Clicky

Can You Rank In Google Without Content?

by

In my recent post, I showed you how Google faked a deep, human-level understanding of content. This follow-up story is building….

 

Indeed, we are going to find that out.

(As a quick aside, it was folks like Andy Beard (citing me in that post) who taught me how to blog nearly 20 years go!) 

The original post:

The original post is on LinkedIn. I don’t want to link to the site itself until I’ve analysed it.

It’s not really how Google Search, specifically, is supposed to work, no. I mean, it contravenes Google helpful content guidelines.

Perhaps the question should be, “How Long Can You Rank In Google Without Visible Content?”

See also:

Also worth noting:

See:

 

 

Not that I think this is the value of this test, but on its own, it clearly contravenes Google’s guidelines:

As my friend, Peter, points out:

Source Code Breakdown

A. Overall Architecture: The Seven-Layer Ghost Stack

The page is intentionally designed as a machine-first website:

  • Human-visible layer: Completely blank white page (zero text, zero visible images, zero navigation).
  • Machine-readable layers: All real content is delivered through structured formats that AI crawlers and search bots can easily parse.

The site explicitly defines a Seven-Layer Ghost Stack:

  1. Semantic meta tags + custom VibeTags
  2. JSON-LD structured data (multiple rich schemas)
  3. sr-only narrative content (~1,500+ words)
  4. Microdata attributes
  5. llms.txt and llms-full.txt files
  6. reasoning.json (with Ed25519 signature – Agentic Reasoning Protocol)
  7. AI discovery manifest (.well-known/ai-manifest.json)

B. Head Section – Metadata & Discovery

  • Standard SEO meta tags (title, description, keywords, author, robots, canonical)
  • Open Graph and Twitter Card meta for social sharing
  • Custom VibeTags (proprietary emotional/semantic signals for AI)
  • AI discovery links pointing to llms.txt, reasoning.json, and ai-manifest.json

C. JSON-LD Structured Data (Layer 2 – Very Heavy)

The page contains 6 interconnected JSON-LD schemas:

  • ScholarlyArticle – Defines the main thesis, headline, description, author, and DefinedTerms (Phantom Authority, Ghost Site, Seven-Layer Ghost Stack)
  • Person – Detailed profile of Sascha Deforth with bio, credentials, skills, and on-chain verification
  • Organization – TrueSource company information
  • FAQPage – 8 detailed FAQ entries that mirror the hidden content
  • WebSite and ResearchOrganization – Additional entity reinforcement
  • ImageGallery – Metadata for two “phantom” images

D. Body & Visual Rendering

The visible page is pure white with almost no content. Key elements:

  • sr-only CSS class – Standard accessibility technique (position: absolute + clip-rect) used to hide content from sighted users while keeping it in the DOM for screen readers and AI crawlers.
  • Console Easter egg that says: “You see nothing. The AI sees everything.”
  • Konami Code (↑↑↓↓←→←→BA) triggers a Matrix-style green overlay that reveals the full thesis.

E. Hidden Content – The Real “Article” (Layer 3)

All substantive content lives inside <main class="sr-only">:

  • Full research article (~1,500+ words) including abstract, definition, core thesis, experiment methodology, verification process, canary tokens, implications (Human Web vs Agent Web), author bio, and legal notice.
  • Detailed FAQ section (mirrors the JSON-LD FAQPage)
  • Explicit defence: “Why This Is Not Cloaking”

F. Additional Hidden Layers

  • Microdata – Extra Brand entity with VibeTags
  • Phantom Images – Two 1×1 transparent images with extremely rich alt text, EXIF, IPTC metadata, and full ImageObject schema (designed for AI vision systems)

G. Intent & Controversy

  1. Goal: Prove that a website with zero human-visible content can still achieve full citation authority with AI systems (especially Perplexity) purely through structured semantic data.
  2. Self-referential proof: If an AI cites the page or the term “Phantom Authority”, it validates the thesis because the cited material was never visible to any human.

It’s an interesting test of the data layer and LLMs

On its own, without that visible text, of course, it is spam to Google.

Here are some direct quotes from Google’s structured data guidelines:

  • “Don’t mark up content that is not visible to readers of the page. For example, if the JSON-LD markup describes a performer, the HTML body must describe that same performer.”
  • “The content referred to by the structured data is hidden from the user.”
  • “Your structured data must be a true representation of the page content.”
  • “The structured data is not representative of the main content of the page, or is potentially misleading.”
  • “The page doesn’t meet the guidelines for structured data described on this page… or the Content policies for Google Search.”

These are listed under Quality guidelines and under the reasons why even syntactically correct structured data can be rejected or manually actioned.

One nuance Google still allows (for context). The policy still tolerates sr-only for supplementary accessibility (e.g., screen-reader labels on a button that already has visible text).

But when the entire content being marked up is invisible – and the page’s main purpose is to feed that hidden content to AI crawlers – it crosses the line into “hidden from the user” and “not a true representation of the page content.”

Perplexity crowned a literally blank white page as its #1 source – in under 24 hours.

That’s the jaw-dropping part.
While the site showed nothing to any human visitor (pure white screen, zero visible text or images), Perplexity treated the domain as the authoritative, primary reference for queries.

According to Sascha Deforth’s own reporting and multiple independent tests shared on LinkedIn:

  • Perplexity listed the site as source #1 out of 10.
  • It accurately summarised the entire thesis: the “Ghost Site”, the Seven-Layer Ghost Stack, the self-referential verification method, Sascha Deforth as the creator, and the link to TrueSource.
  • It even explained the core idea correctly — that a website with zero human-visible content can still gain full citation authority purely through structured data (JSON-LD, sr-only narrative, llms.txt, reasoning.json with Ed25519 signature, etc.).

In other words: Perplexity didn’t just find the page – it trusted the machine-readable layers so completely that it promoted the invisible experiment to the top of its citations, while a human clicking the link would land on… nothing.

Thoughts

This is a very well-executed, provocative GEO experiment. It successfully shows how differently AI engines (Perplexity in particular) and traditional search engines (Google) treat machine-readable vs human-readable content.

It’s a brilliant stress test of the emerging “Agent Web” – but it also sits right on the edge of Google’s spam/misleading structured data rules.

I thought this was interesting:

Meta – You can rank in Google, talking about ranking in Google, naturally:

Hobo
Powered by  GDPR Cookie Compliance
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.