About & FAQ
Project overview
The Leakage Theory is an independent research project that proposed a 5D braneworld explanation for the S₈ and Lyman-α cosmological tensions, derived it from first principles, tested it, and published the negative result. It is maintained by a single author (Sparky / GeometricCosmo) in Cape Town and published openly for criticism.
Methodology
This project demonstrates a simple four-step method:
- Propose a bold hypothesis — extra dimensions explain the tensions.
- Derive it from first principles — the 5D Einstein–Gauss-Bonnet solution.
- Test it independently — the CAMB Boltzmann pipeline.
- Report honestly — publish the falsification when the test fails.
The value of the project lies not in confirming a theory but in showing how to properly test an exotic hypothesis, identifying a structural observational constraint, maintaining integrity when the result is negative, and leaving a roadmap for resurrection attempts (Path C). See the Stage 3 results.
Contact
- Email: geometriccosmo.illusion559@passinbox.com
- Location: Cape Town (UTC+2)
- Response time: < 48 hours
- Github: gb-leakage-cmb
Frequently Asked Questions
Changelog
Current version: v1.9.0 — updated 2026-08-01.
Stage 3 complete — honest falsification. The full 5D Einstein–Gauss-Bonnet system was solved as a boundary-value problem (residual < 6×10⁻⁶) and holographic entanglement entropy uniquely determined the transfer function, giving a k^(−1/2) power law from geometry rather than fitting. Direct CAMB testing then showed that the S₈ constraint (k ≈ 0.5 h/Mpc) and the Lyman-α constraint (k ≈ 5 h/Mpc) cannot be satisfied by a single monotonic transfer function; the two sit about a decade apart in wavenumber. The incompatibility is structural, not a tuning problem, so the power-law radion-leakage mechanism is treated as observationally excluded (<20%). Both tensions remain real and open. Three resurrection paths are under consideration, with Path C (partially-coupled dark matter) receiving current effort; it is speculative and not yet derived.
Performance & CLS pass. Font-metric overrides for Inter and Instrument Serif eliminate the layout shift that used to hit when webfonts swapped in. The cosmic-backdrop starfield now animates only on ≥768px viewports with prefers-reduced-motion: no-preference — mobile keeps the static gradient and drops a full-viewport repaint loop. Below-the-fold diagrams use content-visibility: auto with intrinsic-size hints, so browsers skip painting them until they scroll near the viewport (SSR HTML still contains full a11y titles/descriptions so crawlers and LLMs read everything). Root head gains color-scheme: dark, a font-CSS preload, and a fonts.gstatic dns-prefetch. Global reduced-motion media query neutralises decorative animations. No content, copy, physics, structure, or routes changed.
Visual polish pass. A global cosmic backdrop (layered radial gradients + subtle drifting starfield) now sits behind every page for a unified feel. The home hero gains a gradient-clipped title, a soft cyan CTA glow, animated pulse dots, and tightened rhythm. Cards across the site adopt a glass treatment (translucent + backdrop blur + inner highlight) with a small hover lift. Diagrams get soft SVG glow filters on the brane, radion dial, and leakage band. Header nav gains an animated underline for active/hovered links; footer gets a small logo dot and gradient dividers replace hard borders throughout. No content, copy, physics, structure, or routes changed.
Explainer pass for both casual browsers and physics-literate readers. Every jargon term now has a plain-language tooltip that links to a new /glossary route. Each conceptual section opens with a four-panel 'What / Why / How / So what' card. Two hand-authored inline SVG diagrams (brane + radion, leakage timeline) anchor the mental model. A layered 'Go deeper' primitive lets researchers expand technical bridges in place without cluttering the surface read. No equations, numbers, or existing brick derivations were changed.
AI & search visibility fix. The /framework page — which holds the full seven-brick theory, all equations, diagrams, FAQ, and references — was previously gated by a client-only disclaimer that ran inside useEffect. Search engines and AI crawlers (Googlebot, GPTBot, ClaudeBot, PerplexityBot, etc.) therefore only received an empty <div> in the SSR HTML and never indexed any of the theory content. The disclaimer is now a client-only dismissible overlay layered above content that is always server-rendered, so crawlers and LLM answer engines can read the full page on first request. Adds a public /llms.txt (llmstxt.org convention) summarising the project and linking every core page for AI indexing. No visible change for human visitors — the disclaimer still appears on first visit with the same wording and 'I understand' button.
Multi-page architecture. The single-page landing site is split into a lean gateway home page plus six focused sub-pages: Understanding (5-minute intro), Framework (seven-brick technical deep dive — the previous landing content, in full), Evidence (predictions, confidence tiers, falsification criteria, four-stage roadmap), Get Involved (contribution paths + critical path), Resources (docs, downloads, references), and About & FAQ. A persistent site header and footer replace per-page nav. No content is removed — the framework page still contains every existing section verbatim; the other pages provide curated entry points that link back for deep detail. Sitemap updated to include all seven routes.