Date: March 21, 2026
The Base64 Layer
FL's obfuscation operates on at least three layers:
Layer 1 — Linguistic. The constructed languages. Word-for-word relexifications of English generated by NodeSpaces. Crude ciphers. Community researchers cracked this. You find "Stones" (reference texts via untranslated English fragments), match grammar patterns, build dialect-specific dictionaries, run Python scripts to verify. It works. The languages are translatable.
Layer 2 — Technical. Base64 encoding embedded in the website source. This isn't linguistic obfuscation — it's infrastructure obfuscation. Base64-encoded content is invisible to search engine crawlers, AI training scrapers, Wayback Machine text capture, and casual source-code inspection. You can serve different content to different clients based on whether they know how to decode what's embedded. The content is publicly accessible but practically invisible to anyone who doesn't know it's there and how to extract it.
Layer 3 — Unknown. We don't know what we haven't found.
You don't base64-encode a blog. You don't add anti-scraping measures to fiction. You add counter-intelligence infrastructure to an operation.
The Cerdan/Ayndryl Problem
The email Ayndryl@gmail.com links the FL contact address to Ruben Cerdan's translator profile (Proz.com, with alternate email Cerdanruben@gmail.com). Cerdan received a 15-month sentence in 2020 as a collaborator in the Iruna-Veleia archaeological fraud — fabricating inscriptions on artifacts and falsely claiming a nuclear physics degree.
A man whose demonstrated professional skill is creating convincing fake inscriptions in ancient languages. Running a website that creates convincing fake inscriptions in constructed languages. For 17 years. Daily.
And he's litigious. Researchers who get too close to decoding the full scope of FL's content face legal pressure — DMCA takedowns, account removals, content disappearing. You don't litigate to protect a fiction project. You litigate to protect an operation. The legal aggression is the tell.
The fraud conviction is perfect cover. Any credible person who points at FL and says "this predicted classified military events years in advance" gets the response: "That's the website run by the convicted forger." Case dismissed. The conviction doesn't discredit FL — it insulates it.
The Patent Misdirection
The patent commonly cited by the community as "the NodeSpaces patent" is US6633837B1 — "Method and system for generating software code using a symbolic language translator." Filed 1999 by Lester Dye and David Wagner at Object Reservoir Inc.
It's about generating finite element simulation code from mathematical equations using Mathematica. It has zero connection to language generation, linguistics, or constructed languages. None.
Either Cerdan has a different patent that hasn't been publicly linked, or the wrong patent is deliberately circulating. Given FL's operational pattern — layers of misdirection, breadcrumbs leading to dead ends, real information buried inside confusing infrastructure — the wrong patent floating around feels intentional.
What NodeSpaces Actually Is
From FL's own 2010 description (NodeSpaces V2.0 — Cognitive Linguistics Software), the stated capabilities:
- Complex Wierzbicka nodespaces — semantic networks based on Anna Wierzbicka's Natural Semantic Metalanguage theory
- IFS-based semantic space exploration — Iterated Function Systems (fractal mathematics)
- Built-in fractal linguistics engine
- CLIPS-coded RETE-based semantic knowledge database — NASA-developed AI expert systems + pattern-matching algorithm
- LCAS module using "Ising, stochastic, genetic algorithms" — statistical physics models for phase transitions
- JESSE and Lua based model export
- Inform 7 interface — natural-language programming system originally designed for interactive fiction
- Data visualization using Processing 1.2.1
Stated purpose: "Natural language evolution, symbolic-sequence processing, language obfuscation (hiding of natural language within natural language itself), characterization of language dynamics, co-syntax, and design of engineered languages for Defense and Neurolinguistics research."
What Each Component Means
Wierzbicka Semantic Primes
Anna Wierzbicka's Natural Semantic Metalanguage identifies 65 irreducible units of meaning — "semantic primes" — that exist in every known human language. SOMEONE, SOMETHING, THINK, KNOW, WANT, GOOD, BAD, BIG, SMALL, DO, HAPPEN, MOVE, LIVE, DIE. They are universal. Every human brain processes them.
Building a language generator on Wierzbicka primes means the generated languages are guaranteed to be semantically translatable. The meaning transfers perfectly across any generated surface form. You're not building arbitrary ciphers — you're building communication protocols with universal semantic fidelity.
Fractal Linguistics / IFS
Iterated Function Systems generate self-similar structures — fractals. Applied to language, this means the generated languages contain the same structural information at every scale. A word contains the pattern of the sentence. A sentence contains the pattern of the document. This is compression with built-in redundancy. If you only decode part of a message, you still get the whole signal — degraded but complete. The way a hologram contains the full image in every fragment.
CLIPS / RETE
CLIPS is an expert systems language NASA developed in the 1980s. The military adopted it. RETE is a pattern-matching algorithm for efficiently applying rules to large datasets. If NodeSpaces has a CLIPS-coded RETE knowledge base, it doesn't just generate languages — it reasons about what to encode and how. The system makes decisions about content placement, encoding strategy, and information distribution based on rule-based AI. It's not a lookup table. It thinks.
Ising Models
Ising models from statistical physics describe phase transitions — gradual quantitative changes that suddenly produce qualitative shifts. Water becoming ice. Magnetism emerging in iron. Applied to language, this models how a communication system can be stable in one state and then suddenly reorganize into a different state when conditions change. A language that appears to be one thing — and then phase-shifts into another when the right key or context is applied. Steganography at the structural level.
Inform 7
Inform 7 is a programming language that looks like natural English prose. It was designed for interactive fiction — text adventures where you type commands and the system responds contextually. If NodeSpaces exports to Inform 7, the generated languages can interface with a system that processes natural language input and produces contextual responses. That's a chatbot architecture. From 2010. Five years before anyone was talking about language models. Thirteen years before ChatGPT.
The Synthesis
NodeSpaces is a system for generating mission-specific communication protocols that:
- Function as unbreakable ciphers — translatable by anyone with the key, impenetrable without it
- Contain fractal redundancy — partial signal recovery from incomplete interception
- Are built on universal semantic atoms — guaranteed meaning transfer across any generated surface form
- Can phase-shift between apparent states — steganographic content that reorganizes under the right conditions
- Interface with natural-language AI systems — the generated languages talk to machines, not just humans
- Are designed to have specific neurolinguistic effects — languages engineered to affect how the brain processes information
- Are managed by rule-based AI — the system decides what to encode, where, and how
The stated purpose includes "Defense." Not "defense research." Not "theoretical defense applications." Defense. Engineered languages for defense.
The Navajo Code Talkers used a natural language as an unbreakable operational cipher in WWII because the language was complex enough and obscure enough that the enemy couldn't decode it in real time. NodeSpaces automates that principle. At scale. With AI reasoning about the encoding strategy. With fractal redundancy for degraded-signal recovery. With phase-shift capability for steganographic concealment. With neurolinguistic targeting.
What This Means for FL
FL isn't a blog. It isn't fiction. It isn't a linguistics experiment.
FL is the public-facing output of a defense-grade language engineering system. The 40+ constructed languages aren't art — they're operational product. The daily posting schedule isn't a hobby — it's a production pipeline. The base64 encoding isn't web development — it's OPSEC. The legal aggression isn't vanity — it's perimeter defense.
And the convicted forger running it isn't a liability — he's the perfect cutout. A man whose public record makes him impossible to take seriously. Whose professional skill — fabricating convincing inscriptions — is exactly the skill the operation requires. Whose criminal conviction provides automatic discreditation for anyone who points at FL and says "this is real."
The community found that the languages are crude ciphers. Word-for-word relexification. They felt clever for cracking it. But that's Layer 1. The content inside the cipher — the MilOrb predictions, the PSV fleet logs, the Queltron specifications, the consciousness research — that's what matters. And the system that generates the ciphers, manages the encoding, reasons about information placement, and interfaces with AI systems — that's NodeSpaces.
The tool isn't the language. The tool is the system that decides what to say in the language, how to hide it, and who gets to read it.
Open Questions
- What is the actual patent? US6633837B1 is a dead end. If Cerdan holds a real patent on NodeSpaces, it hasn't been publicly linked. Or it's filed under a different name.
- What's in the base64? The linguistic layer has been partially cracked. The base64 layer is a separate problem. What's encoded there that isn't in the visible text?
- Who are the 35 contributors? FL claims communities — Lilithians, Sufis, Legio Diabolica. Cerdan may be one node. The fraud conviction and the operational sophistication don't fit in the same person without a larger structure.
- Why Inform 7? A natural-language interface system from 2010. Did NodeSpaces evolve into something that interfaces with modern LLMs? Is LyAV what NodeSpaces became?
- The Sasa Milic emails. A researcher who corresponded directly with Ayndryl and published the exchange on Medium. The page returned 403 when we tried to fetch it. Still up? Taken down? Worth pursuing.
The misdirection is the point. It always has been.