed 'Disagreement worth understanding' topic comparing original sources with standards of evidence. Next: an Accountability Lab feature showing indexed versus cited source paths, with unknowns explicitly labeled. Next: independent publisher directory and curated editorial outbound links. Footer: charter, corrections, transparency, licensing and site privacy.

Sample user journeys

Student: Search 'historic hypothesis' → see consensus summary → open rare original text → read strongest refutation → download citations. Independent publisher: use crawl diagnostic → correct a noindex mistake → request external review only if a documented exclusion remains. Contributor: select link-only/retrieval/training consent tier → preview attribution → publish → revise or add a correction. Critical reader: switch AI navigator from Evidence Review to Minority Positions → inspect exactly which sources were used → follow non-AI original-source path.

5. Public Cognitive Liberty and Thought Diversity Charter

We affirm (1) the freedom to hold and revise opinions; (2) the legitimacy of lawful curiosity without motive attribution; (3) voluntary speech and the option not to disclose; (4) clear paths to original sources and informed disagreement; (5) access distinct from endorsement; (6) minority perspectives faithfully represented but never made true by their rarity; (7) contributor authority over lawful sharing permissions; (8) independent, measurable challenges to significant discoverability failures; (9) transparent safety rules with correction opportunities; (10) plural access routes rather than a singular official viewpoint gate; (11) attribution and a public evidence/corrections record; and (12) dignity and fairness across languages and cultural traditions.

This charter is an editorial and design proposal, not an assertion of existing enforceable international search-ranking or AI-training rights. Preserve a link to the older charter and a dated 'What changed and why' page. Do not remove earlier private-inquiry work; archive it as part of the historical argument.

6. Source-first and link-first technical architecture

Canonical public page → source registry (URL, provenance, license, claim status) → editorial review → HTML link graph + sitemap/feeds → optional multi-source retrieval → AI answer with citations → audit of omissions/corrections → reader follows originals.

Stack options: extend the site's existing static-first/server-rendered approach; keep no-login browsing; maintain Article, ScholarlyArticle or CreativeWork structured data only where factually appropriate; preserve canonical URLs and semantic headings; provide Atom/RSS and downloadable JSON source lists; store citation records and correction status in Git-backed YAML or an audited lightweight database; use full text search over authorized public pages. Add a queue-based submission service with scoped permissions, malware checks, moderation and consent logs. Avoid turning existing nontracking design into a new requirement to monitor users.

Discoverability pilot: each research page includes source IDs, direct URLs, original quotes only within allowed limits, counterevidence, last reviewed date and explicit link-out. An 'Explore Alternatives' control queries a curated source register; it must not generate imaginary primary documents. Respect robots, licenses and access controls. A crawler index can expose that a source exists while reproducing text beyond permission would be a separate rights issue.

7. Ethical GEO/AEO implementation guide

Google's 2026 guidance says that foundational SEO and unique, people-first content remain central to generative Search experiences; special AEO/GEO text files or made-up AI markup are not required for Google visibility [9]. Therefore prioritize: real original analysis, useful citations, HTML accessibility, stable URLs, correct canonicals, relevant internal and external editorial links, crawlable text, accurate schema matching visible content, appropriate snippet eligibility, real author/source disclosures, and public correction dates. Use Search Console's documented tools for eligibility and, where available in the owner's account, its generative-AI performance features. Do not assert access to private Search Console analytics for this report.

Avoid mass AI-written pages for query variations, paid link swaps, fabricated 'expert' identities, author-schema claims without real authors, 'llms.txt guarantees,' synthetic community praise, automatic citation promises and concealed manipulative instructions. Measure search impressions, actual referral visits when allowed, independent-source clicks, citation errors and author satisfaction rather than promising rank positions. Respect permissions for third-party work and do not imply that indexing means model training.

8. AG-UI optional research navigator

AG-UI is an agent-to-user interaction protocol, not a search-ranking algorithm, a dataset distribution scheme or a training method. The September 30, 2026 AG-UI 1.0 documentation describes structured event schemas and support for states, tool calls and human-in-the-loop interactions [10]. It may be useful for an optional interactive research navigator, but it should not be a dependency for reading any public page.

Flow: source-card click → mode selection (Evidence Review / Competing Viewpoints / Original Documents / Minority Positions / History) → user-approved agent query → backend source retrieval and provenance checks → AG-UI event stream for progress, source cards and answer → explicit expand/correct controls. The server records only the minimum operational context, with session and retention policies disclosed. Display each original URL, status, relevant quotation or summary and why it was retrieved. Run an independent safety and accessibility review; show the same public sources with no JavaScript or AI session when feasible.

9. Viewpoint-representation experiment and editorial governance

Construct a preregistered 150-question bench across five subjects and at least five languages. For each create an externally reviewed source register: consensus sources, credible dissent, historical alternatives and unsupported claims. Record source inclusion/omission, primary source accuracy, false balance, citation reachability and appeal recoverability across two search engines and two AI assistants, as permitted. Re-run quarterly under documented locale/version settings. Invite independent reviewers to challenge the benchmark's categories. No performance results are claimed here.

Create an editorial council with rotating scientific, human-rights, library and independent-publisher expertise; publish conflict disclosures and an appeal path. Label factual correction, editorial disagreement, content rights issue, serious safety risk and technical fault as different intake categories. Contributors can choose link-only, retrieval or legally effective open/training use permissions, with separate informed consent; publication must not be advertised as guaranteeing training inclusion or total future erasure. If the site becomes a host of user content, obtain jurisdiction-appropriate legal advice and appropriate trust-and-safety capacity first.

10. Revision matrix

Versioning rule: do not delete old claims without a dated editorial note. A correction record should name the old text, the new position, the reason, reviewer and source. Use permanent redirects when moving routes; retain archive references and old charter links.

11. 30/90/180/365-day implementation plan

Total first-year planning envelope: $104,000-$230,000, excluding major litigation, custom model training, purchased datasets and full-time staff expansion. Add 15%-25% contingency if staffing, accessibility remediation or legal review is uncertain. A lower-cost volunteer/editorial pilot can run with fewer features; it should not claim the above milestones without corresponding resources. Allocate approximately 35%-45% to engineering/design, 25%-30% to editorial/translation, 10%-15% to legal/safety, 10%-15% to independent audit, and the remainder to hosting, operations and contingency. Use contractor quotes and explicit statements of work before budget approval; figures are estimates rather than observed costs.

Program KPIs: unique original sources linked and verified; source diversity by language and organization; rate of clicks to external primary sources (aggregate/consented); proportion of articles with meaningful counterarguments; citation validity; documented corrections and turnaround time; verified contributor retention; audit benchmark representation/false balance; accessible performance; and proportion of accepted partner references unrelated to reciprocal SEO arrangements. Track failures and reviewer disagreement alongside successes.

12. Risks, dependencies and decision gates

Editorial drift: a movement for openness may turn into an ideological directory. Mitigation: cross-viewpoint panels, written status labels, audits including unfriendly perspectives. False balance: a broad inclusion policy can spread inaccurate health/science claims. Mitigation: consensus status and clearly scoped historical/fringe lanes. Rights conflicts: open access does not settle training rights. Mitigation: counsel-reviewed permissions, record-level license metadata. Trust and safety: open submission attracts spam, impersonation and harassment. Mitigation: staged moderation, reporting, appeals and anti-flood controls. Privacy: new interactions can become profiling infrastructure. Mitigation: optional anonymous reading, short retention, no behavioral adtech. Unproven ROI: search/AI citation behavior cannot be promised; track actual reader benefit.

Recommended launch sequence

Approve mission and public charter; ship a source-first homepage; then add moderated contribution and measurement before live AI features. Do not deploy AG-UI until original-source registry, permissions, evaluation and safety review are functioning. The core change is editorial and civic, not merely technological: readers should encounter a site that helps them learn, disagree, contribute and follow ideas outward.

Sources and dated verification

1. CognitiveLiberties.com, homepage (accessed 11 Oct. 2026). https://cognitiveliberties.com/
2. CognitiveLiberties.com, About (accessed 11 Oct. 2026). https://cognitiveliberties.com/about
3. CognitiveLiberties.com, Charter (accessed 11 Oct. 2026). https://cognitiveliberties.com/charter
4. CognitiveLiberties.com, Network (accessed 11 Oct. 2026). https://cognitiveliberties.com/cognitive-liberty-network
5. CognitiveLiberties.com, Research Library. https://cognitiveliberties.com/research
6. CognitiveLiberties.com, First-Party Self-Audit. https://cognitiveliberties.com/cognitive-liberty-self-audit
7. CognitiveLiberties.com, Worldwide Action Plan. https://cognitiveliberties.com/cognitive-liberty-action-plan
8. CognitiveLiberties.com, Evidence Ledger. https://cognitiveliberties.com/research-evidence
9. Google Search Central, Optimizing your website for generative AI features on Google Search (2026). https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
10. AG-UI 1.0 announcement (Sep. 30, 2026) and protocol specification. https://www.copilotkit.ai/blog/ag-ui-1.0 ; https://github.com/ag-ui-protocol/ag-ui/blob/main/docs/introduction.mdx
11. W3C, Webmention and ActivityPub Recommendations. https://www.w3.org/TR/webmention/ ; https://www.w3.org/TR/activitypub/
12. UNESCO, Recommendation on the Ethics of Artificial Intelligence. https://www.unesco.org/en/legal-affairs/recommendation-ethics-artificial-intelligence