Understanding the Inflation and Borrowed Authority Patterns (12-18) in Humanizer

The "Inflation and Borrowed Authority" category in the blader/humanizer repository defines six specific AI-writing tells—cataloged as patterns 12 through 17 in SKILL.md—that detect inflated significance, vague associations, and unverified authority claims in prose.

The open-source Humanizer project provides a structured framework for identifying and neutralizing common AI-generated text artifacts. Within the repository's master pattern list at SKILL.md, the Inflation and Borrowed Authority group (entries 12-18) targets rhetorical strategies that artificially bolster claims without adding substantive information. These patterns map to internal identifiers I2 through I7, with pattern 18 marking the section boundary rather than defining a distinct tell.

Overview of the Inflation and Borrowed Authority Category

According to the source documentation in [SKILL.md](https://github.com/blader/humanizer/blob/main/SKILL.md), this category captures "AI-writing" tells where writers substitute grandeur for precision. The validation logic in scripts/validate-package.py ensures these pattern entries maintain consistent numbering and metadata integrity across the package. While the section spans entries 12-18, only six active patterns (I2-I7) are defined, with I8 representing the terminal boundary of the group.

The Six Anti-Patterns Defined

I2 – Inflated Significance

Inflated significance occurs when ordinary facts are dressed as "pivotal," "crucial," or "landmark" moments without evidentiary support. This pattern flags language that implies historical weight beyond what the source material confirms.

In SKILL.md (lines 13-14), the canonical example transforms:

Before: "The Statistical Institute of Catalonia was officially established in 1989, marking a pivotal moment in the evolution of regional statistics."

After: "The Statistical Institute of Catalonia was established in 1989, part of a wider decentralisation of administrative functions in Spain."

The revision strips the unsupported "pivotal moment" framing while preserving factual accuracy.

I3 – Vague Connection or Association

Vague connection or association detects phrases like "associated with," "linked to," or "tied to" that obscure the actual relationship between entities. This pattern demands specificity about the nature of connections.

As documented in [SKILL.md lines 24-31](https://github.com/blader/humanizer/blob/main/SKILL.md#L24), the transformation clarifies:

Before: "He is associated with the Rajhans Orchestra, which he founded and conducts."

After: "He founded and conducts the Rajhans Orchestra."

The revision eliminates the vague "associated with" in favor of the direct founding relationship.

I4 – Shallow -ing Riders

Shallow -ing riders identifies gratuitous participial phrases—such as "highlighting," "underscoring," or "symbolizing"—that attach to simple facts without adding meaningful content. These riders function as stylistic inflation devices.

The [SKILL.md entry (lines 35-40)](https://github.com/blader/humanizer/blob/main/SKILL.md#L35) demonstrates the compression:

Before: "The temple's colour palette ... symbolizing Texas blue-bonnets ... reflecting the community's deep connection."

After: "The temple is painted blue, green, and gold, colours meant to evoke Texas blue-bonnets and the Gulf of Mexico."

I5 – Sales Language

Sales language flags promotional adjectives and buzz-phrases—such as "groundbreaking," "breathtaking," or "vibrant"—that convert factual descriptions into marketing copy. This pattern enforces objective tone by removing hyperbolic qualifiers.

From [SKILL.md lines 46-49](https://github.com/blader/humanizer/blob/main/SKILL.md#L46):

Before: "Nestled within the breathtaking region of Gonder ... vibrant town with a rich cultural heritage."

After: "Alamata Raya Kobo is a town in the Gonder region of Ethiopia."

I6 – Borrowed Authority

Borrowed authority targets citations of unnamed experts, "some critics," or "industry reports" that lend unverifiable weight to claims. This pattern requires either concrete attribution or removal of the unsupported authority reference.

The example in [SKILL.md (lines 52-58)](https://github.com/blader/humanizer/blob/main/SKILL.md#L52) shows:

Before: "Experts believe it plays a crucial role."

After: "Researchers and conservationists study the Haolai River for its unusual characteristics."

I7 – Avoiding is/are/has

Avoiding is, are, and has detects circumlocutions like "serves as," "functions as," or "boasts" that replace simple copular verbs. This pattern promotes direct, economical prose by restoring basic verb structures.

As shown in [SKILL.md lines 64-71](https://github.com/blader/humanizer/blob/main/SKILL.md#L64):

Before: "Gallery 825 serves as LAAA's exhibition space ... features four separate spaces and boasts over 3,000 sq ft."

After: "Gallery 825 is LAAA's exhibition space ... has four rooms totaling 3,000 sq ft."

Implementation and Validation

The Humanizer package structure enforces pattern integrity through [scripts/validate-package.py](https://github.com/blader/humanizer/blob/main/scripts/validate-package.py), which checks that entries 12-18 maintain consistent metadata and numbering schemes. The agent configuration in [agents/openai.yaml](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) points to SKILL.md as the canonical source for these definitions, ensuring that implementations across different interfaces reference the same detection criteria.

Summary

• Inflated Significance (I2) removes unsupported "pivotal" or "crucial" framing from ordinary facts.
• Vague Connection (I3) demands specificity instead of phrases like "associated with" or "linked to."
• Shallow -ing Riders (I4) eliminates participial phrases that add no substantive information.
• Sales Language (I5) strips promotional adjectives that turn facts into marketing copy.
• Borrowed Authority (I6) requires named sources rather than appeals to unnamed experts.
• Avoiding is/are/has (I7) replaces verbose phrases like "serves as" with direct copular verbs.

Frequently Asked Questions

What distinguishes Inflated Significance (I2) from Sales Language (I5)?

Inflated Significance specifically targets claims about historical or conceptual importance ("pivotal moment," "lasting legacy"), while Sales Language targets promotional descriptors ("vibrant," "breathtaking," "groundbreaking"). The former overstates importance; the latter overstates quality or appeal.

How does the validator ensure pattern consistency in Humanizer?

The scripts/validate-package.py script checks that SKILL.md entries maintain sequential numbering, that each pattern includes the required "What to watch for" and "Core problem" fields, and that the agent metadata in agents/openai.yaml correctly references the skill definition file.

Why does the Inflation category list patterns 12-18 when I8 is undefined?

The numbering in SKILL.md reserves entry 18 as the terminal marker for the Inflation and Borrowed Authority group. Patterns 12-17 map to I2-I7, while entry 18 signifies the section end, creating a consistent indexing scheme across the seven primary categories documented in the repository.

Can these patterns be applied to non-English text?

The current implementation in blader/humanizer targets English linguistic patterns—particularly English promotional rhetoric and copular verb avoidance. Extending detection to other languages would require adding localized pattern definitions to SKILL.md and updating the validation logic to handle multilingual metadata structures.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}