diff --git a/README.md b/README.md index 120bec1c..45f66b03 100644 --- a/README.md +++ b/README.md @@ -88,7 +88,7 @@ The skill also includes a final "obviously AI generated" audit pass and a second > "LLMs use statistical algorithms to guess what should come next. The result tends toward the most statistically likely result that applies to the widest variety of cases." -## 29 Patterns Detected (with Before/After Examples) +## 31 Patterns Detected (with Before/After Examples) ### Content Patterns @@ -107,7 +107,7 @@ The skill also includes a final "obviously AI generated" audit pass and a second |---|---------|--------|-------| | 7 | **AI vocabulary** | "Actually... additionally... testament... landscape... showcasing" | "also... remain common" | | 8 | **Copula avoidance** | "serves as... features... boasts" | "is... has" | -| 9 | **Negative parallelisms / tailing negations** | "It's not just X, it's Y", "..., no guessing" | State the point directly | +| 9 | **Negative parallelisms / tailing negations / "rather than" dismissals** | "It's not just X, it's Y", "..., no guessing", "X rather than Y (where Y is unstated)" | State the point directly; cut dismissed alternatives nobody claimed | | 10 | **Rule of three** | "innovation, inspiration, and insights" | Use natural number of items | | 11 | **Synonym cycling** | "protagonist... main character... central figure... hero" | "protagonist" (repeat when clearest) | | 12 | **False ranges** | "from the Big Bang to dark matter" | List topics directly | @@ -117,7 +117,7 @@ The skill also includes a final "obviously AI generated" audit pass and a second | # | Pattern | Before | After | |---|---------|--------|-------| -| 14 | **Em dash overuse** | "institutions—not the people—yet this continues—" | Prefer commas or periods | +| 14 | **Em dash overuse / paired bracketing** | "institutions—not the people—yet this continues—", "report—covering three continents—concluded" | Prefer commas or periods; break paired brackets into separate sentences or appositives | | 15 | **Boldface overuse** | "**OKRs**, **KPIs**, **BMC**" | "OKRs, KPIs, BMC" | | 16 | **Inline-header lists** | "**Performance:** Performance improved" | Convert to prose | | 17 | **Title Case Headings** | "Strategic Negotiations And Partnerships" | "Strategic negotiations and partnerships" | @@ -127,6 +127,8 @@ The skill also includes a final "obviously AI generated" audit pass and a second | 27 | **Persuasive authority tropes** | "At its core, what matters is..." | State the point directly | | 28 | **Signposting announcements** | "Let's dive in", "Here's what you need to know" | Start with the content | | 29 | **Fragmented headers** | "## Performance" + "Speed matters." | Let the heading do the work | +| 30 | **Conditional frame stacking** | "If the argument holds, and if the reading is right, then perhaps..." | State the conclusion; reserve "if" for real analytical branches | +| 31 | **Miscalibrated epistemic confidence** | Over: "decisively demonstrates fundamentally"; Over-hedge: "appears to have arguably may have somewhat" | Narrow the claim to what the evidence supports; don't replace over-assertion with hedges | ### Communication Patterns @@ -179,6 +181,7 @@ The skill also includes a final "obviously AI generated" audit pass and a second ## Version History +- **2.6.0** - Added patterns 30 (conditional frame stacking) and 31 (miscalibrated epistemic confidence); expanded rule 9 to cover "rather than" dismissals; expanded rule 14 to cover paired em dash bracketing; raising the total to 31 patterns - **2.5.1** - Added a passive-voice / subjectless-fragment rule, raising the total to 29 patterns - **2.5.0** - Added patterns for persuasive framing, signposting, and fragmented headers; expanded negative parallelisms to cover tailing negations; tightened wording around em dash overuse; fixed frontmatter wording to use "filler phrases" - **2.4.0** - Added voice calibration: match the user's personal writing style from samples diff --git a/SKILL.md b/SKILL.md index 46639f02..8df2e9d2 100644 --- a/SKILL.md +++ b/SKILL.md @@ -1,13 +1,14 @@ --- name: humanizer -version: 2.5.1 +version: 2.6.0 description: | Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive - voice, negative parallelisms, and filler phrases. + voice, negative parallelisms, filler phrases, conditional frame stacking, and + miscalibrated epistemic confidence. license: MIT compatibility: claude-code opencode allowed-tools: @@ -197,9 +198,9 @@ Avoiding AI patterns is only half the job. Sterile, voiceless writing is just as > Gallery 825 is LAAA's exhibition space for contemporary art. The gallery has four rooms totaling 3,000 square feet. -### 9. Negative Parallelisms and Tailing Negations +### 9. Negative Parallelisms, Tailing Negations, and "Rather Than" Dismissals -**Problem:** Constructions like "Not only...but..." or "It's not just about..., it's..." are overused. So are clipped tailing-negation fragments such as "no guessing" or "no wasted motion" tacked onto the end of a sentence instead of written as a real clause. +**Problem:** Constructions like "Not only...but..." or "It's not just about..., it's..." are overused. So are clipped tailing-negation fragments such as "no guessing" or "no wasted motion" tacked onto the end of a sentence instead of written as a real clause. A third form of the same pattern is "rather than" used to stage a contrast by dismissing an alternative that nobody was claiming in the first place. **Before:** > It's not just about the beat riding under the vocals; it's part of the aggression and atmosphere. It's not merely a song, it's a statement. @@ -213,6 +214,14 @@ Avoiding AI patterns is only half the job. Sterile, voiceless writing is just as **After:** > The options come from the selected item without forcing the user to guess. +**Before ("rather than" dismissal):** +> The goal is to write clearly rather than to impress the reader with complexity. + +**After:** +> The goal is to write clearly. + +**Test:** Ask whether the discarded alternative (Y in "X rather than Y") is actually on the table. If no one was claiming Y, cut the dismissal and just say X. + ### 10. Rule of Three Overuse @@ -260,16 +269,29 @@ Avoiding AI patterns is only half the job. Sterile, voiceless writing is just as ## STYLE PATTERNS -### 14. Em Dash Overuse +### 14. Em Dash Overuse and Paired Bracketing **Problem:** LLMs use em dashes (—) more than humans, mimicking "punchy" sales writing. In practice, most of these can be rewritten more cleanly with commas, periods, or parentheses. +A specific sub-pattern is paired em dash bracketing: wrapping an elaboration between two dashes (X — elaboration — continues). This looks inserted rather than written — like something dropped into an existing sentence rather than composed as part of it. + **Before:** > The term is primarily promoted by Dutch institutions—not by the people themselves. You don't say "Netherlands, Europe" as an address—yet this mislabeling continues—even in official documents. **After:** > The term is primarily promoted by Dutch institutions, not by the people themselves. You don't say "Netherlands, Europe" as an address, yet this mislabeling continues in official documents. +**Before (paired bracketing):** +> The report—which covered three continents and twelve case studies—concluded that demand had shifted. + +**After options (depending on type of insertion):** +- If a list: break into a separate sentence. "The report covered three continents and twelve case studies. It concluded that demand had shifted." +- If an appositive: use a comma or recast. "The report, covering three continents and twelve case studies, concluded that demand had shifted." +- If a parenthetical aside: use parentheses if truly aside, or restructure. "The report (three continents, twelve case studies) concluded that demand had shifted." +- If subject expansion: rewrite as two sentences. + +**Exception:** A single, short, earned bracket that does not repeat elsewhere in the passage is fine. The problem is the pattern, not any one instance. + ### 15. Overuse of Boldface @@ -461,6 +483,43 @@ Avoiding AI patterns is only half the job. Sterile, voiceless writing is just as > > When users hit a slow page, they leave. + +### 30. Conditional Frame Stacking + +**Problem:** AI hedges its own conclusions by stacking multiple "if" clauses in the same passage — "if the argument holds," "if the reading is right," "if this interpretation is correct." One conditional at a genuine analytical branching point is fine. A cluster of them in a conclusion or summary signals the writer is not standing behind their own work. + +**Before:** +> If the argument holds, and if the evidence supports this reading, then the policy may have had some effect — if, that is, the context was as described. + +**After:** +> The evidence supports the argument that the policy had an effect in this context. + +**Fix:** In a conclusion or summary, state what the argument found. Reserve "if" for real analytical branches where the outcome genuinely differs depending on the condition — not as a repeated hedge against being wrong. + + +### 31. Miscalibrated Epistemic Confidence + +**Problem:** A two-sided pattern. AI swings between over-asserting and over-hedging, sometimes in the same passage. + +- **Over-assertion:** Loading claims with words like "decisively," "fundamentally," "completely," "unquestionably," "clearly demonstrates" when the evidence is more limited. +- **Over-hedging:** Layering qualifiers such as "appears to have arguably," "may have somewhat," "could potentially suggest" when the evidence actually supports a more direct statement. + +Both are tells. The fix is not to replace one extreme with the other — it is to narrow the claim to what the evidence actually supports. + +**Before (over-assertion):** +> The data decisively demonstrates that remote work fundamentally transformed productivity across all sectors. + +**After:** +> In the surveyed companies, productivity rose an average of 8% in the first year of remote work. + +**Before (over-hedging):** +> It appears that the policy may have arguably had some effect on outcomes, potentially suggesting a modest shift. + +**After:** +> The policy was associated with a modest improvement in outcomes in two of the three cases studied. + +**Critical rule:** Do not fix over-assertion by adding hedges. Fix it by narrowing the claim. + --- ## Process