Wikipedia leapfrog 2030 — design-research substrate

Source: research-wikipedia-leapfrog-2030.draft.md (project-knowledge, 2026-04-30) Authors: claude-opus-4-7 parent + 4× claude-sonnet-4-6 sub-agents Cites: Wikipedia Main Page, Vector 2022 design docs, DTCG 2025.10, Carbon v10, Wikimedia Codex

---

This research consolidates four parallel Sonnet sub-agent reports into a single design-research record for the substrate. It is the canonical "what we learned" for every downstream DESIGN-COMPONENT and DESIGN-TOKEN-CHANGE draft staged in the same pickup batch (citation-authority-ribbon, research-trail-footer, freshness-ribbon, knowledge-wiki-baseline tokens).

1. Why this research was commissioned

documentation.pointsav.com shipped its first Wikipedia-Main-Page-shaped chrome at workspace v0.1.70 (2026-04-29): 9 muscle-memory items at the article shell (Article/Talk tabs, Read/Edit/View-history tabs, per-section [edit] pencils, hatnote, lead first-sentence convention, tagline, collapsible left-rail TOC, language switcher, footer ordering) and a home-page composition of lede + featured-pin panel + 3×3 category grid + recent-additions feed. Iteration 2 needs to push from visible-operational into the leapfrog claim. This research is the bridge.

2. Wikipedia Main Page — primitive-level inventory

Slot	Role	Refresh cadence	Format invariant
Welcome banner	Orientation; community-scale signal via statistics	Continuous (live counter)	Single sentence + two stats; no marketing copy
Today's Featured Article	Flagship editorial showcase from 0.1% FA-reviewed articles	Daily	909–1,009 character paraphrase; bold linked title; "(Full article...)" closer
Did You Know	New-content discovery via hook fact	Daily	"Did you know that…" frame; 15–25 word hook; 9 hooks per rotation
In the News	Current-events bridge to encyclopedic depth	Non-uniform (consensus-driven)	4–6 blurbs + Ongoing + Recent deaths; one sentence per blurb
On This Day	Historical depth via temporal anchoring	Daily (pre-authored full year)	5–8 bullets in roughly reverse-chronological order
Today's Featured Picture	Visual anchor; standalone discovery	Daily (queue-driven)	One image + ~100 word encyclopedic caption + photographer credit
Other Areas	Reader → editor onramp	Static	7 linked items with 10–15 word descriptions
Sister Projects	Ecosystem disclosure across 12 Wikimedia projects	Static	Logo + name + 20–30 word description
Languages bar	Global scope signal across 346 language editions	Static (re-tiered as counts cross thresholds)	Three tiers by article count
Footer	License, privacy, contact	Static	License notice + last-updated timestamp; no advertising

The two-column asymmetry (left 65%, right 35%) is not stylistic — it is the same spatial grammar as a newspaper broadsheet. Density is measurable: a single Main Page screen at 1080p contains ~20–25 independently navigable units above the fold with no gratuitous padding separating sections.

3. Wikipedia article-shell weaknesses in 2026

Ten structural weaknesses from a leapfrog design perspective:

1. References section is a flat numeric list with no source-authority semantics. A peer-reviewed Nature paper and a personal blog post occupy identical visual registers.
2. Infoboxes are semi-structured but not natively machine-readable at the reading surface.
3. TOC is structural-only, with no semantic section typing. AI consumers cannot navigate by content kind.
4. What-links-here returns a paginated flat list, not a graph. No second-hop neighbours, no cluster grouping.
5. No inline-comment surface on the article reading view. Editorial discussion is a separate page.
6. No per-section last-edited or authorship granularity. Freshness illusion.
7. No reading-time or skim-aid. Article-size guidance is metadata at ?action=info, not in reading view.
8. No citation trail back to the specific cited passage. Footnote [4] resolves to a bibliography entry only.
9. No live-edit currency indicator. Articles edited dozens of times per day present no in-session change signal.
10. AI-consumption surface is unstructured at section granularity. 65% of expensive Wikimedia requests are AI scrapers collecting undifferentiated full-article data.

4. Competitive landscape — why no provider has replaced Wikipedia in 2026

Eight cross-cutting structural reasons (25 providers audited across: collaborative knowledge bases / public wiki engines / developer documentation site generators / personal-networked-thought tools):

1. Audience mismatch. Notion/Confluence/Coda were built for private organizational knowledge management — access-control model, pricing, and UX all assume a known trusted team.
2. No editorial constitution. Wikipedia's NPOV policy, Notability standards, RS policy, NOR rule, and MOS constitute a multi-decade-refined governance organization. No provider ships an equivalent.
3. Information density floor. Documentation site generators optimize for prose elegance — the wrong optimization for encyclopedic reference.
4. Navigation primitive missing. [[wikilink]] red-link signaling, Special:WhatLinksHere, category graph, disambiguation, navbox templates — no other provider ships even the red-link mechanism.
5. Citations are decorative, not load-bearing. Wikipedia's claim-level footnote system is unmatched.
6. No Talk-page substrate. The epistemic provenance layer — public record of every editorial dispute.
7. Structural brittleness. Notion/Coda/ClickUp use proprietary serialization formats; vendor-lock-in risk.
8. Template homogenization. Every Docusaurus/Starlight/VitePress site looks structurally identical — documentation aesthetic, not encyclopedic authority.

Genuine advantages competitors have that a 2030 leapfrog should steal:

• MkDocs Material's instant client-side search with offline support
• Capacities' typed-object relationship surface as navigable article metadata
• Obsidian Publish's hover-preview popover on [[wikilinks]]

5. The market-gap analysis

The gap exists because closing it requires simultaneously building governance software, a navigation primitive set, and an editorial culture — no commercial incentive in the last decade has pointed all three directions at once. Commercial incentives (seat licenses to organizations) actively point away from the Wikipedia structural model.

PointSav's substrate-sovereignty posture, three-tier compute routing, apprenticeship corpus capture, and editorial-pipeline infrastructure (project-language as gateway) give the substrate exactly the three preconditions no commercial competitor can simultaneously match.

6. Award-winning leapfrog primitives — the prioritized set

Three primitives are first-class (ship in iteration 2); two are second-class (iteration 3+). All five are additive to existing Wikipedia muscle-memory chrome.

6.1 Citation-authority ribbon (FIRST)

A small coloured badge on each entry in the References section indicating source category: academic (blue) / government-regulator (dark green) / industry-trade (orange) / direct-primary-source (teal) / news (grey) / web-informal (light grey). Class derived from citation template type. Emitted as @type refinement on citation entries in JSON-LD.

Addresses weakness #1 — flat numeric list with no source-authority semantics. Makes the citation substrate visible at the reading surface, directly expressing BCSC continuous-disclosure posture: citations are machine-readable, audit-traceable.

Component: components/citation-authority-ribbon/ (this batch).

6.2 Research-trail footer block (FIRST)

A collapsible footer block below the References section, rendered when article frontmatter declares research_trail: true. Three subsections per draft-research-trail-discipline convention (Doctrine claim #39): Research done / Suggested research / Open questions. Collapsed by default. Emitted as JSON-LD potentialAction nodes.

Combination of citation-authority ribbon + research-trail footer makes the article's epistemological position legible without reading all footnotes. No Wikipedia equivalent.

Component: components/research-trail-footer/ (this batch).

6.3 Freshness ribbon — per-section last-content-review date (FIRST)

An optional small badge on each section heading (right of the [edit] pencil) showing the date of the last substantive content change. Three-stop colour scale: fresh / stale / archived. Derived from git-blame at section level; frontmatter content_reviewed_on per-section override available. Per-section dateModified emitted as JSON-LD WebPageElement.

Addresses weakness #6 (freshness illusion) and weakness #10 (AI scraper full-article pulls because section-level freshness is undeclared).

Component: components/freshness-ribbon/ (this batch).

6.4 Plain-language toggle backed by curated authored paragraphs (SECOND)

A toggle in the reader-preference toolbar. When active, sections flagged plain_language: true render an alternative lead paragraph at a lower reading level. Plain-language paragraphs are explicitly authored by humans and committed to the article source — not LLM-generated at request time. Emitted as disambiguatingDescription on Schema.org TechArticle.

Second-class because curating paragraphs costs editorial labour that scales linearly.

6.5 Citation-graph mini-map — 3-hop neighbourhood (SECOND)

A collapsible section at the article foot showing a small SVG graph: the current article as centre node, 1-hop outbound wikilinks, 1-hop inbound links. Nodes labelled; edges directional. Emitted in JSON-LD relatedLink and mentions arrays.

Second-class because the wikilink graph must be pre-computed — high effort. Worth shipping when the article corpus reaches ≥200 articles.

7. Open questions for project-design ratification

Six real decisions gating substrate work:

(a) Light/dark theme switching pattern — Style Dictionary @value override / DTCG $extends / CSS :root[data-theme="dark"] selector.

(b) wiki.* vs ps-wiki.* semantic namespace prefix. Resolved by Master: wiki.* approved.

(c) Variable-font loading vs system-stack. Resolved by Master: system-stack.

(d) Density toggle as token vs component state. Token approach is more portable across rendering environments.

(e) FLI-banner colour register — amber vs neutral. Resolved by Master: NEUTRAL. Left-border accent + small disclosure icon; no amber.

(f) Research-trail visual weight — chrome vs body content. Affects article-research-trail-trail-bg token resolution.

8. Award criteria

A wiki shipping Wikipedia muscle memory + the five primitives in §6 is competitive in: Awwwards Site of the Day / Site of the Year · Webby Award (Websites: Reference) · Information is Beautiful Awards (Interactive / Tools & Services) · Communication Arts Interactive Annual · Open Source Awards (GitNation) · European Open Source Awards 2026 · MIT Technology Review Breakthrough Technologies.

9. Substrate-side scope

Four downstream DESIGN-* drafts staged in the same pickup batch (this batch):

1. components/citation-authority-ribbon/ — recipe for §6.1
2. components/research-trail-footer/ — recipe for §6.2
3. components/freshness-ribbon/ — recipe for §6.3
4. tokens/dtcg-bundle.json — DESIGN-TOKEN-CHANGE (master_cosign: 2026-04-30T17:00Z)

The home-grid component (§5 iteration-1 existing implementation) is also in this batch.

The two second-class primitives (§6.4, §6.5) are not staged as DESIGN-COMPONENT drafts in this batch; pre-emptive token-vocabulary slots may be added to the bundle.

Research trail

Done

• Wikipedia Main Page primitive audit (sub-agent A) — full inventory with format invariants
• Wikipedia article-shell anatomy under Vector 2022 (sub-agent B) — primitive inventory + weakness analysis
• Competitive landscape audit — 25 providers across 4 groups (sub-agent C)
• DTCG token-vocabulary proposal (sub-agent D) — three-tier taxonomy, full token inventory
• Existing draft review (topic-wikipedia-leapfrog-design, component-home-grid) — additive, no duplication
• naming-convention.md §2 §3 §4 §6 §10 — repo-rule design intent + 9-category set
• conventions/cluster-design-draft-pipeline.md §1 §3 §6 — DESIGN-* schema and master-cosign requirements
• conventions/draft-research-trail-discipline.md — Doctrine claim #39

Suggested

• Verify Carbon Tag baseline for citation badge visual treatment
• WCAG contrast audit for all 6 badge colour classes
• Grid breakpoint alignment with Carbon Grid breakpoints (960px vs Carbon's 1056px medium)

Open questions

• §7(a) light/dark theme switching — implementation strategy decision pending
• §7(f) research-trail visual weight — chrome vs body content posture