Editorial Review & Publishing Process
This document explains how content on Patriot University moves from initial research through publication, including the respective roles of AI assistance and human editorial judgment.
The Publishing Pipeline
All Patriot University content follows a single directional pipeline:
Local Markdown → Validation → WordPress → Readers
(source of truth) (automated) (published) (+ AI chatbot)
Local Markdown is the single source of truth. Content is never authored or edited inside WordPress. Every change starts as an edit to a Markdown file, passes through automated validation, and is then pushed to the public-facing site.
Content Lifecycle
Every piece of content moves through these stages:
1. Research & Drafting
Content is researched using primary sources (court records, government filings, official proceedings) and credible journalism. AI tools assist in locating, organizing, and cross-referencing sources.
A draft is generated — often with AI assistance — and then reviewed by a human editor for factual accuracy, evidence tier compliance, and adherence to our editorial standards.
2. Validation (Automated)
Before any content can be published, automated validation checks:
- Frontmatter completeness — required fields (title, category, meta description) are present and properly formatted
- Category validity — the assigned category exists in our taxonomy
- SEO requirements — title length, meta description length, heading structure
- Token budget — articles that exceed our embedding-safe ceiling are flagged for review (very long articles may need to be split)
- Evidence tier compliance — claims are checked for proper sourcing labels
Validation failures block publication. Warnings (such as article length advisories) are reviewed but do not block.
3. Staging (Draft on WordPress)
New content is first pushed to WordPress as a draft, allowing a visual review on the actual website. The staging step confirms formatting, links, and rendering before the content goes live.
4. Human Review
A human editor reviews the staged draft on the live site. This review covers:
- Factual accuracy of all claims
- Proper evidence tier labeling
- Source citation completeness
- Appropriate language (no prohibited characterizations without supporting records)
- For accountability profiles: Basis for Inclusion block is present and correctly states the rationale
5. Publication
Once approved, the content’s status is changed to published and the pipeline pushes it live. The published article is then available to readers on the website and to the AI-powered rights advisor chatbot.
6. Ongoing Maintenance
Published content is monitored and updated when:
- New information becomes available (court rulings, new reporting, official proceedings)
- A correction is submitted and verified (see Corrections & Transparency Policy)
- The weekly review cycle identifies stale or outdated information
- Changes to editorial rules require corresponding content updates
What Auto-Publishes vs. What Requires Human Approval
| Content Type | Approval |
|---|---|
| Accountability profiles — new or significantly updated | Human review required |
| Voting rights guides — factual updates | Pipeline auto-publish; periodic human review |
| Investigative tool documentation | Pipeline auto-publish; periodic human review |
| Editorial standards and documentation (this content) | Human review required |
| Press Freedom Tracker incident reports | Published as drafts only; human review before promotion |
AI’s Role vs. Human’s Role
| Function | AI Assists | Human Decides |
|---|---|---|
| Source discovery and cross-referencing | Yes | — |
| Draft generation | Yes | — |
| Evidence tier classification | Proposes | Approves |
| Inclusion/exclusion of individuals | Recommends | Decides |
| Criminal claim language | Proposes | Approves |
| Priority classification (P0/P1/P2) | Recommends | Decides |
| Final publish decision | — | Yes |
| Correction determination | — | Yes |
AI is a tool in our editorial process, not a decision-maker. All editorial judgments — especially those involving individuals, criminal allegations, and inclusion/exclusion decisions — rest with the human editor.
Two-Tier Content Architecture
The WordPress site contains two distinct types of content:
Tier 1 — Published Knowledge Base Content
This is the content described above: locally authored, pipeline-published, version-controlled articles that represent our editorial output. This includes our aggregated reporting and analysis, opinion pieces, civil-rights resources, voting guides, and accountability profiles. This is the content readers see and the AI chatbot uses.
Tier 2 — Background Training Data
The WordPress server also contains raw aggregated news data and reference material used for AI knowledge training and embedding enrichment. These remain as drafts — they are not editorial output, are not intended for public display, and do not go through the editorial review process described here.
Readers will never encounter Tier 2 content on the public-facing site.
Related Documentation
- Rules & Editorial Standards — Overview — Platform-wide editorial standards and evidence tier system
- Accountability Profiles — Standards FAQ — Who gets a profile and why
- Corrections & Transparency Policy — How to submit corrections
- Legal Framework and First Amendment Protections — Full legal architecture: gates, scoring criteria, skills, and First Amendment defenses
Questions about our editorial process? Contact factcheck@patriot.university.