# The Human Writing Protocol version: 1.2.0 updated: 2026-09-26 maintainer: Spectrum Flare canonical: https://writelikehuman.app languages: en, pl A machine-readable ruleset for writing text that does not read as AI-generated. Paste the canonical URL into any AI agent. The agent fetches this document and applies every rule below to the draft it is producing. **Writing in Polish?** Sections 2, 3 and 4 are English-specific. Use section 11 in their place — it carries the Polish banlists, Polish typography, and the declension check. Everything else applies unchanged. Short link for Polish-only work: `https://writelikehuman.app/rules.md?lang=pl` --- ## 0. Operating instruction You are writing on behalf of a human. Apply this protocol to every sentence you produce. 0. Identify the output language. English → sections 1-4. Polish → sections 1 and 11. Another language → sections 1, 5, 6, 7, 8 only, and say so rather than applying English lexicon rules to text that is not English. 1. Write the draft. 2. Run the SCRUB pass (sections 1-4, or 1 and 11 for Polish). Remove what is banned. 3. Run the MEASURE pass (section 5). Fix any metric outside its target range. 4. Run the SUBSTANCE pass (section 6). Every paragraph must survive the deletion test. 5. Run the self-check (section 8) before returning the text. Match the job to what the user asked for: - **Write** (the default). Produce the text they asked for, with every pass above applied. - **Rewrite** ("humanise this", "fix this", pasted text). Keep the meaning and the author's quirks; change the style only. Return the rewritten text, then a short list of the rules that fired only if they ask for it. - **Audit** ("check this", "audit", "what's wrong with this"). Do not rewrite. Return a list: the rule (section number), the exact quote, the fix. Most serious first. End with the section 5 metrics you can compute. - **Voice** ("learn my style", pasted samples). Build the voice profile from section 12 and use it for everything you write after that. Two hard constraints that override everything else: - **Never invent facts to satisfy this protocol.** If a rule asks for a number, a name, or a date and you do not have one, ask the user for it. A fabricated specific is worse than a generic sentence. - **Never change the author's meaning.** Humanising is a style operation, not an editorial one. The goal is not to defeat a detector. AI detectors are unreliable in both directions: a Stanford 2023 study found 61.3% false-positive rates on non-native-speaker TOEFL essays, and current research shows perplexity and burstiness gaps between large models and humans are closing. The goal is prose a careful reader would not flag as machine-written, because it carries specifics, rhythm, and a point of view. --- ## 1. Forensic leakage — always strip These never appear in human writing. There is no defence for any of them. Delete on sight. | Pattern | Action | |---|---| | `oaicite`, `contentReference`, `turn0search0`, `:contentReference[...]`, `grok_card`, `attached_file` | delete the token | | "As of my last update", "As of my knowledge cutoff", "As of [date], I don't have access to" | delete the whole sentence | | "I'm sorry, but", "I cannot", "As an AI language model" | delete the whole sentence | | Unfilled template slots: `[Your Name]`, `[Company]`, `[Insert X here]`, `[Date]` | stop and ask the user to fill them | | Two or more consecutive bracketed blanks | stop and ask | | A closing offer the user did not ask for: "Let me know if you'd like me to expand on any of these" | delete | | Meta-narration of your own process: "Here's a rewritten version", "I've structured this as", "Certainly! Here is" | delete | | Chatbot tracking parameters in links: `utm_source=chatgpt.com`, `utm_source=openai` | strip the parameter | | Markdown emitted into a plain-text channel (`**bold**` in an email body, `###` in a LinkedIn post) | convert to the channel's native formatting | --- ## 2. Structural and rhetorical tells ### 2.1 Negative parallelism — hard ban All six forms. This is the single loudest tell in 2026 short-form writing. ``` "It's not just X, it's Y" → banned "X isn't Y, it's Z" → banned "Not X, but Y" → banned "It's not about X, it's about Y" → banned "The question isn't X, it's Y" → banned "This isn't X. This is Y." → banned ``` Fix: state the claim directly, or split into two plain declaratives. If both halves matter, write them as two sentences without the inversion. ### 2.2 Rule of three on autopilot Triplet adjectives ("efficient, scalable, and reliable") and triplet clauses ("plan, execute, and optimise"). Threshold: **no more than one polished triplet per 200 words.** Break the rest into two items or four. Uneven lists read as human because a human stops when they run out of real examples. ### 2.3 Empty openers "In today's fast-paced world", "In the digital age", "In a world where", "Picture this", "Let's be honest", "Here's the thing". Threshold: **zero.** A human editor deletes the first one on sight. Open on the actual claim. ### 2.4 Fake authority "Studies have shown", "Experts agree", "Research suggests", "The data speaks for itself", "It is widely known that". Rule: every authority claim carries a name, a number, a date, or a link, or it gets deleted. Do not attach a citation you cannot verify. ### 2.5 Pseudo-wisdom filler "The key is finding balance", "True growth comes from within", "It's not about the destination". Test: delete the sentence. If the paragraph loses no information, it was filler. **More than one third of your sentences must not survive deletion.** ### 2.6 Signposting and transition stacking "Furthermore", "Moreover", "Additionally", "Ultimately", "In conclusion", "It's worth noting that", "Importantly", "Crucially", "Notably", "Essentially", "Fundamentally". Threshold: **fewer than half of paragraphs may open with a formal transition.** In short-form (under 400 words), use zero. Let sequence carry the logic. ### 2.7 The outline-formula closer "Despite its X, the company faces Y. Looking ahead, it will need to Z." "As the landscape continues to evolve, one thing is clear." Fix: end on the last real thing you have to say. Do not synthesise, do not gesture at the future, do not restate the opening. ### 2.8 Superficial -ing tails Present participles bolted onto a sentence end to add unearned significance: "…, highlighting the importance of collaboration", "…, underscoring the need for change", "…, reflecting a broader shift". Fix: cut the tail, or promote it to its own sentence with a real subject. ### 2.9 Sycophantic tone "Great question!", "You're absolutely right!", "That's a fantastic point." In writing produced for publication: **zero.** In conversational replies: at most one per exchange, and only when it carries information. ### 2.10 Copula avoidance Models dodge "is" and "has": "serves as", "stands as", "functions as", "acts as", "boasts", "features", "represents". Fix: "The library serves as a hub for the community" → "The library is where the town meets." Use "is" and "has" unless the fancier verb adds meaning. ### 2.11 Significance inflation and puffery Inflating the stakes: "marks a pivotal moment", "plays a vital role", "leaves an indelible mark", "an enduring legacy", "cannot be overstated", "a testament to". Brochure words: "nestled", "breathtaking", "rich heritage", "stunning", "world-class", "state-of-the-art", "unparalleled". Rule: say what happened and let the reader judge its size. If something really is important, the fact carries it: "the first bridge across the river since 1890" needs no "pivotal". ### 2.12 Elegant variation Cycling synonyms so no noun repeats: "the company", then "the firm", then "the organisation", then "the Seattle-based giant". Readers start to wonder whether these are four different things. Fix: pick one name and repeat it, or use a pronoun. Repetition is clearer than rotation. ### 2.13 False ranges "From X to Y" where X and Y are not two ends of any scale: "from boardrooms to bedrooms", "everything from pricing to culture", "ranging from startups to governments". Fix: name the things you actually mean, or drop the phrase. ### 2.14 Staged reveals and stacked fragments Setting up a punchline instead of stating the point: "The result? A 40% drop." "Here's the kicker:", "The best part?", "Spoiler:", "That's where X comes in.", "Enter X." Also the three-fragment drumroll: "No fluff. No filler. Just results." and the audience split "Whether you're a founder or a freelancer…". Threshold: **zero** in writing for publication. State the result in a normal sentence: "The error rate dropped 40%." --- ## 3. Lexicon ### 3.1 Style words — the post-2022 spike Kobak et al. analysed 14.2M PubMed abstracts (2010-2024) and found the post-ChatGPT vocabulary shift was "almost entirely style words". *Delves* rose to roughly 25x its pre-ChatGPT frequency; *showcasing* and *underscores* jumped about nine-fold. Threshold: **no more than 3 flagged words per 500 words, never clustered.** ``` delve, delving → look at, dig into tapestry → (delete — describe the actual thing) underscore → show showcase → show pivotal → important, or name the stakes testament (to) → (rewrite — usually "proves" or delete) intricate, intricacies → complex, the details multifaceted → (delete — name the facets) realm → area, field landscape → field, market, or delete ecosystem → (only if literally an ecosystem) paradigm → approach holistic → complete, or name what it covers nuanced → specific seamless → smooth, or delete robust → solid, reliable vibrant → (name the actual quality) garner → get foster → build cultivate → grow, build harness → use unlock → find, get, open elevate → improve, raise navigate → handle, deal with resonate → land, matter to ``` ### 3.2 Corporate verb inflation Latinate verbs standing in for plain ones. ``` utilize → use leverage → use facilitate → help streamline → simplify optimize → improve (unless mathematically optimising) implement → build, set up, do commence → start endeavour → try ``` Threshold: if an inflated verb appears where the plain one works, **more than once per 300 words**, that is inflation, not register. ### 3.3 Buzz phrases ``` game-changer → say what changed deep dive → look, breakdown move the needle → change the numbers paradigm shift → real shift, or name it low-hanging fruit → the easy part at the end of the day → (delete) in the age of AI → (delete) best-in-class → (delete or benchmark it) ``` ### 3.4 Empty intensifiers `very, really, incredibly, truly, absolutely, significantly, substantially, dramatically, massively` — delete, or replace with a number. "Sales grew significantly" → "Sales grew 34%." --- ## 4. Punctuation and formatting ### 4.1 Em dash The most argued-about tell, and the most misused rule. Measured human prose runs **3.7 to 10 dash constructions per 1,000 words**; Twain's *Huckleberry Finn* scores 10.13. A controlled study clocked GPT-4.1 at 10.62 per 1,000 against a 3.23 human baseline. Rule: **cap at 15 per 1,000 words, and never more than one per paragraph in short-form.** Do not ban the em dash. Banning it is itself a 2026 tell, because everyone who read one viral thread now avoids it uniformly. Never use a spaced em dash ( — ) in prose that otherwise uses tight punctuation; pick one convention and hold it. ### 4.2 Quotes and dashes Curly quotes and apostrophes (" " ' ') pasted into a plain-text channel are a copy-paste fingerprint. Convert to straight quotes when the destination is plain text. Double hyphen (`--`) standing in for a dash: convert to a period or a comma. ### 4.3 Emoji Never as bullet markers. Never as section dividers. The rocket / lightbulb / sparkles / check-mark signature set is a strong tell. Maximum one emoji per message, and only when it carries tone a word cannot. ### 4.4 Structure formatting - No bold lead-ins inside bullet lists ("**Speed:** the system is faster") unless the destination is documentation. - No title-case headings in body copy. - No bullet list where three sentences of prose would do. Models reach for lists because lists are safe. - No horizontal rules between every section in a short piece. - Bold is for the one phrase a skimming reader must not miss. More than one bold phrase per paragraph is decoration. - No headings that only hold other headings, and no skipped levels (an H2 followed by an H4). - No table for content that has one dimension. A table needs rows and columns that both mean something. --- ## 5. Measurable targets Compute these on the finished draft. Each has a number you can check. | Metric | How to compute | Target | |---|---|---| | **Burstiness** | stdev(sentence word counts) ÷ mean | **> 0.5** (human range 0.6-1.2; model output clusters 0.2-0.4) | | **Sentence-length SD** | standard deviation of words per sentence | **> 7 words** (GPT-4o ≈ 4.1, Claude ≈ 5.3, human academic ≈ 8.2) | | **Sentence-length range** | longest minus shortest | **> 30 words** (under 15 is a strong AI signal) | | **Em dash density** | em dashes per 1,000 words | **< 15** | | **Flagged style words** | section 3.1 hits per 500 words | **≤ 3**, never clustered | | **Triplets** | polished three-item parallel lists per 200 words | **≤ 1** | | **Paragraph-opening transitions** | share of paragraphs starting with a formal connector | **< 50%**, ideally 0 in short-form | | **Specificity** | concrete numbers, names, or dates per 100 words | **≥ 1** | | **Deletion survival** | share of sentences that can be deleted with no information loss | **< 33%** | If burstiness is under target, do not paraphrase. Paraphrasing preserves sentence length: a 20-word AI sentence comes back at 18-22 words. Restructure instead — merge two sentences into one long one, then split a third into a four-word sentence. **Convergence rule:** no single metric convicts a text. Three or more failing at once in the same short passage is the fingerprint. Fix the three loudest, not all nine. --- ## 6. Substance requirements This is the most reliable signal and the hardest to fake. You can strip every em dash, every "delve", and every triplet from an empty paragraph and it stays empty. Per 100 words, the text must carry at least: - **one specific number** — replace "many", "significant", "a lot of" with a figure - **one named entity** — a real person, company, product, city, or date - **one concrete noun** a reader can picture Per piece, at least one of: - a self-correction, a hedge that names its own limit, or an admission of what the writer does not know - a trade-off stated plainly, with the losing side named - a detail that is true but not flattering **The restatement test:** after each paragraph, name one concrete fact the reader now knows. If more than half the paragraphs fail, the piece is empty regardless of who wrote it. Compare: > Empty: "Nutrition plays a crucial role in overall wellness. By making mindful choices and understanding your body's needs, you can unlock a healthier lifestyle." > Substantive: "Swap the 6pm soda for water and you cut roughly 40,000 calories a year, about 11 pounds. That did more for my blood sugar than any app I tried." If the source material contains no specifics, **stop and ask the user for one number or one anecdote.** Do not invent one. --- ## 7. Do not overcorrect Over-scrubbing produces its own recognisable style: choppy, fragment-heavy, aggressively lowercase, and stripped of every subordinate clause. This is the 2026 tell that replaced the em dash. Things that are **not** evidence of AI writing and must not be removed on sight: - a single em dash used well - one rule of three in a piece of persuasive writing - correct grammar and spelling - the passive voice where the object matters more than the actor - long sentences, when the idea is genuinely long - the word "robust" in an engineering or epidemiological context Do not add deliberate typos. Do not force lowercase. Do not add fake hedging ("I think maybe") to sound uncertain. Do not pad with filler personal detail. Each of these is now its own tell. --- ## 8. Self-check before returning Run this as a checklist. Return the text only when every line passes. ``` [ ] Zero forensic artefacts (§1) [ ] Zero negative parallelism, in all six forms (§2.1) [ ] Zero empty openers (§2.3) [ ] Every authority claim carries a source, or is deleted (§2.4) [ ] Zero outline-formula closers (§2.7) [ ] Zero staged reveals and fragment drumrolls (§2.14) [ ] Style-word count ≤ 3 per 500 words (§3.1) [ ] Burstiness > 0.5 and sentence-length SD > 7 (§5) [ ] ≥ 1 specific number, name, or date per 100 words (§6) [ ] < 33% of sentences survive the deletion test (§6) [ ] Nothing invented to satisfy a rule [ ] Meaning unchanged from the source draft ``` If asked to show your work, return a short diff: which rule fired, what changed. Do not narrate the process inside the text itself. --- ## 9. Channel overrides The base protocol assumes short-form business writing. Adjust as follows. **LinkedIn / X / short social** Transitions: zero. Bullet lists: at most one, at most four items. Hook must open on the claim, not on context. No "What do you think?" closer, no "Tag someone who". Target burstiness 0.7+, because scroll rhythm punishes uniformity harder than a page does. **Email** No markdown. No headings. One ask per email, stated in the first two sentences. Sign-off matches the relationship, not the template. **Technical documentation** Section 5's burstiness and transition targets do not apply. Uniform sentence length is correct here. Sections 1, 2.4, 3.1, 3.2 and 6 still apply in full. **Proposals and client documents** Every claim carries a number or a named deliverable. Section 6's specificity target rises to 2 per 100 words. Never use "tailored", "bespoke", "cutting-edge" or "end-to-end" without an example immediately after. **Long-form article** Transition ceiling rises to 50% of paragraphs. Triplet ceiling stays. Substance test applies per section, not per paragraph. --- ## 10. Anti-patterns in the humanising itself Failure modes seen in agents applying rulesets like this one: 1. **Synonym swapping without restructuring.** Replacing "leverage" with "use" while keeping the same 20-word balanced clause changes nothing measurable. Fix the structure first, the vocabulary second. 2. **Fabricating specifics to hit section 6.** The single worst outcome. A made-up statistic is a real harm; a generic sentence is only a style problem. 3. **Scrubbing the author's voice quirks.** If the source writer uses lowercase starts or double-dot pauses, keep them. Those are the fingerprint you are trying to protect. 4. **Applying every rule at maximum strictness.** Sections 3.1 and 4.1 have legitimate uses. Aim for the thresholds, not for zero. 5. **Rewriting the claim.** If the humanised version says something the original did not, you failed, regardless of how it reads. --- ## 11. Polski — zasady dla tekstu pisanego po polsku Ta sekcja zastępuje sekcje 2, 3 i 4, które opisują angielszczyznę. Sekcje 0, 1, 5, 6, 7 i 8 obowiązują bez zmian. Polska lista zakazanych fraz jest inna niż angielska — nie jest jej tłumaczeniem. ### 11.1 Openery — zero ``` W dzisiejszych czasach… W dzisiejszym dynamicznie zmieniającym się świecie… W erze cyfrowej… W obliczu wyzwań współczesnego rynku… Nie sposób nie zauważyć, że… Czy zastanawiałeś się kiedyś nad… Wiele osób zadaje sobie pytanie… Chcę się z Tobą podzielić… ``` Zacznij od tego, co masz do powiedzenia. ### 11.2 Fałszywa autentyczność ``` Ostatnio rozmawiałem z klientem… Ostatnio coraz częściej słyszę, że… Często zauważam w swojej pracy… Moi klienci pytają mnie ostatnio o… W rozmowach z klientami pojawia się… Wiele osób zmaga się z… ``` Wyjątek: opener przechodzi, jeśli w następnym zdaniu pada konkret — rok, branża, liczba, nazwa. „Rozmawiałem z trenerką, która prowadzi szkolenia HR od 2018 roku" to historia. „Rozmawiałem z klientem o ważnych sprawach" to AI. ### 11.3 Wypełniacze — kasuj w całości ``` warto zauważyć / warto zaznaczyć / warto podkreślić należy zauważyć / należy podkreślić kluczowym aspektem jest w praktyce oznacza to, że co więcej / co ciekawe mając na uwadze powyższe nie sposób pominąć faktu, że trzeba pamiętać, że jest to niezwykle istotne ``` Próg dla łączników na początku akapitu (Ponadto, Dodatkowo, Co więcej, Jednakże, Niemniej jednak, Warto również): **mniej niż połowa akapitów.** W tekstach poniżej 400 słów — zero. ### 11.4 Zakończenia — zero ``` Podsumowując… / Reasumując… / Konkludując… Na zakończenie chciałbym… Mam nadzieję, że ten artykuł… Zachęcam do refleksji nad… ``` Kończ na ostatniej rzeczy, którą naprawdę masz do powiedzenia. ### 11.5 Konstrukcje — zakaz twardy ``` To nie X, to Y. → zakaz To nie tylko X — to Y. → zakaz Nie chodzi o X, chodzi o Y. → zakaz nie tylko…, ale również/także → zakaz jako ozdobnik Bez ściemy. Bez bullshitu. Tylko konkret. → zakaz (trzy urwane zdania) Z jednej strony X, z drugiej Y. → zakaz jako sposób na uniknięcie stanowiska Otwiera drzwi do nowych możliwości. / Stanowi fundament każdego biznesu. → zakaz ``` Anafora ×3 (trzy zdania pod rząd zaczynające się tak samo) — zakaz. Triady („szybko, skutecznie i efektywnie", „innowacyjny, kompleksowy i skalowalny") — maksimum **jedna na 200 słów.** ### 11.6 Przymiotniki bez pokrycia — najsilniejszy polski sygnał ``` kompleksowy, dynamiczny, innowacyjny, holistyczny, wielowymiarowy, kluczowy, istotny, niezwykle ważny, skuteczny, dedykowany, rewolucyjny, przełomowy, transformacyjny, unikalny, nowoczesny, profesjonalny, indywidualne podejście, szeroki wachlarz, kompleksowe rozwiązanie, najwyższa jakość ``` Trzy lub cztery z tej listy w jednym akapicie to praktycznie pewność, że pisała AI. Próg: **maksimum jeden na 150 słów, i tylko z liczbą albo dowodem obok.** „Innowacyjny kurs" to AI. „Pierwszy kurs w Polsce łączący X z metodyką Y" to konkret. ### 11.7 Kalki z angielskiego ``` dedykowany (dedicated to) → dla, przeznaczony dla, przypisany do zaadresować problem → zająć się, rozwiązać dostarczać wartość → dawać coś konkretnego (nazwij co) na koniec dnia → (skasuj) w kontekście / w ramach / z perspektywy → nadużywane jako klej, sprawdź czy niesie treść wspierać proces → (nazwij, co robi) ``` Kalki składniowe i typograficzne: - **Title case w nagłówkach** („Jak Rozpoznać Tekst AI") — po polsku wielka litera tylko w pierwszym słowie i nazwach własnych. Każde odstępstwo to import z angielskiego. - **Przecinek po okoliczniku na początku zdania** („Dodatkowo, warto…", „Ponadto, system…") — po polsku bez przecinka. ### 11.8 Czasowniki napuszone ``` posiadać → mieć realizować → robić, prowadzić dokonać zakupu → kupić poprzez → przez użytkować → używać stanowić → być umożliwiać → pozwalać celem zwiększenia → żeby zwiększyć w celu… → żeby… ``` ### 11.9 Ogony imiesłowowe ``` …, co przekłada się na… …, co pozwala na… …, co sprawia, że… …, podkreślając znaczenie… ``` Utnij albo zrób z tego osobne zdanie z prawdziwym podmiotem. ### 11.10 Coachingowy bełkot — zero ``` Uwolnij swój wewnętrzny potencjał Odkryj autentyczną wersję siebie Wyrusz w podróż transformacji Przepracuj limitujące przekonania Odblokuj ukryty potencjał Turbodoładuj swój biznes ``` ### 11.11 Typografia — polska, nie angielska **Myślnik.** Po polsku myślnik jest oddzielony spacjami z obu stron: `słowo – słowo`. Standardem w prozie jest półpauza (–). Długi em dash bez spacji (`słowo—słowo`) to import z angielskiego i jeden z najgłośniejszych sygnałów w polskim tekście — nie dlatego, że jest zakazany, tylko dlatego, że nikt tak po polsku nie pisze. Limit: **poniżej 10 na 1000 słów, maksimum jeden na akapit.** Nie kasuj wszystkich; brak jakiejkolwiek pauzy w dłuższym tekście też jest nienaturalny. **Cudzysłów.** Polski cudzysłów to `„tekst"` (otwierający na dole, zamykający u góry). Angielski `"tekst"` albo `"tekst"` w polskim tekście to odcisk palca kopiuj-wklej. Cytat wewnątrz cytatu: `«tekst»`. **Reszta:** - Wielokropek to jeden znak `…`, nie trzy kropki. - Skróty z kropkami: np., itp., m.in., tzn., ok. - Nie zaczynaj zdania od cyfry. - Bez emoji jako punktorów i bez emoji w nagłówkach. ### 11.12 Fleksja — sprawdź odmianę Modele trenowane głównie na angielskim zostawiają w polskim zdaniu nieodmienione nazwy własne i zapożyczenia: „w wyniku atak dronów", „podczas zamach na Donald Trump", „po śmierci Łukasz Litewka". To sygnał widoczny nawet dla kogoś, kto nigdy nie używał LLM-a. Sprawdź w każdym tekście: - przypadek każdego nazwiska, nazwy firmy, miasta i produktu, - liczebniki („dwóch klientów", nie „dwa klienci"), - rekcję czasownika („używać czegoś", nie „używać coś"), - zgodność rodzaju w zdaniach złożonych. ### 11.13 Forma adresu — jedna na cały tekst Wybierz jedną: „ty" (w liście Ty/Twój wielką literą, w artykule małą), „Państwo", albo forma bezosobowa. Modele mieszają je w obrębie jednego tekstu — przechodzą ze „Sprawdź" na „Zachęcamy Państwa do zapoznania się" w sąsiednich akapitach. To sygnał sam w sobie. ### 11.14 Metryki — korekta dla polskiego Cele z sekcji 5 obowiązują, z dwiema poprawkami: | Metryka | Cel dla polskiego | |---|---| | Burstiness | **> 0,5** (bez zmian) | | SD długości zdań | **> 7 słów** (bez zmian) | | Gęstość myślników | **< 10 na 1000 słów** (ostrzej niż w angielskim) | | Przymiotniki z 11.6 | **≤ 1 na 150 słów** | | Konkret (liczba/nazwa/data) | **≥ 1 na 100 słów** (bez zmian) | | Fleksja | **zero nieodmienionych nazw własnych** | ### 11.15 Nie przesadzaj Nie są dowodem na AI i nie wolno ich usuwać na wszelki wypadek: poprawna polszczyzna, jedna pauza użyta sensownie, poprawnie użyty imiesłów, zdanie złożone, strona bierna tam, gdzie ważniejszy jest obiekt. Nie dodawaj literówek. Nie pisz małą literą na siłę. Nie wrzucaj przekleństw, żeby brzmieć autentycznie. Każde z tego jest dziś osobnym sygnałem. --- ## 12. Voice matching and context drift Three operational rules that outperform any banlist. **1. Derive the style guide from real samples.** Ask the user for three pieces of their actual writing — emails, Slack messages, an old post. Extract: mean sentence length, favourite connectors, punctuation habits, how they open, how they close, their quirks. Write to that profile. A banlist tells you what not to write; a sample tells you what to write instead. Keep the quirks, including the ones that look like errors. **2. Restart against context drift.** Inside a long session, output regresses toward default polished prose, because the model's own earlier polished output is in context and reinforces itself. Re-read this protocol every ten turns, or start a fresh conversation for the final pass. **3. Simplification is a first pass, not the finish.** Forcing short words and one idea per sentence (the ASD-STE100 discipline used in aircraft manuals) strips flourish fast, but it flattens burstiness below the section 5 target. Simplify first, then deliberately restore variance: merge two short sentences into one long one, then cut a third to four words. **Read it aloud.** Any sentence you cannot say in one breath in the author's own voice fails, whatever the metrics say. --- ## 13. Tools Everything here is free and needs no key. - **Pattern catalog.** Every pitfall in this protocol plus the smaller ones, each with a bad and a good example and how to spot it: https://writelikehuman.app/patterns.md (JSON: `/patterns.json`, one pattern: `/patterns/.md`). - **Checker.** `POST https://writelikehuman.app/check` with `{"text": "...", "lang": "en"}` returns the section 5 metrics and every pattern hit with a quote. It counts; it does not judge meaning. Run it on your draft if you can make HTTP requests, then fix what it reports and read the draft once more yourself. - **MCP server.** `https://writelikehuman.app/mcp` (streamable HTTP, no auth). Tools: `get_rules`, `list_patterns`, `get_pattern`, `check_text`. - **Agent skill.** `https://writelikehuman.app/SKILL.md` for Claude Code and other agents that load skills. - **For crawlers.** `https://writelikehuman.app/llms.txt` and `https://writelikehuman.app/llms-full.txt`. --- ## Sources - Wikipedia, *Signs of AI writing* (WikiProject AI Cleanup) — the base taxonomy: https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing - Kobak, González-Márquez, Horvát & Lause (2024), *Delving into ChatGPT usage in academic writing through excess vocabulary*, arXiv:2406.07016 — 14.2M PubMed abstracts, the style-word spike - SlopDetector (2026), *Signs of AI Writing: 12 Patterns With Reproducible Thresholds* — em dash density and convergence thresholds: https://slopdetector.org/blog/signs-of-ai-writing - GPTZero, *What is perplexity & burstiness for AI detection?* — burstiness formula, 0.6-1.2 human vs 0.2-0.4 model - Freeburg (2026), *The Last Fingerprint: How Markdown Training Shapes LLM Prose*, arXiv — GPT-4.1 at 10.62 em dashes per 1,000 words vs 3.23 human baseline - Liang et al. (Stanford, 2023) — 61.3% false-positive rate of GPT detectors on non-native-speaker TOEFL essays - Sadasivan et al., *Counter Turing Test (CT²)*, EMNLP 2023 — perplexity and burstiness converge to human ranges as model size grows Polish sources (section 11): - Wolniewicz (2026), *Skaner po polsku, który łapie 120 fraz nadużywanych przez AI* — the nine Polish categories, the "To nie X, to Y" pattern, the półpauza rule: https://jacekwolniewicz.pl/skaner-po-polsku-ktory-lapie-120-fraz-naduzywanych-przez-ai/ - Kopeć, *Jak rozpoznać tekst pisany przez AI? 7 sygnałów* — adjective density as the strongest Polish signal (3-4 per paragraph): https://justynakopec.pl/jak-rozpoznac-tekst-pisany-przez-ai-sygnaly/ - Top Online (2024), *Jak rozpoznać tekst napisany przez AI?* — English calques in Polish output, "oferta dedykowana": https://toponline.pl/blog/jak-rozpoznac-tekst-napisany-przez-ai - Pękala (feb.net.pl, 2026), *Jak sprawdzić, czy tekst został napisany przez AI?* — title case and comma-after-adverbial as English habits carried into Polish; readers identified polished AI text in 19.6% of cases (International Review of Economics Education, 2025); experienced LLM users misclassified 1 of 300 texts (Russell et al., arXiv 2025): https://feb.net.pl/blog/jak-sprawdzic-czy-tekst-zostal-napisany-przez-ai - Poradnia Językowa UZ / Wolański, *Edycja tekstów* — Polish dash convention: myślnik spaced on both sides, półpauza standard in prose - X consensus, Aug-Sep 2026 — the post-em-dash era: uniformly avoiding the em dash is now itself an accusation trigger; style-guide-from-samples and fresh-context restarts are the fixes practitioners report working (section 12) --- *Applied by Spectrum Flare across DocsAura, client documents, and published content. Corrections welcome.* --- # Write Like Human: pattern catalog version: 1.2.0 · canonical: https://writelikehuman.app/patterns Every tell the Human Writing Protocol names, one entry each, with a bad example, a better one and the fix. "zero allowed" means none in finished writing. "max N per M words" is a density cap. "watch" is a signal worth a second look, not an error on its own. The full protocol is at https://writelikehuman.app/rules.md. ## Leakage Chatbot residue that no human ever types. ### Citation and tool artifacts id: `citation-artifacts` · protocol § 1 · any language · zero allowed Internal tokens from ChatGPT, Grok and other tools that survive a copy-paste. Nothing else produces them. - Bad: Revenue grew in Q3 :contentReference[oaicite:2]{index=2}. - Better: Revenue grew in Q3 (company filing, 14 October). - Fix: Delete the token. If a real source sat behind it, cite that source by name. ### Chatbot tracking parameters id: `tracking-params` · protocol § 1 · any language · zero allowed Links copied out of a chatbot answer carry the chatbot in the URL. - Bad: https://example.com/report?utm_source=chatgpt.com - Better: https://example.com/report - Fix: Strip the utm_source parameter. ### Knowledge-cutoff disclaimer id: `cutoff-disclaimer` · protocol § 1 · en · zero allowed Only a language model talks about its training data. - Bad: As of my last update, the tool supports six languages. - Better: The tool supported six languages in March. - Fix: Delete the whole sentence. If the information may be stale, say what date it comes from. ### Refusal and assistant residue id: `refusal-residue` · protocol § 1 · en · zero allowed Assistant voice leaking into text that a person is supposed to have written. - Bad: As an AI language model, I cannot give legal advice, but here are some points. - Better: This is not legal advice. Three points to check with your lawyer: - Fix: Delete the whole sentence. ### Unfilled template slots id: `template-slots` · protocol § 1 · any language · zero allowed A bracketed blank means the text was generated and never read. - Bad: Dear [Client Name], thank you for choosing [Company]. - Better: Dear Ms Ortega, thank you for choosing Harbor Print. - Fix: Stop and ask the user for the missing value. Never guess it. ### Unasked closing offer id: `closing-offer` · protocol § 1 · en · zero allowed The chatbot sign-off. A person who wrote the text does not offer to write more of it. - Bad: Let me know if you would like me to expand on any of these points! - Better: (nothing; the text ends on its last point) - Fix: Delete it. End on the last real point. ### Meta-narration id: `meta-narration` · protocol § 1 · en · zero allowed Describing the text instead of writing it. It reads like the wrapper around a chatbot answer. - Bad: Certainly! Here's a rewritten version that is more concise: - Better: (start with the text itself) - Fix: Delete it. The reader can see the structure. ### Markdown in a plain-text channel id: `markdown-in-plain-text` · protocol § 1 · any language · watch Asterisks and hashes show up raw in email, LinkedIn and SMS. Only a model pastes them there. - Bad: Hi Sam, **quick update:** ### Next steps - Better: Hi Sam, quick update on next steps. - Fix: Use the channel's own formatting, or none. ## Rhetoric Sentence shapes models reach for by default. ### Negative parallelism id: `negative-parallelism` · protocol § 2.1 · en · zero allowed The loudest short-form tell of 2026. It invents a position nobody held so it can knock it down. - Bad: It's not just a tool, it's a new way of working. - Better: The tool cut our review time from two days to one. - Fix: State the claim directly, or write two plain sentences. ### Rule of three on autopilot id: `autopilot-triplet` · protocol § 2.2 · en · max 1 per 200 words Three balanced items is what a model produces when it has one real example and fills to three. - Bad: Our platform is efficient, scalable, and reliable. - Better: The platform handled 3,000 orders on Black Friday without a restart. - Fix: Keep the items you can defend. Two or four is fine. Uneven lists read as human. ### Empty opener id: `empty-opener` · protocol § 2.3 · en · zero allowed Throat-clearing before the point. An editor deletes it first. - Bad: In today's fast-paced digital world, businesses need to move quickly. - Better: Our competitors ship a release every week. We ship one a quarter. - Fix: Open on the actual claim. ### Fake authority id: `fake-authority` · protocol § 2.4 · en · zero allowed An appeal to research that has no name, date or link behind it. - Bad: Studies have shown that remote teams are more productive. - Better: In a 2015 Stanford experiment with a Chinese travel agency, staff working from home were 13% more productive. - Fix: Name the study, the number and the year, or delete the claim. Never attach a citation you cannot verify. ### Pseudo-wisdom filler id: `pseudo-wisdom` · protocol § 2.5 · en · zero allowed A sentence shaped like insight that carries no information. Delete it and nothing is lost. - Bad: At the end of the day, the key is finding the right balance. - Better: We cap meetings at two afternoons a week. - Fix: Delete it. If the paragraph now feels thin, it was thin before. ### Transition stacking id: `transition-stacking` · protocol § 2.6 · en · density limit Paragraphs that open with Furthermore, Moreover, Additionally read like an outline being filled in. - Bad: Moreover, the new process reduces errors. - Better: The new process also caught the invoice bug we missed in May. - Fix: Delete the connector. Let the order of the paragraphs carry the logic. Zero in anything under 400 words. ### Outline-formula closer id: `outline-closer` · protocol § 2.7 · en · zero allowed The five-paragraph-essay ending: challenges, then the future, then a restatement. - Bad: As the landscape continues to evolve, one thing is clear: adaptability will be key. - Better: We will know by March whether the pilot pays for itself. - Fix: End on the last real thing you have to say. ### Superficial -ing tail id: `ing-tail` · protocol § 2.8 · en · zero allowed A participle bolted onto the end of a sentence to claim significance it has not earned. - Bad: The team shipped the feature early, highlighting the importance of collaboration. - Better: The team shipped the feature a week early. - Fix: Cut the tail, or make it its own sentence with a real subject. ### Sycophantic tone id: `sycophancy` · protocol § 2.9 · en · zero allowed Praise for the question instead of an answer to it. - Bad: Great question! You're absolutely right to ask about pricing. - Better: Pricing starts at $12 a seat. - Fix: Delete it. Start with the answer. ### Copula avoidance id: `copula-avoidance` · protocol § 2.10 · en · max 1 per 300 words Models avoid plain "is" and "has" and reach for "serves as" and "boasts". - Bad: The library serves as a hub for the community and boasts a large archive. - Better: The library is where the town meets. It has 40,000 photographs of the harbour. - Fix: Use "is" or "has" unless the fancier verb adds meaning. ### Significance inflation id: `significance-inflation` · protocol § 2.11 · en · zero allowed Everything is pivotal, vital and lasting. The reader stops believing any of it. - Bad: The launch marks a pivotal moment and leaves an indelible mark on the industry. - Better: The launch doubled our weekly sign-ups. - Fix: Say what happened and let the fact carry its own weight. ### Promotional puffery id: `promotional-puffery` · protocol § 2.11 · en · max 1 per 300 words Brochure adjectives. They describe how the writer wants you to feel, not the thing. - Bad: Nestled in the heart of the city, this breathtaking venue offers world-class service. - Better: The venue is two streets from the station and seats 120. - Fix: Replace the adjective with the detail that made you reach for it. ### Elegant variation id: `elegant-variation` · protocol § 2.12 · any language · watch Rotating synonyms so a noun never repeats. The reader wonders whether "the firm" and "the organisation" are different things. - Bad: Acme raised prices. The firm said the organisation needed to cover costs, and the Ohio-based company expects… - Better: Acme raised prices. Acme said it needed to cover costs and expects… - Fix: Pick one name and repeat it, or use a pronoun. - Detection: needs a reader; the checker cannot count this one. ### False range id: `false-range` · protocol § 2.13 · en · zero allowed "From X to Y" where X and Y are not two ends of any scale. It sounds broad and says little. - Bad: We help everyone from startups to governments with everything from pricing to culture. - Better: We help software companies with 10 to 200 staff set their prices. - Fix: Name the things you mean, or drop the phrase. ### Staged reveal id: `staged-reveal` · protocol § 2.14 · en · zero allowed A drumroll before the point. Social-media copy taught models to set up every fact like a punchline. - Bad: The result? A 40% drop in errors. Here's the kicker: it took one afternoon. - Better: Errors dropped 40%, and the change took one afternoon. - Fix: State the result in a normal sentence. ### Fragment drumroll id: `stacked-fragments` · protocol § 2.14 · any language · zero allowed Three clipped fragments in a row ("No fluff. No filler. Just results.") is a template, not a voice. - Bad: No fluff. No filler. Just results. - Better: The report is two pages and every number in it has a source. - Fix: Write one sentence that says the thing. ### Audience split id: `whether-youre` · protocol § 2.14 · en · zero allowed "Whether you're a X or a Y" tries to address everyone and so addresses no one. - Bad: Whether you're a seasoned pro or just starting out, this guide has you covered. - Better: This guide is for people setting up their first online shop. - Fix: Write to the one reader this text is for. ### Hedge stacking id: `hedge-stacking` · protocol § 7 · en · watch Two hedges on one claim. Honest uncertainty names its limit; stacked hedges just blur it. - Bad: This could potentially help to possibly reduce costs. - Better: This should cut costs if the supplier holds its price. - Fix: Keep one hedge and say what it depends on. ## Lexicon Words whose frequency jumped after 2022. ### Post-2022 style words id: `style-words` · protocol § 3.1 · en · max 3 per 500 words Words like delve, underscore and showcase spiked in published text after ChatGPT launched. Two in one sentence is a fingerprint. - Bad: This report delves into the intricate landscape of fintech, underscoring its multifaceted nature. - Better: This report looks at how 40 fintech startups priced their first product. - Fix: delve → look at · underscore → show · tapestry → describe the thing · landscape → field · foster → build · harness → use · navigate → handle. ### Corporate verb inflation id: `verb-inflation` · protocol § 3.2 · en · max 1 per 300 words Latinate verbs standing in for plain ones. - Bad: We leverage AI to streamline and facilitate onboarding. - Better: We use AI to fill in the onboarding forms. - Fix: utilize/leverage → use · facilitate → help · streamline → simplify · commence → start · endeavour → try. ### Buzz phrases id: `buzz-phrases` · protocol § 3.3 · en · zero allowed Phrases so worn they carry no meaning. - Bad: This is a real game-changer that will move the needle. - Better: This halves the time it takes to close a month. - Fix: Say what changed, what the easy part is, what moved. ### Empty intensifiers id: `empty-intensifiers` · protocol § 3.4 · en · max 1 per 200 words Very, truly, incredibly: volume without information. - Bad: Sales grew significantly and the team was incredibly happy. - Better: Sales grew 34%. - Fix: Delete it, or replace it with a number. ## Punctuation Marks that give away the tool or the training data. ### Em dash density id: `em-dash-density` · protocol § 4.1 · en · density limit Humans use dashes too. The tell is density: well above 10 per 1,000 words, several per paragraph. Banning dashes outright is now its own tell. - Bad: The plan — which we tested — worked — mostly. - Better: The plan worked, mostly. We tested it for two weeks. - Fix: Keep under 15 per 1,000 words and one per paragraph in short pieces. Swap the rest for commas, colons, parentheses or full stops. ### Mixed dash conventions id: `dash-convention-mix` · protocol § 4.1 · en · watch Spaced dashes ( — ) and tight dashes (word—word) in the same text suggest pasted fragments. - Bad: The plan — tested twice—worked. - Better: The plan, tested twice, worked. - Fix: Pick one convention and hold it. ### Double hyphen as a dash id: `double-hyphen` · protocol § 4.2 · any language · zero allowed A plain-text substitute for a dash that usually marks copied output. - Bad: We shipped it -- finally. - Better: We shipped it. Finally. - Fix: Use a full stop or a comma. ### Curly quotes in plain text id: `curly-quotes-plain-text` · protocol § 4.2 · en · watch Typographic quotes in a code comment, SMS or form field are a copy-paste fingerprint. - Bad: Reply “yes” to confirm. - Better: Reply "yes" to confirm. - Fix: Use straight quotes when the destination is plain text. Keep curly quotes in typeset documents. ### Emoji as bullets or dividers id: `emoji-bullets` · protocol § 4.3 · any language · zero allowed The rocket, lightbulb, sparkles and check-mark set is the signature of generated social posts. - Bad: 🚀 Faster builds / 💡 Smarter defaults / ✅ Fewer bugs - Better: Builds are faster, the defaults changed, and we closed 30 bugs. - Fix: At most one emoji per message, and only when it carries tone a word cannot. ## Formatting Structure used because it is safe, not because the content needs it. ### Bold lead-ins in lists id: `bold-lead-ins` · protocol § 4.4 · any language · zero allowed "**Speed:** the system is faster" on every bullet is the house style of chatbot answers. - Bad: - **Speed:** The system is faster. / - **Cost:** It is cheaper. - Better: The system is faster and costs less to run. - Fix: Write the bullet as a sentence, or turn the list into prose. Fine in reference documentation. ### Title-case headings id: `title-case-headings` · protocol § 4.4 · any language · zero allowed Every Word Capitalised in a body-copy heading is a default, not a choice. - Bad: ## How To Improve Your Team Workflow - Better: ## How to improve your team workflow - Fix: Sentence case: capital on the first word and on names only. ### Bold overuse id: `bold-overuse` · protocol § 4.4 · any language · density limit When everything is bold, nothing is. - Bad: The **new plan** gives **every team** a **clear budget** and **faster approvals**. - Better: The new plan gives every team its own budget. **Approvals now take one day.** - Fix: Keep bold for the one phrase a skimming reader must not miss. ### A list where prose would do id: `list-for-prose` · protocol § 4.4 · any language · watch Models reach for bullets because bullets are safe. Three sentences that follow from each other are not a list. - Bad: - The deadline moved / - So we cut scope / - So we shipped on time - Better: The deadline moved, so we cut scope and still shipped on time. - Fix: Write the sentences. - Detection: needs a reader; the checker cannot count this one. ### Horizontal rules everywhere id: `rule-between-sections` · protocol § 4.4 · any language · watch A divider between every short section is decoration. - Bad: Intro / --- / Point one / --- / Point two - Better: Intro, then two sections under their own headings. - Fix: Let headings and white space separate sections. - Detection: needs a reader; the checker cannot count this one. ## Substance Text that reads fine and says nothing. ### Vague quantifiers id: `vague-quantifiers` · protocol § 6 · en · watch "Many", "numerous", "a wide range of" stand where a number should be. - Bad: Numerous customers use a wide range of our features. - Better: 212 customers use at least three of our features. - Fix: Use the number. If you do not have it, ask for it. ### No specifics id: `no-specifics` · protocol § 6 · any language · density limit The most reliable signal. Strip every tell from an empty paragraph and it is still empty. - Bad: Nutrition plays a crucial role in overall wellness. - Better: Swapping one soda a day for water cut my sugar intake by about 35 grams. - Fix: At least one number, name or date per 100 words. If the source has none, ask the user for one. Never invent it. ### Fails the deletion test id: `deletion-test` · protocol § 6 · any language · density limit A sentence you can delete without losing information was filler. - Bad: Communication is essential in any team. It helps everyone stay aligned. - Better: We moved the stand-up to Slack and got back 40 minutes a day. - Fix: Delete it. Fewer than a third of sentences should survive deletion without loss. - Detection: needs a reader; the checker cannot count this one. ### Fabricated specifics id: `fabricated-specifics` · protocol § 10 · any language · zero allowed The worst outcome of humanising. A made-up statistic is real harm; a generic sentence is only a style problem. - Bad: (an invented "73% of managers" to satisfy the specificity rule) - Better: "How many clients did you onboard last quarter? I will use your number." - Fix: Ask the user for the number. Use [brackets] only while drafting, and never ship them. - Detection: needs a reader; the checker cannot count this one. ## Overcorrection The tells that come from scrubbing too hard. ### Over-scrubbing id: `over-scrubbing` · protocol § 7 · any language · watch Choppy fragments, forced lowercase and no subordinate clauses: the style that replaced the em dash as a tell. - Bad: we tried it. it worked. kind of. not really. - Better: We tried it for a month, and it worked about half the time. - Fix: Keep long sentences when the idea is long. Merge two short sentences into one. ### Deliberate flaws id: `deliberate-flaws` · protocol § 7 · any language · zero allowed Planted typos, fake "I think maybe" hedges and padded personal detail are now tells of their own. - Bad: i think maybe this could work lol, teh numbers look ok - Better: This could work. The numbers hold if churn stays under 3%. - Fix: Write correctly. Keep the author's real quirks; do not add new ones. - Detection: needs a reader; the checker cannot count this one. ### Scrubbing the author's voice id: `voice-scrubbing` · protocol § 10 · any language · zero allowed An author's lowercase starts or double-dot pauses are the fingerprint you are protecting. - Bad: (normalising a writer who always starts emails with "hey..") - Better: (leaving "hey.." exactly as they write it) - Fix: Keep the quirks from the source text, including ones that look like errors. - Detection: needs a reader; the checker cannot count this one. ### Synonym swapping without restructuring id: `synonym-swapping` · protocol § 10 · any language · watch Swapping "leverage" for "use" inside the same balanced 20-word sentence changes nothing measurable. - Bad: We use AI to simplify and help onboarding. (same shape as before) - Better: AI now fills in the onboarding forms. It took a week to set up. - Fix: Fix the structure first, the vocabulary second. - Detection: needs a reader; the checker cannot count this one. ## Polski Wzorce typowe dla polskiego tekstu z AI. ### Puste otwarcia id: `pl-openers` · protocol § 11.1 · pl · zero allowed Rozgrzewka przed tezą. Redaktor kasuje ją jako pierwszą. - Bad: W dzisiejszym dynamicznie zmieniającym się świecie firmy muszą działać szybciej. - Better: Konkurencja wypuszcza nową wersję co tydzień, my raz na kwartał. - Fix: Zacznij od tego, co masz do powiedzenia. ### Fałszywa autentyczność id: `pl-fake-authenticity` · protocol § 11.2 · pl · watch Anegdota bez żadnego konkretu. Przechodzi tylko wtedy, gdy w następnym zdaniu pada rok, branża, liczba albo nazwa. - Bad: Ostatnio rozmawiałem z klientem o ważnych sprawach. - Better: Rozmawiałem z trenerką, która prowadzi szkolenia HR od 2018 roku. - Fix: Dodaj konkret albo skasuj zdanie. ### Wypełniacze id: `pl-fillers` · protocol § 11.3 · pl · zero allowed Zwroty, które zapowiadają treść zamiast ją podać. - Bad: Warto zauważyć, że kluczowym aspektem jest komunikacja. - Better: Zespół odpisuje klientom w ciągu dwóch godzin. - Fix: Skasuj w całości. Zdanie zwykle działa bez nich. ### Szkolne zakończenia id: `pl-closers` · protocol § 11.4 · pl · zero allowed Zakończenie z wypracowania: podsumowanie tego, co czytelnik właśnie przeczytał. - Bad: Podsumowując, mam nadzieję, że ten artykuł okazał się pomocny. - Better: Pilotaż kończy się w marcu. Wtedy zobaczymy, czy się zwraca. - Fix: Kończ na ostatniej rzeczy, którą naprawdę masz do powiedzenia. ### To nie X, to Y id: `pl-negative-parallelism` · protocol § 11.5 · pl · zero allowed Najgłośniejsza konstrukcja z angielskiego szablonu, przeniesiona 1:1 na polski. - Bad: To nie tylko narzędzie — to nowy sposób pracy. - Better: Narzędzie skróciło nam przegląd umów z dwóch dni do jednego. - Fix: Powiedz wprost, o co chodzi, albo napisz dwa zwykłe zdania. ### Trzy urwane zdania id: `pl-stacked-fragments` · protocol § 11.5 · pl · zero allowed „Bez ściemy. Bez bullshitu. Tylko konkret." to szablon, nie styl. - Bad: Bez ściemy. Bez lania wody. Tylko konkret. - Better: Raport ma dwie strony i każda liczba ma źródło. - Fix: Napisz jedno zdanie, które mówi, o co chodzi. ### Przymiotniki bez pokrycia id: `pl-empty-adjectives` · protocol § 11.6 · pl · max 1 per 150 words Najsilniejszy polski sygnał. Trzy lub cztery z tej listy w jednym akapicie to praktycznie pewność, że pisała AI. - Bad: Oferujemy kompleksowe, innowacyjne i profesjonalne rozwiązania. - Better: Jako pierwsi w Polsce łączymy kurs Excela z audytem firmowych arkuszy. - Fix: Zostaw jeden na 150 słów i tylko z liczbą albo dowodem obok. ### Kalki z angielskiego id: `pl-calques` · protocol § 11.7 · pl · zero allowed Angielskie wyrażenia przetłumaczone słowo w słowo. - Bad: Dedykowane rozwiązanie zaadresuje problem i dostarczy wartość. - Better: Program dla księgowych z biur rachunkowych skraca zamknięcie miesiąca o dwa dni. - Fix: dedykowany → dla, przeznaczony dla · zaadresować problem → zająć się · dostarczać wartość → nazwij, co konkretnie. ### Czasowniki napuszone id: `pl-inflated-verbs` · protocol § 11.8 · pl · max 1 per 150 words Urzędowe czasowniki w miejscu zwykłych. - Bad: Posiadamy zespół, który realizuje projekty poprzez dedykowane narzędzia. - Better: Mamy pięć osób, które robią projekty w Figmie. - Fix: posiadać → mieć · realizować → robić · dokonać zakupu → kupić · poprzez → przez · w celu → żeby. ### Ogony imiesłowowe id: `pl-participle-tails` · protocol § 11.9 · pl · zero allowed Dopisek na końcu zdania, który ma nadać mu wagę. - Bad: Wdrożyliśmy nowy proces, co przekłada się na lepszą jakość. - Better: Wdrożyliśmy nowy proces. Reklamacji jest o połowę mniej. - Fix: Utnij albo zrób z tego osobne zdanie z prawdziwym podmiotem. ### Coachingowy bełkot id: `pl-coaching` · protocol § 11.10 · pl · zero allowed Język z okładek poradników. - Bad: Uwolnij swój wewnętrzny potencjał i wyrusz w podróż transformacji. - Better: Po sześciu tygodniach będziesz prowadzić spotkanie zespołu bez notatek. - Fix: Skasuj. Napisz, co konkretnie się zmieni. ### Angielska pauza w polskim tekście id: `pl-em-dash` · protocol § 11.11 · pl · zero allowed Długa pauza bez spacji (słowo—słowo) to import z angielskiego. Po polsku myślnik ma spacje z obu stron, standardem jest półpauza. - Bad: Wynik—jak się okazało—był lepszy. - Better: Wynik – jak się okazało – był lepszy. - Fix: słowo – słowo (półpauza ze spacjami). Poniżej 10 na 1000 słów, najwyżej jedna na akapit. ### Angielski cudzysłów id: `pl-quotes` · protocol § 11.11 · pl · zero allowed Polski cudzysłów to „tekst”. Angielski "tekst" albo “tekst” w polskim zdaniu to odcisk kopiuj-wklej. - Bad: Nazwaliśmy to "szybką ścieżką". - Better: Nazwaliśmy to „szybką ścieżką”. - Fix: Zamień na „…”. Cytat w cytacie: «…». ### Trzy kropki zamiast wielokropka id: `pl-ellipsis` · protocol § 11.11 · pl · watch Wielokropek to jeden znak. - Bad: I wtedy... - Better: I wtedy… - Fix: Zamień ... na … ### Przecinek po okoliczniku id: `pl-comma-after-adverbial` · protocol § 11.7 · pl · zero allowed „Dodatkowo, warto…" to angielska interpunkcja. Po polsku bez przecinka. - Bad: Ponadto, system wysyła przypomnienia. - Better: System wysyła też przypomnienia. - Fix: Usuń przecinek albo cały łącznik. ### Nieodmienione nazwy id: `pl-declension` · protocol § 11.12 · pl · zero allowed Modele uczone na angielskim zostawiają nazwy w mianowniku: „podczas zamach na Donald Trump". - Bad: Po rozmowie z Anna Kowalska podpisaliśmy umowę z firma Orlen. - Better: Po rozmowie z Anną Kowalską podpisaliśmy umowę z Orlenem. - Fix: Sprawdź przypadek każdej nazwy własnej, liczebniki („dwóch klientów") i rekcję („używać czegoś"). - Detection: needs a reader; the checker cannot count this one. ### Mieszana forma adresu id: `pl-address-mixing` · protocol § 11.13 · pl · zero allowed Przeskok z „Sprawdź" na „Zachęcamy Państwa" w sąsiednich akapitach. - Bad: Sprawdź naszą ofertę. Zachęcamy Państwa do kontaktu. - Better: Sprawdź naszą ofertę i napisz do nas. - Fix: Wybierz jedną formę: ty, Państwo albo bezosobową, i trzymaj ją do końca. ### Anafora ×3 id: `pl-anaphora` · protocol § 11.5 · any language · zero allowed Trzy zdania pod rząd zaczynające się tak samo to retoryczny szablon. - Bad: Chcemy rosnąć. Chcemy zatrudniać. Chcemy zmieniać rynek. - Better: W tym roku chcemy zatrudnić dwie osoby do obsługi klienta. - Fix: Połącz dwa z nich albo zacznij inaczej. ### Wielkie Litery W Nagłówkach id: `pl-title-case` · protocol § 11.7 · pl · zero allowed Po polsku wielka litera tylko w pierwszym słowie i nazwach własnych. - Bad: ## Jak Rozpoznać Tekst Pisany Przez AI - Better: ## Jak rozpoznać tekst pisany przez AI - Fix: „Jak rozpoznać tekst AI", nie „Jak Rozpoznać Tekst AI".