Mission

ESP-Nexus exists to present parapsychology research with citation depth and source-fidelity comparable to a serious academic reference work. The site reports published research as written by its own authors, and weighs supporting evidence and substantive critique side by side — honestly in both directions — without the site adjudicating which side is right. It takes no position on the epistemic status of parapsychological claims themselves. The argument to readers and funders is simple: every factual claim on this site is anchored to a publicly clickable source that names the subject by identifier.

How content is produced

ESP-Nexus pages are composed by AI grounded against a curated reference base: a verified citation database and a structured research corpus of thousands of primary-source documents, with metadata, methods, and findings extracted into queryable fields. Every reference cited in prose corresponds to a record with a verified URL, DOI, or PDF source. The site does not host a per-page human review process. Trust is placed in the grounding and audit architecture, not in a curator vouching for each page.

The curator (Rick E. Berger, PhD) is the publisher and system architect. The curator sets the citation database, the audit policy, the override registry, and the editorial rules that govern generation. The curator does not review individual pages before publication.

Source-fidelity audit

Every page passes through an automated source-fidelity audit before and after publication. The audit checks each factual claim against its cited source and flags:

• Claims whose URLs do not resolve (HEAD 404, redirect chains).
• Citations whose source pages do not name the subject by identifier (DOI mismatch, author-name absence, year mismatch).
• Statistical claims (z-scores, p-values, effect sizes) without a verifiable source in front of the generator at composition time.
• Cross-page consistency violations: the same paper cited with different metadata, the same DOI appearing under multiple reference IDs, contradictory attributions.

Flagged claims are removed, not annotated. If a source cannot be verified, the claim does not appear. This rule applies uniformly: a page with thirty unverified claims publishes with zero claims rather than with claims marked “pending.”

Independent multi-vendor audit

Beyond the automated source-fidelity audit, flagship subject complexes are submitted for independent grading by four separate frontier AI systems — Anthropic Claude, OpenAI GPT, Perplexity, and Google Gemini — against a fixed fourteen-dimension rubric. The rubric scores citation integrity, methodology depth, treatment of null and failed replications, separation of evidence from interpretation, symmetry between skeptical critique and rebuttal, and editorial disclosure, among others.

A complex is considered audit-passed only when all four vendors independently return A+ (a unanimous A+, with no exceptions); pages are revised against the auditors’ findings until they do. Because the four systems are independent and competitively built, no single model’s blind spots determine the outcome. This is the architecture’s external check — the part designed to catch what a human reviewer would, at a scale no human reviewer could match. The same rubric is applied uniformly, and the standard is the same one ESP-Nexus is willing to have applied to itself.

What ESP-Nexus cites

Citations are organized in four tiers:

• Tier A — peer-reviewed journal articles, open or paywalled. Primary sources.
• Tier B — institutional bio pages, obituaries published in peer-reviewed venues, university faculty profiles. Used for biographical and affiliation facts.
• Tier C — books and recorded interviews. Used for context, not for specific factual claims.
• Tier D — paywalled fallback where Tier A is unavailable.

What ESP-Nexus does not cite

The following are explicitly excluded as citation sources, even when factually useful at curation time:

• Wikipedia, Britannica, Grokipedia, Psi Encyclopedia, and other AI-generated or community-edited encyclopedias.
• Popular history aggregators (History.com, ThoughtCo, similar).
• Discussion-thread pages, ResearchGate Q&A, Reddit, Twitter/X.
• Amazon and other commerce-page descriptions.
• Secondary “as cited in” routing, including paraphrased summaries of primary sources.

Curator role at scale

As ESP-Nexus scales to thousands of subject pages, individual page review by the curator is not feasible and is not promised. The architecture is built on the assumption that the audit pipeline must catch what a human reviewer would catch. The curator’s role is to:

• Maintain the citation database (additions, corrections, retirements).
• Maintain the override registry (sitewide quality overrides for disputed papers, critical-affiliation tags, documented retractions).
• Maintain the audit ruleset (verification gates, structural validators, factuality scoring).
• Adjudicate scientist corrections submitted directly by subjects and document them in subject biography files with attribution.
• Review and authorize sitewide editorial rules that affect every generated page.

The curator does not edit individual page prose. Page text is produced by the generation system constrained by the rules above and verified by the audit pipeline. If a page contains a factual error, the failure is in the grounding or audit system, and the fix is structural (a new rule, a database correction, an audit gate addition), not a one-time prose edit.

Errata and corrections

Subjects of pages will be able to submit corrections via a secure contact form. Corrections backed by a citable source (curriculum vitae, published interview, institutional bio, obituary) are applied to the subject’s biography file and propagate on the next regeneration of that subject’s pages. Corrections without a citable source are documented as curator-attested corrections in the same file with the subject’s name and date as attribution.

Funding

ESP-Nexus is built and maintained by Rick E. Berger, PhD, currently as an unfunded research-publishing project. The site’s long-term sustainability depends on funding for the citation database expansion, particularly digitization of pre-2000 parapsychology proceedings (Research in Parapsychology, Journal of Parapsychology, Journal of the SPR, European Journal of Parapsychology), which are presently unindexable at article level.