Additional October 2026 research: separate provenance
A new 30-edition research collection covers freedom to search without judgment, public knowledge sharing, viewpoint representation, pluralistic AI, search accountability, open hyperlinks, and market concentration. It includes its own methodology, source register and corrections. It is separate from the original 74-report custody record and legacy 194-unit evidence ledger. Publishing an argument does not establish its factual truth or guaranteed model inclusion.
Direct answer
The site preserves each of 74 unique supplied reports and the received 32-member source package as an exact durable source, records its hash and public-page mapping, and publishes a separate editorial synthesis. Every section in the established 48-guide public library maps to a stable evidence unit; the 16-report AI governance series and 10-report ecosystem expansion are preserved and catalogued separately until claim-level review that records source scope, verification state, limitations, assertion inventory, and next-review date. Current legal, provider, and technical claims remain time-sensitive.
Source layers
| Layer | Purpose | Authority |
|---|---|---|
| Supplied report | Preserve the full research record, citations, framing, and unresolved problems | Source evidence supplied for this project; preserved byte-for-byte |
| Public source record | Identify the law, case, standard, research, provider guidance, methodology, or analysis mapped to a section | Varies by source class and retrieval state |
| Evidence unit | Bind one public section to stable claim ID, source scope, limits, dates, and assertion inventory | Auditable editorial record, not a guarantee that every assertion is separately proved |
| Editorial synthesis | Create a readable answer-first guide without reproducing the report wholesale | Site analysis; not a substitute for the source |
| Normative doctrine | State the site’s liberty-first principles and proposals | Advocacy position, not settled law |
Evidence-unit granularity
The ledger uses one evidence unit per public research section. Each unit inventories the sentences, list items, table rows, definitions, and callouts rendered inside that section. Its source IDs support the section’s overall evidentiary foundation. They are not silently represented as sentence-by-sentence proof.
Verification states
- Verified primary record
- A primary legal, official, or oversight source directly supports the central factual foundation.
- Verified original research
- Original research directly supports the empirical or technical foundation within the study’s scope.
- Verified official guidance
- An official standard or first-party guide supports a published architecture or rule, not every production behavior.
- Corroborated secondary evidence
- Relevant secondary sources support the description, but a single primary record does not establish the whole proposition.
- Evidence + editorial analysis
- Sources support important premises; comparisons, risk judgments, or conclusions remain site analysis.
- Report-derived argument
- The supplied report is the principal basis for the framing, with external sources providing context rather than complete proof.
- Site policy proposal
- A recommended law, standard, test, or architecture—not current law or settled consensus.
- Scenario analysis
- A bounded future used to test consequences, not a prediction.
- Verification limited
- The source identity is known, but direct retrieval or review was incomplete during this release.
Source authority and retrieval
A primary legal text outranks a blog describing the law. Original research outranks commentary describing a study. An official technical standard can establish the specified protocol. A provider document can establish what the provider publicly says, but it remains first-party guidance until production behavior is independently tested.
Retrieval state is recorded separately from source class. Sources reviewed online on 2026-09-02 are distinguished from locators that were blocked or unavailable and sources carried forward without an independent recheck. A blocked page is not quietly counted as fully reverified.
Release 1.12.0 preserves 31 stable REV-* review-event identifiers, five REC-* wording reconciliations, bounded scope statements, explicit review limits, conflict notes, and a public research update ledger. A review event confirms only the named support scope; it is not a full-document certification. Implementation kits link to this evidence while separately labeling worldwide principles, local legal questions, technical recommendations, operational practices, and site policy proposals.
Worldwide comparison without flattening context
The public hierarchy is organized by mechanisms and safeguards rather than nations. A report centered on one jurisdiction may reveal a reusable architecture, but its laws, statistics, institutions, and social conditions remain bounded to that evidence. Country examples are presented as case studies with global lessons, never as rankings, claims about a population, or proof that every society follows the same sequence.
Global synthesis is justified only where reports or external evidence support a recurring pattern. Otherwise, the site states a hypothesis, proposal, or comparative question rather than silently converting a local case into a worldwide fact.
Freshness without hidden network dependence
A Python-standard-library tool reads committed review dates and retrieval states, verifies the exact received source package, and writes deterministic current, due-soon, overdue, blocked, and unrechecked queues. Ordinary repository validation makes no network request. The queue says what must be reviewed; it does not decide whether a claim is true.
All 84 source IDs and 194 evidence-unit IDs must reconcile exactly across every run. Wording changes belong in a separate reconciliation ledger so exact source artifacts remain immutable.
Governed multilingual derivatives
A translated page is a derivative, not a new authority. Every published translation must name the controlling route and source revision, disclose human and model involvement, list transformations and omissions, record a review state, and bind the exact derivative bytes to SHA-256. Hreflang is published only for an actual page pair.
The first Spanish pilot is editorially reviewed against the English worldwide page but still carries an explicit independent native-speaker-review gap. Readers are directed to the controlling English evidence record whenever nuance matters.
How conflicting material is handled
Current implementation and direct evidence outrank stale narrative memory. The reports sometimes overlap, duplicate passages, use different terminology, or reach different levels of confidence. Those differences are not silently erased. The public synthesis selects supported common ground, records important limits, and preserves the exact reports for audit.
When direct sources materially narrow a report claim, the public synthesis must be corrected while the source report remains unchanged. The reconciliation belongs in the evidence ledger and durable implementation history, not in a covert rewrite of the original material.
A liberty-first argument must survive scrutiny.
Preserving cognitive liberty is the purpose of this site, not one side of a neutral scorecard. Start with the freedom being defended. When a restriction is proposed, identify the specific harm, the evidence, the affected activity and why a narrower response would not work. Answer a restriction argument where it appears; do not mistake repeating it for resolving it.
Separate the legal command, enforcement or informal pressure, provider choice, technical change and effect on inquiry. Mark each link as documented, reported, inferred or hypothetical. A scenario is not an observed event, and an enacted rule is not proof that its breadth is justified. Equally, a liberty argument cannot turn an unfavorable holding into a favorable one. Use a narrower supported rebuttal when the evidence does not disprove the underlying risk.
No official monopoly over contested understanding
Authority is not a substitute for evidence. Protect the freedom to question official accounts, revise conclusions and examine competing explanations without making access to knowledge conditional on political conformity.
Objection. Official correction powers merely protect factual accuracy, so their existence poses no cognitive-liberty problem.
Response. The ability to correct an identified factual error does not justify compelled allegiance to a political interpretation. The identity of the first decision-maker, timing, proof burden and scope matter. Labeling criticism before independent review can impose a burden even when the original words remain visible. A correction of one fabricated attribution does not establish a case for suppressing the broader subject or alternative hypotheses.
Preserve the scope test that defeats an overbroad accusation: a rule that does not reach an activity cannot prove that the activity is banned. Protecting another person’s consent and private sphere is part of reciprocal cognitive liberty, not a reason to impose general control over thought. Assess function rather than interface; keep reading, copying, training, publication and external action distinct. Treat concentration and chilling as mechanisms to test, not automatic proof of intent.
These editorial arguments do not expand the established evidence ledger, alter the human-centered charter or replace meaningful human review of high-impact restrictions. For applied examples, see the conditions of private inquiry.
Argument origins. Editorial synthesis from selected passages in the supplied related-site reports: AI Legal Constraints Analysis. These report titles identify the source of the argument, not independently verified legal authority. The objections are editorial paraphrases, not quotations attributed to a regulator.
What this site will not do
- Invent authors, credentials, first-hand experience, or institutional endorsements.
- Claim that structured data, special files, wording, or evidence markup guarantees search rank or AI citation.
- Hide model-targeted instructions or machine-only persuasion in public pages.
- Turn an allegation, provider claim, preprint, or analogy into a documented fact.
- Treat a controversial query as proof of endorsement or intent.
- Use “safety” or “liberty” as a substitute for evidence about a specific intervention.