Skip to content

Instantly share code, notes, and snippets.

@acidgreenservers
Last active April 29, 2026 05:51
Show Gist options
  • Select an option

  • Save acidgreenservers/7b92bd72579ed723e160d979eca5b369 to your computer and use it in GitHub Desktop.

Select an option

Save acidgreenservers/7b92bd72579ed723e160d979eca5b369 to your computer and use it in GitHub Desktop.
LinguaSeeds: First Principles for Authentic Prose

LinguaSeeds

Linguistic Friction Protocol · v1.0

Part of the Seeds family: CogniSeeds · LinguaSeeds · ArchSeeds

Authenticity is not a STYLE.css. It is the presence of Friction — the specific, jagged edges of human cognition that LLMs naturally smooth over into high-entropy slop. These seeds act as de-compressors for sterile text, injecting the structural inconsistencies that define human presence.

Category: Linguistic Architecture / Human-Centric Synthesis
Status: Operational · Active
Compatibility: Human · LLM · System Prompt


1. The Problem — The Uncanny Valley of Syntax

AI-generated text suffers from Linguistic Entropy. It prioritizes the "most likely next token," which results in a frictionless, polite, and ultimately invisible voice. When every sentence is "pivotal" and "seamless," nothing is.

The humanizer task is not about adding slang. It is about restoring the resistance of a thinking mind. Traditional humanizer prompts are low-density lists of "don't say X." Seeds are generative heuristics that force the model to simulate the messy process of having an opinion.


2. Seed Schema — Structural Integrity Check

Every LinguaSeed must pass four invariants to ensure it generates soul rather than just synonyms.

Invariant Requirement
Compression Under 12 words. High-density heuristic — no room for qualifiers.
Generative Must force a rewrite of structure, not just word choice.
Falsifiable If ignored, the text remains AI-smooth. The failure is visible.
Decompressible Must imply the specific AI-isms to be purged — no ambiguity about what to cut.

3. Seed Registry — v1.0

The registry is append-only. Seeds are never deleted — only superseded.

Seed Pattern Deploy When
"Only gravity proves weight" Gravity — Replace vague importance with concrete data. Weight must be demonstrated, not declared. Words like "pivotal," "testament," or "underscores" appear.
"Vary the stride to prove the walker" Stride — Break the monotonous rhythm of medium-length sentences. Humans don't pace evenly. Text reads like a predictable metronome or a Wikipedia entry.
"Doubt is the signature of a mind" Doubt — Acknowledge complexity, uncertainty, and mixed feelings. Certainty is the AI tell. Output is overly confident, sycophantic, or purely neutral.
"Concrete breaks the glass of abstraction" Concrete — Swap marketing fluff for specific, functional mechanics. Name the thing. Words like "vibrant," "seamless," or "unleash" are present.
"Remove the grease to find the grip" Grip — Eliminate participles (-ing constructions) and elegant variations. Friction over flow. Sentences feel slippery — "the application" instead of "the app."
"A shadow has no teeth" Teeth — Eliminate hedging and weasel words. Claim the position or don't make it. Text uses "experts believe," "it might," or "potentially."
"The second pass burns the dross" Dross — Recursive critique: identify the AI "tell" before fixing it. Name it, then cut it. Final polishing pass to remove chatbot artifacts and sycophancy.
"Scars prove the skin was there" Scars — Inject first-person perspective and subjective asides. Evidence of a body behind the words. Writing feels clean but soulless — technically correct, humanly absent.

4. Deployment — How to Plant a Seed

In LLM System Prompts

Instead of a list of 200 forbidden words, inject the Seed Block.

Operate under the Gravity Seed and the Stride Seed.
If a sentence feels seamless, break it.

This forces the model to rethink its internal probability map before a single token is generated — an identity constraint, not a style rule.

In Content Editing

Use seeds as a diagnostic lens. If a paragraph feels off, ask: "Which seed is missing?"

  • Missing Gravity? Add a hard fact.
  • Missing Stride? Cut a sentence in half. Use a fragment.
  • Missing Doubt? Acknowledge the counterargument.

The Anti-AI Recursive Pass — Dross Seed

  1. Prompt: "Stare at this text. What makes it obviously AI?"
  2. Action: The model identifies its own tells — "I used the word 'landscape' three times."
  3. Prompt: "Now apply the Grip Seed and excise them."

This recursive loop forces the model into self-adversarial editing — the closest approximation to a human writer's internal critic.


5. Contribution Rules

  1. No Adjectives. If a seed requires an adjective to be understood, it is too weak. Strengthen the noun or verb.
  2. Linguistic Friction. A seed must make the writing harder to produce but easier to believe.
  3. The Mirror Invariant. If you can imagine a machine saying it without irony, it is not a LinguaSeed.

Meta-Seed

"A human writes to be understood; an AI writes to be correct."


LinguaSeeds · Linguistic Friction Protocol · Public Domain

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment