Rulebook

38 tells, and which way each one points

Each rule is a habit that model writing and people's writing show at different rates. Most point toward a model, a few point the other way, and none decides anything alone. The meter weighs the 36 it can measure together with 140 measures of rhythm and word choice, using weights learned from data.

r-001 · Words

Stock vocabulary

Words that machine text over-uses relative to human prose (delve, tapestry, pivotal, intricate, multifaceted, foster, bolster, garner, leverage, seamless, robust, nuanced, meticulous, embark, underscore, showcase).

Leans machine
In 7.7% of machine paragraphs, 3.2% of human

With the tell
We delve into the intricate tapestry of pivotal ideas that foster robust growth.
Without it
We looked at the ideas that helped the town grow.

FixUse the plain word, or cut it; most of these carry no meaning in context.

r-002 · Words

Copula avoidance

Fancy substitutes for "is" and "has" ("serves as", "stands as", "acts as", "boasts", "features a").

About even
In 1.7% of machine paragraphs, 1.1% of human

With the tell
The library serves as a hub and boasts a large collection.
Without it
The library is a hub and has a large collection.

FixWrite "is" or "has".

r-003 · Words

Filler phrases

Phrases that announce content instead of stating it ("it's important to note", "it is worth noting", "in today's fast-paced world", "when it comes to", "at the end of the day").

About even
In 1.6% of machine paragraphs, 2.4% of human

With the tell
It's important to note that, when it comes to sleep, in order to rest you need quiet.
Without it
You need quiet to rest.

FixDelete the phrase and start at the point.

r-004 · Words

Chatbot artifacts

Assistant openers and sign-offs left in the text ("Certainly!", "Great question", "I hope this helps", "let me know if", "feel free to").

Leans machine
In 1.4% of machine paragraphs, 0.7% of human

With the tell
Certainly! Here are some tips. I hope this helps, and let me know if you need more.
Without it
Here are the tips I use.

FixDelete them.

r-005 · Words

Promotional adjectives

Brochure words (breathtaking, stunning, nestled, vibrant, renowned, world-class, unparalleled, cutting-edge, must-visit).

Leans machine
In 3.2% of machine paragraphs, 1.3% of human

With the tell
Nestled in the hills, this breathtaking, world-class resort is a must-visit.
Without it
The hotel is in the hills and costs $90 a night.

FixReplace with a concrete detail, or drop the adjective.

r-006 · Words

Abstract metaphor nouns

Figurative nouns doing no work (tapestry, landscape, realm, journey, symphony, beacon, cornerstone, testament, mosaic, fabric).

Leans machine
In 2.4% of machine paragraphs, 0.9% of human

With the tell
The rich tapestry of the culinary landscape is a testament to the city.
Without it
The city has good food.

FixName the actual thing.

r-007 · Words

Stacked transitions

Formal connectives at sentence starts (Moreover, Furthermore, Additionally, Consequently, Notably, Importantly, Ultimately).

About even
In 1.0% of machine paragraphs, 1.3% of human

With the tell
It rained. Moreover, it was cold. Furthermore, the bus was late. Additionally, I lost my keys.
Without it
It rained. It was cold. The bus was late and I lost my keys.

FixDrop the connective; if the logic needs it, use "and", "but", "so".

r-008 · Words

Knowledge-cutoff disclaimers

Model hedges about its own knowledge ("as of my last update", "while specific details are limited", "I don't have access to").

Leans machine
In 0.5% of machine paragraphs, 0.0% of human

With the tell
As of my last update, the company had 40 staff, while specific details are limited.
Without it
The company had 40 staff in 2019.

FixDelete, and either find the fact or leave it out.

r-009 · Punctuation

Em-dash density

Em dashes (or spaced double hyphens) per sentence well above the human baseline.

Leans machine
In 11.1% of machine paragraphs, 4.4% of human

With the tell
The plan — bold as it was — failed. The team — tired — went home. The end — finally.
Without it
The plan failed, and the team went home tired.

FixUse a comma, a period, or parentheses; keep at most one dash per paragraph.

r-010 · Punctuation

Colon as connector

Mid-sentence colons that set up a reveal ("The answer is simple: consistency.").

About even
In 4.4% of machine paragraphs, 4.4% of human

With the tell
The answer is simple: consistency beats talent every time you practice.
Without it
Consistency beats talent.

FixMake it two sentences or state it directly.

r-011 · Punctuation

Bold-label lists

List items that start with a bolded label and a colon ("**Speed:** …").

Leans machine
In 1.3% of machine paragraphs, 0.0% of human

With the tell
- **Speed:** it is fast - **Cost:** it is cheap - **Scale:** it grows
Without it
- it is fast - it is cheap - it grows

FixWrite sentences, or a plain list without labels.

r-012 · Punctuation

Title-case headings

Short heading lines with Every Major Word Capitalized.

About even
In 0.7% of machine paragraphs, 0.5% of human

With the tell
Why Remote Work Changes Everything Forever Some people disagree about this, and they have reasons.
Without it
Why remote work changes things Some people disagree about this, and they have reasons.

FixSentence case.

r-013 · Punctuation

Decorative emoji

Emoji used as bullets or section decoration (🚀 ✨ 💡 ✅ 👉).

Leans human
In 0.3% of machine paragraphs, 0.4% of human

With the tell
🚀 Launching today! ✨ New features. 💡 Big ideas.
Without it
Launching today with new features.

FixRemove them.

r-014 · Structure

Negative parallelism

Sentence pairs in a "not just X, it's Y" shape ("It's not just a tool, it's a movement", "not only ... but also").

Leans machine
In 1.4% of machine paragraphs, 0.7% of human

With the tell
It's not just a tool, it's a movement. This is not only fast but also cheap.
Without it
The tool is fast and cheap.

FixState the point directly.

r-015 · Punctuation

Quote style

Typographic curly quotes and apostrophes in contexts where people type straight ones.

Leans machine
In 19.1% of machine paragraphs, 13.2% of human

With the tell
“Hello,” she said. “It’s late.”
Without it
"Hello," she said. "It's late."

FixNone needed for meaning; this is a provenance signal, not a style error.

r-016 · Punctuation

List density

The share of lines that are bullets or numbered items.

Leans machine
In 5.7% of machine paragraphs, 0.6% of human

With the tell
Steps: - one - two - three
Without it
We did one thing, then two, then three.

FixFold short lists back into sentences.

r-017 · Rates

Uniform sentence length

Low variance in sentence length (low burstiness); human prose mixes very short and very long sentences.

About even
Measured on every paragraph

With the tell
The system works well in most cases. The team built it over many months. The users like it a lot now. The results are good for all of us.
Without it
No. The system, which the team built over many long and difficult months in a cramped office above a bakery, mostly works. Users like it.

FixVary the rhythm; let one sentence run, cut another to three words.

r-018 · Rates

Adverb density

More -ly adverbs per 100 words than human prose uses.

About even
Measured on every paragraph

With the tell
She quickly and carefully and quietly and deliberately closed the door.
Without it
She closed the door.

FixPick a stronger verb or delete the adverb.

r-019 · Rates

Passive ratio

The share of sentences in the passive voice (a form of "be" plus a past participle).

About even
Measured on every paragraph

With the tell
The report was written by the team. The data was collected in May. Errors were found.
Without it
The team wrote the report. We collected data in May. We found errors.

FixName who did it.

r-020 · Rates

Hedging stacks

Two or more hedges in one sentence (may, might, could, potentially, arguably, generally, often, typically, somewhat).

About even
Measured on every paragraph

With the tell
This may potentially help and could possibly improve results. It might generally work.
Without it
This helps. It improved results by 4%.

FixCommit to the claim, or say exactly what is uncertain.

r-021 · Rates

Groups of three

More "X, Y, and Z" triads per sentence than human prose uses.

Leans machine
Measured on every paragraph

With the tell
It is fast, cheap, and reliable. It is simple, clear, and useful. It is small, light, and strong.
Without it
It is fast. It costs less than the old one.

FixKeep the one or two items that matter.

r-022 · Rates

Lexical diversity

Words that rarely repeat for the paragraph's length (a high moving type-token ratio), a side effect of synonym cycling.

Leans machine
Measured on every paragraph

No one sentence shows this. It's a rate over the whole paragraph, so it only shows up in text long enough to count.

FixRepeat the right word instead of rotating synonyms.

r-023 · Rates

Long-word share

The share of words with 10 or more letters.

Leans machine
Measured on every paragraph

With the tell
Unquestionably, the extraordinary organizational transformation necessitates comprehensive implementation strategies.
Without it
The big change needs a plan we can follow.

FixPrefer the short word when it says the same thing.

r-024 · Structure

Recap openers

Paragraphs that open by summarizing ("In summary", "Overall", "In conclusion", "As mentioned", "To sum up").

About even
In 0.1% of machine paragraphs, 0.0% of human

With the tell
In conclusion, the plan worked and the team was happy with how it went overall.
Without it
The plan worked and the team was happy with how it went.

FixCut the recap; start with the new point.

r-025 · Structure

Stop-start fragments

Runs of very short verbless fragments ("No fluff. No filler. Just results.").

About even
In 3.3% of machine paragraphs, 3.3% of human

With the tell
No fluff. No filler. Just results. That is what we promise every single customer who calls.
Without it
We promise every customer results without any filler.

FixJoin them into one sentence that says the thing.

r-026 · Structure

False ranges

A "from X to Y" phrase where X and Y are not ends of a scale ("from ancient rituals to modern cuisine").

About even
In 2.0% of machine paragraphs, 2.2% of human

With the tell
The festival covers everything from ancient rituals to modern cuisine.
Without it
The festival runs from 9am to 5pm.

FixList the specific things, or pick one.

r-027 · Structure

Openers with no antecedent

Paragraphs that open by pointing at nothing ("This", "These", "Here's the thing", "That's why", "The result?").

About even
In 3.9% of machine paragraphs, 4.0% of human

With the tell
This is why the approach matters so much for teams who ship every week.
Without it
Weekly releases make the approach matter for teams.

FixName the subject.

r-028 · Structure

Participle tail clauses

Sentences ending in ", highlighting/underscoring/reflecting/showcasing/emphasizing …" analysis tails.

Leans machine
In 2.7% of machine paragraphs, 0.6% of human

With the tell
The city grew fast, highlighting its importance to the region.
Without it
The city grew fast.

FixDelete the tail, or make its claim a sentence with evidence.

r-029 · Structure

Staged reveal questions

A short rhetorical question answered immediately ("The result? …", "Why does this matter? Because …").

About even
In 0.6% of machine paragraphs, 0.9% of human

With the tell
We tried everything for weeks. The result? A faster app for everyone who uses it.
Without it
We tried everything for weeks and the app got faster.

FixState the answer.

r-030 · Structure

Repeated sentence openers

Many sentences in a paragraph starting with the same word.

About even
In 13.9% of machine paragraphs, 17.5% of human

With the tell
The app is fast. The app is cheap. The app is simple. The app is new.
Without it
The app is fast. It costs little. Setup takes a minute. Version two ships soon.

FixVary the subject or merge sentences.

r-031 · Meaning

Significance inflation

Claims of importance with no specifics ("plays a pivotal role", "stands as a testament", "marks a significant shift", "enduring legacy").

Leans machine
In 0.5% of machine paragraphs, 0.1% of human

With the tell
The bridge plays a pivotal role and stands as a testament to its enduring legacy.
Without it
The bridge carries 40,000 cars a day.

FixSay what happened and let the reader judge.

r-032 · Meaning

Vague attribution

Unnamed authorities ("experts say", "studies show", "critics argue", "observers have noted", "it is widely believed").

About even
In 0.3% of machine paragraphs, 0.1% of human

With the tell
Experts say the diet works, and studies show it helps.
Without it
A 2019 trial of 400 adults found the diet cut weight by 2 kg.

FixName the source, or drop the claim.

r-033 · Meaning

Challenges-and-prospects arc

Formulaic "despite its challenges … continues to thrive" framing.

About even
In 0.1% of machine paragraphs, 0.0% of human

With the tell
Despite these challenges, the town continues to thrive.
Without it
The mill closed in 1987 and the town lost a third of its jobs.

FixDescribe the specific problem and its specific outcome.

r-034 · Meaning

Absence of voice

Paragraphs long enough to expect first person, contractions or opinion markers, with none of them.

About even
In 42.6% of machine paragraphs, 44.7% of human

With the tell
The organization provides services to customers in many regions. The services include support, training, and consulting. The organization was founded to address needs in the market and has expanded over time into several new areas of operation across the country.
Without it
I think the company's support is good. We've used it for years, and honestly I'd pick it again over the others we tried, even though it's pricier than I'd like.

FixSay what you think and why; write the way you talk.

r-035 · Meaning

Generic upbeat conclusion

Closing lines about a bright future or exciting possibilities ("the future looks bright", "exciting times ahead", "the possibilities are endless").

About even
In 0.1% of machine paragraphs, 0.1% of human

With the tell
The future looks bright, and the possibilities are endless.
Without it
The next release is due in March.

FixEnd on the last concrete point.

r-036 · Meaning

Synonym cycling

The same referent renamed each sentence (the city, the metropolis, the urban center).

Not scored: the meter doesn't use this rule

The meter has no detector for this one, so there's nothing to mark. Watch for it yourself.

FixPick one name and repeat it.

r-037 · Meaning

Uniform register

An even, polite, mid-formal tone with no shifts, jokes, or asides.

Not scored: the meter doesn't use this rule

The meter has no detector for this one, so there's nothing to mark. Watch for it yourself.

FixLet the register move with the content.

r-038 · Meaning

Both-sides wrap-up

Closing balance that avoids a position ("ultimately, it depends", "there is no one-size-fits-all", "both have their merits").

About even
In 0.0% of machine paragraphs, 0.0% of human

With the tell
Ultimately, it depends on your needs; there is no one-size-fits-all answer.
Without it
Pick Postgres unless you need offline sync.

FixTake a position, or say what it depends on.