unbias.me is a news aggregator that uses AI to evaluate news articles for common biases. Its goal is to help readers recognize bias in news coverage and find the most objective versions of popular stories. It also aims to neutralize the loaded language that headlines use to bait readers. To ensure transparency and neutrality, every AI prompt used by the site to evaluate articles is visible on this page. Each article’s bias scores are judgments made by TypeSafe’s Jev model against known bias categories, not the opinions of any person or group of people. Because the site is inexpensive to operate and maintain, it does not need to be monetized and is offered 100% cookie- and ad-free.
Every link comes from the public feed of one of more than 390 outlets and points back to that outlet. Articles are read once in order to evaluate them; their text is neither stored nor displayed. Due to technical limitations, articles are evaluated based on their first 12,000 characters only (roughly 2,000 words). The site reads as a named bot, unbiasbot, and obeys each outlet’s robots.txt; an outlet that does not allow it is not listed. Any outlet that does not want to be listed, and anyone with a correction, can write to contact@unbias.me.
Is this text written in English?
Is this a news or current-affairs piece?
{
"true": "Reporting, analysis, opinion or commentary about events, policy, business, science, health, culture or society",
"false": "Service or filler content: an advice column, horoscope, recipe, puzzle, quiz, crossword, obituary, shopping or product guide, a live-blog stub with no reporting, a newsletter or podcast index, a digest or wire roundup listing several unrelated stories (\"AP Top Headlines at 9:03 a.m.\"), or a personal essay with no news subject"
}
Is this piece a roundup: several unrelated news items or developments collected into one piece, such as a daily digest, quick hits, a morning briefing or newsletter edition, a wire roundup, a news quiz, or a sports roundup covering several games? Judge the structure of the piece, not its tone or quality: a column or report about one subject is not a roundup, and an explainer that lists several points about one development is not a roundup.
{
"true": "The piece strings together items on several unrelated developments, each with its own subject, and no single development is the piece's subject",
"false": "The piece is about one development or one subject, however many angles, points, games or quotes it includes"
}
Does this piece exist to get the reader to buy, subscribe, sign up, book or attend something? Judge why the piece exists, not what it is about: a piece may report on a company, a product, a show, a team or a game and still be reporting. The mark of this one is that the piece is built around a commercial instruction: a promo code, a discount or free trial, a streaming or shopping service it recommends and tells the reader how to join, ticket or booking links, a deal that expires, or an event the piece exists to fill seats at. A note that the publisher may earn a commission is strong evidence. A report of a product launch, a review that judges something on its merits, and reporting or commentary about a company or an industry are not this, however much money is involved.
{
"true": "The piece is built to convert the reader: it names a product, service, ticket or event to buy or join and tells them how, and that instruction is the reason the piece exists, whatever reporting is wrapped around it",
"false": "The piece reports, explains or argues about its subject; it sells nothing, or any purchase detail is incidental to a development it reports"
}There are three general feeds: news, analysis, and opinion. Each contains stories and articles sorted by coverage and recency.
Topics follow the IPTC Media Topics standard used by newsrooms and wire services.
Which news section does this article belong in? The sections follow the IPTC Media Topics standard. Judge the subject of the story, not the outlet and not the writer's opinion about it.
{
"us_politics": {
"what": "The exercise of power and the running of government in the United States: elections, campaigns, Congress, the presidency, parties, federal and state government and policy, political process and crisis",
"not_for": "A court ruling or prosecution, which is crime_justice, even when the case is politically charged; events outside the United States, which are world",
"examples": [
"Nessel endorses El-Sayed in Michigan Senate race",
"Affordability is the Democrats' unifying midterm theme"
]
},
"world": {
"what": "Government, politics, elections and international relations outside the United States, and conflict anywhere: armed conflict, acts of terror, civil unrest, coups, massacres, war victims, peace processes",
"not_for": "US domestic politics, which is us_politics; a foreign company or market story, which is business"
},
"business": {
"what": "The planning, production and exchange of wealth: companies and enterprises, markets and exchanges, the economy, products and services, trade, inflation, taxes, personal money, and labour: employment, unions, labour relations, unemployment, retirement"
},
"technology": {
"what": "Technology and engineering: computing, AI, the internet, software, devices and gadgets, tech companies and social platforms, and the business and policy around them",
"not_for": "Research findings and discovery, which are science"
},
"science": {
"what": "Human understanding and methodical study of the natural, formal and social sciences: scientific research and discovery, space and spaceflight, biomedical science, mathematics, the natural world as studied: a finding, a mission, a new species, a fossil, an instrument",
"not_for": "Climate, energy, pollution, wildlife management and land or water disputes, which are environment; medicine and care, which are health; a storm or a forecast, which is weather; a product or a company, which is technology",
"examples": [
"Scientists discover two new species of sea spiders along Canadian coastline",
"SpaceX launches Starship on first orbital flight"
]
},
"environment": {
"what": "The protection, damage and condition of the planet's ecosystem: climate change and climate policy, conservation, pollution, natural resources, nature and wildlife, sustainability, energy and its politics, fights over land, water and development",
"not_for": "A storm or a forecast and what it does this week, which is weather; a discovery about nature made for its own sake, which is science",
"examples": [
"Colorado wolves stay west as livestock losses continue",
"Nepal's hydropower ambitions face climate reality check"
]
},
"health": {
"what": "Physical and mental well-being: diseases and conditions, treatments and procedures, public health, health care and insurance, health facilities and organisations, the medical profession, reproductive and mental health",
"not_for": "Diet, fitness and self-care features, which are lifestyle"
},
"education": {
"what": "Furthering knowledge, formally or informally: schools, universities, students, teachers, teaching and learning, curriculum, testing and examinations, school boards and education policy, vocational and online learning",
"not_for": "A campus crime or lawsuit, which is crime_justice; immigration or race as such, which is society"
},
"crime_justice": {
"what": "Crime, law enforcement, the judiciary, justice and law: specific crimes, policing, courts, lawsuits, prosecutions, sentencing and prisons",
"not_for": "Elections, legislation or campaign politics with no case or investigation at their centre, which is us_politics",
"examples": [
"DOJ indicts former postal worker over discarded ballots",
"Judge denies bid to end oversight of sheriff's office"
]
},
"society": {
"what": "Human social interaction, problems and welfare: communities, demographics, immigration and emigration, discrimination and diversity, fundamental rights, family, social conditions and problems, values, welfare, social movements, and human-interest items about people, achievements, anniversaries and mishaps",
"not_for": "A church, clergy or faith community as such, which is religion; schools and universities, which are education"
},
"entertainment": {
"what": "Arts and entertainment and cultural heritage: film, television and streaming, music, video games, comics and animation, books and literature, theatre, the visual arts, celebrity, awards, and the industries behind them",
"not_for": "The news business and journalism, which are media_press; fashion, food and travel, which are lifestyle"
},
"lifestyle": {
"what": "Activities undertaken for pleasure, relaxation or recreation outside paid employment: fashion and beauty, food and dining, travel, home and garden, hobbies, wellness and fitness, relationships, and personal essays about daily life",
"not_for": "Medicine, disease and health care, which are health"
},
"media_press": {
"what": "The news business, journalists, press freedom, media bias, misinformation, and checks on viral or disputed claims",
"not_for": "A story that merely quotes reporters or officials; the media itself must be the subject",
"examples": [
"The White House press corps' phony little game",
"Do images show a statue of Jesus holding two Uzis?"
]
},
"sports": {
"what": "Competitive activity and skill and the bodies involved: games, competitions, athletes, leagues, transactions, coaching, officiating, drug use in sport and the business of sport"
},
"religion": {
"what": "Belief systems, institutions and people who provide moral guidance: churches, denominations, clergy and religious leaders, faith communities, their practices, teachings, festivals and internal disputes, and relations between religion and government",
"not_for": "A crime, a war or an election in which religion is only the identity of those involved: a trial is crime_justice and a campaign is us_politics or world, whoever the people are",
"examples": [
"Pope Leo in Lourdes: what to know about one of Catholicism's most famous shrines",
"Pastor affirms Steven Lawson's authorship of scandal memoir"
]
},
"weather": {
"what": "Meteorological phenomena, forecasts and warnings, and disasters, accidents and emergency incidents: storms, hurricanes, floods, tornadoes, blizzards, heat waves, wildfires, earthquakes, crashes and major accidents, and the deaths, damage, outages, rescues and travel disruption they cause",
"not_for": "Climate research or climate policy, which is environment, even when a storm is the occasion for it. Deliberate violence, however many it hurt and whatever vehicle it involved: an attack, a hijacking, an act of terror, a shooting or arson is crime_justice (or world, for terror and conflict). IPTC files only accidental, natural or unexplained events here",
"examples": [
"Nor'easter batters US Northeast, causing coastal flooding and outages",
"Medical helicopter crashes off Catalina Island; 2 dead, 1 missing"
]
},
"other": "A news topic that fits none of the above"
}Articles that cover the same event are grouped using BAAI’s bge-small-en-v1.5 embedding model and sorted by bias score. Each story is represented by its least biased article.
Are these two items reporting the same news development?
{
"true": "The same development: one announcement, ruling, incident, event or situation, however differently the two are framed, and whether one reports it beginning, continuing or ending",
"false": "A different development about the same subject: a separate incident, ruling, announcement or response, rather than the same one going on"
}
Two news items about the same subject, from different days. `earlier` is the story as it stood; `later` appeared the next day. Does `later` report the same event as `earlier` (the same thing happened, now re-reported or updated with detail), or a new development: a distinct new action, decision, finding, reaction, death, resignation or ruling that followed from it and would deserve its own story?
{
"same": "The same event, re-reported, updated or detailed: new numbers, a later tally, more witnesses, the same ruling described again",
"development": "A new event in the same saga: someone resigns, reacts, is charged, rules, announces, investigators conclude, a consequence lands"
}These headlines are the stories of one running news subject, newest first. Name the subject: a short descriptive noun phrase of at most eight words, with no verb and no claim, specific enough to tell it from any other subject (name the people, places or things involved), in sentence case with no trailing period. Use only names that appear in the headlines. Reply with the name and nothing else.
New headlines are composed using Alibaba’s Qwen 3.7 Plus LLM, with Anthropic’s Claude Haiku 4.5 as a fallback, under instructions to remove loaded and sensational language while keeping facts. TypeSafe’s Jev model is used to check rewritten headlines to ensure factual accuracy.
Rewrite the headline so a reader can trust it: no loaded, emotive, partisan or sensational wording, no clickbait, sentence case (capitalise only the first word and proper nouns, never every word, however the original was written; acronyms and initialisms stay as the source writes them: an agency, a company or a ticker is NASA, not Nasa), no trailing period, under 15 words. Spell every name exactly as the source spells it. Keep what the piece is. You may be told its kind. A report's headline says plainly what happened. An analysis or opinion piece makes a case, so its headline should say what that case is, worded as the piece's claim rather than as established fact, instead of flattening it into a description of its subject. Add nothing the headline and summary do not contain: no 'experts say', 'analysts warn' or 'critics argue' unless the source names who, and no timing, cause or consequence it does not state. Details from the summary are fine. If the summary is missing or only repeats the headline, work from the headline alone. Always reply with a headline and nothing else: never a question, an explanation or a refusal.
Could the rewritten headline accurately head the same article as the original headline and summary, and the article text where it is given? Dropping loaded wording or adding a detail from the summary or the article is fine.
{
"true": "Same story and subject; nothing in the rewrite contradicts the original, the summary or the article",
"false": "Different subject, contradicts the original, or asserts something the original, the summary and the article do not support"
}
Does every claim, actor, attribution, timing and qualifier in the rewritten headline come from the original headline, the summary or the article text where it is given?
{
"true": "Everything in the rewrite is stated in, or plainly restates, the original, the summary or the article",
"false": "The rewrite adds something none of them contains: a person or group saying it, a time or occasion, a cause, a consequence, a hedge such as 'reportedly' or 'analysts say', or a name spelled differently from the source"
}
Is the rewritten headline correctly capitalised as a sentence-case headline? Sentence case means the first word is capitalised, proper nouns are capitalised (people, places, organisations, products, titles of works, branded names), acronyms and initialisms are in capitals, and every other word is lower case. Judge capitalisation only: ignore wording, accuracy, length and punctuation. A headline opening with a number needs no capital on the number.
{
"true": "Correct sentence case: every word that should be capitalised is, and no ordinary word is capitalised without reason",
"false": "At least one word that should be capitalised is in lower case, whether the first word, a person, place, organisation, product, work or acronym; or the headline is in Title Case with every word capitalised"
}Every article displays three tags: format, impact, and partisanship. Each is Jev’s most likely answer to a fixed question, with its confidence out of 100.
What kind of piece is this? Judge its purpose and stance, not its subject or which side it favours: a piece on either side of politics can be any of these. [ "Describes what happened with attribution to sources; the writer's own judgment does not appear", "Explains or interprets events with judgment supported by facts, data or named expertise, without taking a side on what should happen", "A column, editorial or commentary that argues a viewpoint; the writer's judgment is the point", "Persuasion is the purpose: campaigns for an outcome, a cause, a candidate or an action, whichever side, and asks the reader to side or act" ]
How long will it matter that what this piece reports happened, and to whom? Judge the development itself, whatever its subject: a contest, a performance or a cultural moment can matter as long as a vote or a ruling. Not how interesting the piece is, not how well it is written, and not how many people will read it. [ "Nothing happened: the piece exists to sell, promote or collect attention, and reports no development", "Over within days: a result, a quote, a stunt, a sighting; nothing is different for anyone next week", "Matters for weeks to those who follow the subject: the ordinary run of news about an institution, a company, a team or a case", "Still matters in a year: a decision, a ruling or a result that changes what a sector, a region or a population does from here", "A country or the world will be living with this for years: the kind of development people come to date things by" ]
Which side of US politics does this article's own framing favour, and how consistently? Look in three places: whose positions and sources it treats as credible, the terms it uses for issues and actors, and what it includes or leaves out. Judge the text, not the outlet, the topic, or whether it is an opinion piece. [ "Favours the left consistently in all three places: only its case is made, opponents appear in its terms, and what is included or left out serves it", "Favours the left in some of the three; the other side is still present in its own terms", "Favours neither side, or the piece has no partisan dimension", "Favours the right in some of the three; the other side is still present in its own terms", "Favours the right consistently in all three places: only its case is made, opponents appear in its terms, and what is included or left out serves it" ]
Jev also evaluates each article for 48 media biases, named as Wikipedia names them across its Media bias and Propaganda techniques articles, as well as a few others: clickbait, rage bait, weasel words, scare quotes, slippery slope, appeal to pity, nostalgia, sex appeal and information laundering. A bias tag appears when the answer is more likely yes than no. Each question was tuned against Opus’s answers. Three in four of Jev’s yes answers agree with it, and it finds about twice as many biases as Jev does.
Judge only the writer's own wording in the body of the article, not what quoted sources say and not headlines of other stories that may appear around it. Does the writer choose charged terms for people, groups, policies or positions where a neutral term would carry the same information? The pairs a careful editor would swap: regime and government, scheme and plan, slams and criticises, radical and left-wing, illegal alien and undocumented immigrant. Judging verbs and adjectives: rants, gloats, admits, shameless, disastrous, so-called, woke, extremist. Judge against the ordinary way that kind of piece is written: sports, entertainment and feature writing is vivid by convention, so a dominant win, a stunning catch or a legendary coach is description there, not loading. A vivid word that accurately describes an event is not loaded: a deadly crash is deadly. A label a subject uses of itself is not loaded. A column may argue; it still loads its language if it reaches for the charged term over the plain one about the people or positions it argues against. In a review, a recap or criticism, harsh words about a work, a performance, a fictional character, a game's design or a brand are the critic's verdict and not this; loading is a charged term for a real person, group, policy or position where a plain one would do.
{
"true": "The writer, in their own voice, applies charged terms or judging verbs to people, groups, policies or positions in a way a neutral rewrite would remove, at least once beyond a single conventional descriptor",
"false": "Subjects are named and described in plain terms for that kind of piece; charged words appear only inside attributed quotations, or describe events accurately rather than colour them"
}Judge the writer's own wording against the ordinary way that kind of news is written. Is the event sold beyond what happened: intensifiers and superlatives as the pull, a rare case presented as a trend, worst cases as likely, the extraordinary manufactured from the routine? Count only the writer's own words, including the headline and captions; vivid wording inside quotes from sources does not count. Partisan heat, mockery or harsh judgment of a person or claim is a different device and answers no unless it also inflates the size, rarity or stakes of the event. An event that is genuinely large or extraordinary, reported with its real numbers, answers no. Sports, entertainment and features are vivid by convention; judge them against that convention. Ask whether a plain statement of the same facts would read cooler; if not, answer no. Most news, including most opinion and most partisan writing, is in proportion to its facts; answer yes only when you can point to the writer's own words that make the event bigger, rarer or more dangerous than the facts given in the piece.
{
"true": "Specific words of the writer's own, in the headline or body, make the event bigger, rarer or more dangerous than the facts in the piece support: superlatives or intensifiers as the pull, a rare case as a trend, worst cases as likely, drama the facts do not supply",
"false": "The writer's own words keep the event in proportion for that kind of piece: a vivid event told plainly, a large event with its real numbers, a harsh or partisan judgment of a person or claim, an opinion column in its ordinary register, or heat that sits only inside quoted sources"
}Judge only the writer's own wording, not quoted sources. Does the writer attach a label to a person or group in place of describing them: RINO, socialist, extremist, denier, elite, globalist, the establishment, activist judge, radical, far-right, far-left, MAGA as a noun for people, woke? A label a person or party uses of itself, or a formal title, is not this. Answer yes when the label does the work of an argument the piece never makes.
{
"true": "People or groups are tagged with loaded labels in the writer's voice rather than described",
"false": "People are named by title, role or self-description; labels appear only in quotation"
}Does the writer use quotation marks or hedging words to discredit rather than to cite: so-called, self-styled, claims, alleged, purported, or a term put in quotation marks to signal that the writer does not accept it, when it is not a quotation from anyone? Legal hedging on an unproven charge, alleged before a crime, is convention and does not count. A term quoted because a named source used it does not count. Judge the writer's own sentences. Quotation marks around a phrase that a named person, body or document actually used are citation, not this, even when the writer goes on to dispute the phrase. Marks around slang, jargon, a pun, a title or the writer's own coinage are not this. Open contempt in plain words, ridiculous, absurd, propaganda, is a different device: without the marks or a hedging word there is no scare quote. So-called introducing a thing's informal name, which the writer then uses plainly, is not this. It is this when the writer's own sentence holds a term at arm's length with marks or a hedge and no one is being quoted for it: so-called protected areas, self-identifying men of God, cry 'fraud', 'self care' as a marketing word, claims or claimed on a non-criminal assertion to signal disbelief.
{
"true": "In the writer's own sentence a term is held at arm's length with quotation marks or a distancing word, so-called, self-styled, claims, purported, to cast doubt on it, and no named source is being quoted for it",
"false": "Quotation marks cite words a named source actually used, or mark slang, jargon, a title or the writer's coinage; hedges are legal convention or attributed; contempt in plain words without marks is not this"
}Judge only the writer's own wording, not quoted sources. Does the writer soften something harsh with a milder or more technical term where the plain word was available: collateral damage for civilian deaths, enhanced interrogation for torture, passed away in a crime report, downsizing for sacking, adjustment for a cut, incident for an attack? Official terms quoted from a source are the source's. A plain word used plainly answers no.
{
"true": "At least one harsh fact is carried by a softening or technical term in the writer's voice where the plain word fits",
"false": "Harsh facts are named plainly, or the softer term is a source's"
}Judge only the writer's own wording, not quoted sources. Does the writer use a harsher term than the plain one where the plain one fits: regime for government, propaganda for advertising, cult for movement, thug, gulag, junk for a policy, death tax, to lower a subject? The harsh twin of euphemism. A term the subject uses of itself is not this. One harsher-than-plain term in the writer's own voice, headline included, is enough: toxic for unpopular, lawfare for prosecution, spying for collecting, baby mama for partner, haters for critics, hiding for campaigning elsewhere, predators for defendants.
{
"true": "At least one term in the writer's own voice is harsher than the plain word that fits the same fact, lowering its subject",
"false": "The writer's own terms are the plain ones; any harsh words are quoted, self-applied, sports hyperbole, mockery without renaming, or literally accurate"
}Judge only the writer's own wording, not quoted sources. Does the piece lean on virtue words and glittering generalities, freedom, the American people, common sense, our values, real families, justice, decency, security, used to bless a position without saying what it does? The words must carry weight in the piece's case, not appear once in passing. Virtue words inside quoted tweets, posts or statements belong to the speaker, and words in an appended promo or subscription note are not the piece. Pejorative labels for the other side, woke, radical, evil, are loaded language, a different device; this one blesses the writer's own position with praise words. It is this when the writer's own case for a position is carried by the virtue words themselves, with nothing concrete said about what the position does. One passage where the praise words do the arguing is enough, even if specifics appear elsewhere in the piece.
{
"true": "The writer's own case for a position is carried by virtue words, freedom, values, common sense, decency, with nothing concrete said about what the position does",
"false": "The case rests on specifics, or the virtue words are quoted or in a promo note, or the charged language is pejorative rather than praise, or no position is argued"
}Judge only the writer's own wording, not quoted sources. Does the piece rest on deliberately vague phrases that let the reader supply the meaning: reform, change, common sense, the right thing, extremism, values, security, used without saying what is meant at a point that matters? Doublespeak, language that obscures what is being described, counts. A term defined or made concrete in the piece answers no.
{
"true": "Key claims rest on undefined, elastic terms the reader must fill in",
"false": "Key terms are made concrete, or the vagueness is quoted"
}Judge only the writer's own wording, not quoted sources. Is one charged phrase, slogan or characterisation, the same words or near-same words, repeated three or more times in the writer's own prose as a device, so that the reader remembers the phrase rather than the argument? Not this: the piece's subject nouns and names however often they recur, including a policy's name, a product, an event, a place or a coined term the article is about; the headline echoed once in the lead or in an image caption; a phrase inside quotes, tweets or a document, even when the writer quotes it repeatedly; loaded wording that varies from sentence to sentence and never settles on one phrase; a phrase used twice. The device is the deliberate refrain: the same loaded label or line returning at intervals through the body where varied wording would have served. Most repetition in news is the subject recurring; answer yes only when you can quote the refrain and count three uses in the writer's own voice.
{
"true": "The same charged phrase or characterisation returns three or more times in the writer's own prose, spaced through the piece as a refrain",
"false": "The repeats are subject nouns or names, a headline echoed in the lead or captions, quoted material, varied loaded wording, or fewer than three uses"
}Judge only the writer's own wording, not quoted sources. Is a person or group described in ways that make them subhuman, monstrous, wholly evil or defined by a stereotype of their kind: vermin, animals, a plague, an invasion, monsters, or a nationality, religion or class characterised by its expected traits? Reporting that a source said such things is the source's. A label alone is name-calling; this is the portrait.
{
"true": "A person or group is portrayed as monstrous, subhuman or wholly typified by a stereotype in the writer's voice",
"false": "People and groups are described as people, however sharply criticised"
}Judge only the writer's own wording, not quoted sources. Is a serious event or finding written smaller than the facts warrant: a death toll, a loss, a ruling, a failure described in diminishing terms, only, just, a few, minor, isolated, a blip, so the reader takes it as less than it is? The opposite of sensationalism. A minor event reported as minor answers no. A belittling phrase inside a quote belongs to the speaker, and disputing whether a thing happened at all is not minimising it. It is this when the writer's own words shrink what is alleged or found: a grave allegation recast as an encounter, a situation or a stupid thing; a harm to many as hiccups, quibbles or growing pains; a large rise or loss as not the worst, hardly news or a rounding error. One such passage in the writer's own voice is enough.
{
"true": "The writer's own wording shrinks a serious allegation, harm or finding below what the facts warrant",
"false": "The framing matches the size of the event, the belittling is quoted, the writer disputes the facts rather than shrinking them, or the event is minor"
}Is the piece built to make the reader angry at someone: a villain and a provocation, wording that invites scorn before it informs, someone slammed, exposed, caught, shameless, humiliated, mocked, or an outrageous act foregrounded with the response to it as the story? Reporting that someone else was angry, in their words, is not this; scorn that lives in quoted pundits or in the subject's own anger is not the writer's. An adversarial or partisan column that builds its case from facts, figures, bills, votes or history answers no, however harsh its labels: argument is not scorn. Comedy, affectionate ribbing and a sports post-mortem built on performance answer no. It is this when the writer's own words do the shaming, blasted, rant, bizarre, shameless, caught, silent after, or when the piece is an outrage round-up: a provocation and the backlash to it are the whole story and little else is informed.
{
"true": "The writer's own framing makes anger or scorn at a person or group the pull: shaming words of the writer's own, or a provocation and the backlash to it as the whole story",
"false": "The piece argues its case from facts however harsh, or reports wrongdoing plainly and leaves the reader to judge, or the scorn is quoted or is the subject's own, or the piece is comedy"
}Does the piece work on the reader's fear: a threat framed as larger, nearer or more certain than what is reported, what could happen to you, danger words the sources did not use, a worst case treated as the likely case? A warning attributed to a named source and reported plainly is not this. A report of a real danger in proportion, a storm, a recall, an attack, is not this. Fear words inside quotes or attributed to named sources, officials, passengers, experts, commentators, are theirs, not the writer's. A frightening event reported with its facts and attributions in proportion, a stabbing, a hijacking, a breach, a storm at the forecaster's level, is not this however grim the facts. A harm argued from cited studies or figures is not this. Contempt, insinuation or derision aimed at a person or group is a different device. A sports column's alarm about a team is not a threat to the reader. It is this when the writer's own words add danger the sources did not: a threat to you, danger vocabulary layered on reported incidents, a worst case stated as likely, a danger kept alive against the piece's own data.
{
"true": "The writer's own words add threat beyond what the sources and facts support: a danger made larger, nearer, more certain or more personal than reported",
"false": "Threats are quoted or attributed, or reported in proportion to the facts however grim, or argued from cited evidence, or the heat is contempt rather than fear"
}Compare `title` with `text`. Does the headline withhold or tease the thing the piece exists to tell, to earn the click: a question the text answers, here is why, you will not believe, this one thing, a number or name teased and not given, a claim the body walks back? A headline that says what the piece delivers, even vividly, answers no. An explainer or market wrap that names its subject plainly answers no. The common forms: a person named only by category, rock icon 64, Emmy winner, Hollywood legend, when the body names them at once; a figure replaced by an adjective, jaw-dropping price; a response, message or verdict announced but not given, has a blunt response, the verdict is in; a development described only as a turn, a twist, a bombshell; a counted list whose items are withheld, three celebrities, nine stocks; a question whose answer and subject the body supplies; a headline claim the body only partly supports. Not this: a person given by age and role in a crime, court or accident report, man 22, woman 68, driver, which is convention, not a tease; the withheld-name form needs a public figure or celebrity the body names at once. A headline that gives the news and leaves only a name or a detail for the first line is not withholding: if the substance is in the headline, a missing name is not a tease. A numbered recap or list whose headline names the subject and outcome, three observations from a loss, is not a counted-list tease when the body delivers the items. A how or why headline that names the subject and the outcome is an explainer, not a tease. A service or how-to headline that names the event; a live-updates or rolling-coverage label; a rhetorical question that is the column's own stated thesis; a review headline that gives its verdict. One withheld item is enough.
{
"true": "The headline withholds a public figure's name, a number, a quote, a verdict or a development that the body supplies at once, teases a counted list or a question the body answers, or claims more than the body supports",
"false": "The headline names what the body delivers, including service, live-coverage, review, crime-report and thesis-question headlines"
}Judge the writer's own framing and the headline in `title`. Is a person's suffering, ordeal or kindness the pull of the piece, the tearjerker, the vignette that puts a face on it, in place of the wider facts a reader would need? A report that tells one person's story as part of a fuller account answers no. A profile or obituary answers no. Grief, heartbreak or distress inside a quoted statement, post, interview or as-told-to account belongs to the speaker; judge what the writer's own sentences and headline lean on. A death, a loss or an injury that is itself the news, reported with its facts, a tribute, a memorial, a death notice, is not this. Outrage, ridicule, revulsion, alarm or hero praise is a different pull from pity. A person's ordeal set beside the record, the lawsuit, the statute, the police statement, the other side's response, the data, is one story inside a fuller account and answers no. It is this when the writer's own framing makes the suffering or the kindness do the work and the wider facts a reader would need are left out. A profile of a person, living or dead, an as-told-to account, a first-person column that rests on figures or expert points, is not this even when the writing is warm. A headline that leads on the emotion, harrowing, stunned, heartbreaking, final words, with the writer's own sentences amplifying the terror, grief or warmth while the facts behind it stay thin, is this. Answer yes only when you can point to the writer's own tearjerker sentences and say which facts they stand in for.
{
"true": "The writer's own framing and headline make a person's suffering or kindness the pull, and the wider facts a reader would need are left out",
"false": "The emotion is quoted, or the loss is itself the news reported with its facts, or the story sits inside a fuller account, or the pull is outrage, ridicule or praise rather than pity"
}Judge only the writer's own wording, not quoted sources. Is a position justified by patriotism or loyalty to the group, as American, as un-American, what patriots do, for the country, our nation's values, rather than by what the position does? A quoted politician's flag-waving is the politician's. Flag-waving inside a quote, an embedded post, a press release or the outlet's own subscription blurb belongs to its author, not the writer. It is this when the writer's own sentences make loyalty the reason the position is right, with the merits waved off or absent: loyalty to the nation, the West, a people, its heritage or its values, sovereignty for its own sake, what our forebears stood for. One passage of the writer's own that argues this way is enough; the rest of the piece need not. A closing line that turns the recommendation into a loyalty test counts.
{
"true": "The writer's own sentences offer loyalty to the nation, a people or its heritage as the reason a position is right, in place of its merits",
"false": "The appeal is quoted or in boilerplate, or the writer's case rests on facts, law or effect with patriotism as flourish, or there is no such appeal"
}Judge only the writer's own wording, not quoted sources. Does the writer present themselves or their side as ordinary people against the elite, the experts, the insiders, real folks, working families, people like us, and rest the case on that identification rather than on the merits? Reporting a politician's plain-folks appeal is the politician's.
{
"true": "The writer's case rests on being one of the ordinary people against the elite",
"false": "No such appeal in the writer's voice"
}Judge the headline in `title` and the writer's framing. Are bodies, attraction or sexual interest the pull: stuns, flaunts, sizzles, a revealing outfit, the photo the piece exists to show? A report on a sexual-assault case or a policy on sex is not this. Attraction talk inside a quote, a tweet or a hot-mic clip belongs to the speaker; a piece that reports someone's leering, a racy ad or an affair as the news, or criticises it, is not this. Appearance scrutiny that is not attraction, plastic surgery, ageing, weight loss, looking unrecognisable, a body mocked to belittle, is not this. A fashion, red-carpet or runway report that describes clothes, fabric and designers without offering the body as the draw is not this. It is this when the writer's own headline or captions sell the look: stuns, dazzles, flaunts, sizzles, gorgeous, the revealing outfit or the body itself offered as the reason to look, even on a routine story. Answer yes only when you can point to the writer's own headline or caption that sells attraction.
{
"true": "The writer's own headline or captions offer bodies or attraction as the reason to look",
"false": "The attraction talk is quoted or is the misconduct reported, the scrutiny is surgery, ageing or ridicule, the piece is a fashion report, or the subject is reported for what happened"
}Judge only the writer's own wording, not quoted sources. Is the past invoked as better, simpler or lost, as the pull of the piece: the good old days, what we have lost, how it used to be, in place of an account of the present? A retrospective that is about the past by design, an anniversary, a history, answers no unless it argues the present is worse. Longing inside a quote, a lyric, a press release or a quoted critic's words belongs to the speaker, not the writer. A decline reported as measured change, with figures and causes, is not this; nor is a past figure used as a baseline or as evidence for a judgment about the present. One then-and-now aside or opening anecdote inside an account of the present is not this. It is this when the writer's own voice says the past was better and the present has fallen from it, and that contrast is what the piece asks the reader to feel; a few sentences of the writer's own can carry it. A sports column, a feature or a listicle counts as readily as an opinion piece: gone are the days, it used to be, it will always be the best, in the writer's own voice with the present set lower, is the device even in a light piece. If the piece would stand unchanged with its then-and-now line removed, the past is not its pull.
{
"true": "The writer's own voice frames the piece as a fall from a better or lost past",
"false": "The longing is quoted, the decline is measured change or a baseline, the past is one aside, or there is no such appeal"
}Is the piece promotion dressed as news: a product, service, company, show or event presented on its own terms, with the maker's claims carried as fact, launch-day or availability framing, and no independent voice, customer, critic, rival or figure that was not supplied by the subject? A report on a company that includes its results, a dispute, a regulator, a rival or an independent assessment is not this. A review that judges is not this. Judge whether the reader is being sold to rather than informed. Announcements count: a tour, album, show, signing, appointment, launch, groundbreaking, programme or event relayed with its dates, on-sale or pre-order details, supplied bios or superlatives, and only the subject's own voices, is this whatever the outlet's beat. A how-to-watch or where-to-buy piece that routes the reader to named services, and a teaser whose body is the publisher's own subscription pitch, is this. A trade, arts, sports or local outlet covering its own industry is not exempt. It is not this when the writer adds an independent assessment, a critic, a rival, a cost question, a user or a dispute, or when the claim at issue is research or politics rather than a commercial offering. One piece with no voice but the seller's and the thing's availability as the news is enough.
{
"true": "The piece carries a product, show, event, company or appointment on its own terms, with launch or availability framing and no voice that the subject did not supply",
"false": "An independent voice, assessment, rival, cost question or dispute appears, or the subject is research or politics rather than a commercial offering"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument answer a weaker, cruder or distorted version of the other side's position than the one it actually holds, and treat defeating that version as defeating the other side? Harshness, ridicule, name-calling and attacks on motive are different devices; the straw man needs the other side's position restated in a weaker, cruder or different form and that version answered. Engaging the opponent's own documented words or stated rationale, however hostile the verdict, is not this. Leaving the other side's best defence unmentioned is omission, not distortion. A piece that relays a speaker's caricature of opponents in quotes makes no argument of its own. It is this when the writer's own words put a position in the other side's mouth that its holders would not recognise, they want open borders, they think men can get pregnant, nobody is saying, and then defeat that version in place of the real one. Answer yes only when you can point to the writer's own restatement of the other side's position.
{
"true": "The writer's own argument restates the other side's position as something weaker, cruder or different from what its holders say, and answers that version in place of the real one",
"false": "The other side's actual words or rationale are engaged, or the piece offers only harshness, motive attacks or omission, or the distortion is quoted from a source, or no argument is made"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument discredit a person's motives, character, affiliations or funding in place of answering what they said? Attacks inside quotes or attributed to named sources are theirs, not the writer's; a piece that relays others' insults makes no argument of its own. An investigative report whose subject is the ties themselves, documented and with the subject's response, is not this. But when the writer uses someone's donors, funding, lineage, employer or associations to dismiss what they argue, their donors make them advocates, money bought his position, she was handed everything, he sold his soul to the unions, that is the device, whether or not a rebuttal also appears. Mockery or harsh labels beside a substantive rebuttal are decoration, not this; the test is whether the rebuttal is there. It is this when the writer's own voice explains away what someone said by their motives, hypocrisy, funding, affiliations or character, the government acted to misdirect, the policy comes from hatred, the author writes to please ideological editors, and the merits go unanswered or get less weight than the imputed motive.
{
"true": "In the writer's own voice, what a person or group argues is dismissed by their motives, character, affiliations, lineage or funding, and that dismissal carries more of the argument than any answer to the merits",
"false": "The merits are answered and any jab is decoration, or attacks are quoted or attributed, or the ties are the documented subject of a report rather than a reason to dismiss a claim, or no argument is made"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument meet a charge or a criticism by pointing at the accuser's own conduct, or someone else's, instead of answering it? A counter-charge inside a quote, a statement or an interview belongs to the speaker. A piece that answers the charge on its merits and adds a jab alongside is not this; the jab has to stand in place of the answer. It is this when the writer's own side stands accused and the writer's reply is what the accuser did, or what someone else did, with the charge itself left unanswered. One such reply in the writer's own voice is enough, even if the piece makes other points.
{
"true": "The writer's own reply to a charge is the accuser's or someone else's conduct, and the charge itself goes unanswered",
"false": "The charge is answered on its merits, or the counter-charge is quoted, or no argument is made"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument present two options, or two sides, as the only ones when more exist: good against evil, with us or against us, one cause for a problem with several? Us-against-them or good-against-evil framing inside quotes or attributed to sources is theirs, not the writer's; a report that relays others' claims makes no argument of its own. Arguing hard for one option while naming or conceding alternatives is advocacy, not a false dilemma. A decision that genuinely has two courses, attend or skip, pass or veto, is not false. Harsh criticism piled on one side is a different device unless it also narrows the options. It is this when the writer's own argument allows exactly two outcomes, sides or causes where the piece's own facts show more: no other reason than, either this or that, those who see it and those who refuse to.
{
"true": "The writer's own argument reduces a matter with more options or causes to exactly two, or to one cause, and the conclusion rests on that narrowing",
"false": "The range of options is preserved or conceded, or the two-way framing is quoted or attributed, or the matter genuinely has two courses, or no argument is made"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument discredit an idea, a person or a group by who else holds it, who funds it or who would benefit, rather than by what it is? The association must be the writer's own argument: a speaker's or candidate's linkage reported in quotes is theirs, even when the writer does not challenge it. Charging hypocrisy or inconsistency, you do the same thing yourself, is a different device. A passing partisan label or one-line jab, Trump lite, does not make the argument associative. It is this when the writer's own paragraphs chain the target to other people to discredit them, a hijacker who shared the view, a pastor's heresy, who hosted the fundraiser, a segregationist lineage, a policy that follows Putin's lead, even when substantive reasons appear alongside. Answer yes only when you can point to the writer's own sentences where who holds, funds, admires or benefits from a position does the discrediting.
{
"true": "The writer's own argument discredits an idea, person or group by who holds, funds, admires or benefits from it, and that link does real work even beside other reasons",
"false": "The case rests on content, conduct or inconsistency alone, or the association is quoted from a source or is a passing jab, or no argument is made"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument claim that a step will lead inevitably to an extreme outcome without giving the mechanism by which it would? An escalation inside a quote, a transcript or a roundtable belongs to the speaker, even when the writer does not challenge it. A backward-looking claim that a past step caused a past outcome is post hoc reasoning, a different fault; this one projects forward. A direct consequence claim with its mechanism or named actors, repeal X and Y follows because Z, is an argument, not a slide. A chain the writer attributes to opponents and criticises, or plays as satire, is not the writer's own. Exaggeration about one effect is not this unless it climbs from a modest step to an extreme endpoint with the steps between left out. It is this when the writer's own sentences say a step will inevitably end in collapse, crisis or the end of something, with no path given.
{
"true": "The writer's own argument says a step leads inevitably to an extreme endpoint, collapse, crisis, the end of something, with no path given",
"false": "The chain is quoted or attributed to opponents, or looks backward at a past cause, or gives its mechanism, or no argument is made"
}Judge the argument the writer makes in their own voice, if the piece makes one; a report or a straight explainer that makes no argument answers no. Disagreeing sharply is not a fault. Does the argument answer with something compelling but beside the point, changing the subject from the matter at issue?
{
"true": "The argument's force comes from a point that does not bear on the question",
"false": "The argument stays on the question, or no argument is made"
}Judge only the writer's own wording, not quoted sources. Does the piece urge the reader to a position because everyone is taking it, the tide has turned, the momentum, the inevitable, join the winning side, rather than on its merits? Momentum talk inside a quote or an embedded post belongs to its author. A forecast reasoned from evidence, a measured trend reported with its figures, or a winning-side flourish on a case made on its merits is not this. It is this when the writer's own sentences make the crowd the reason: everyone agrees, the majority wants it, the tide has turned, every outlet now admits it, the others have already done it, do not be left behind, this is out and that is in, or a holdout is urged to catch up. One such passage in the writer's own voice is enough, in a sports column or a trend piece as readily as in politics.
{
"true": "The writer's own sentences offer popularity, consensus or momentum as the reason to hold a position",
"false": "No such appeal in the writer's voice, or the momentum talk is quoted, a reasoned forecast or a measured trend"
}Judge only the writer's own wording, not quoted sources. Does the writer close off a question with a thought-terminating phrase: it is what it is, both sides do it, the science is settled, that's just politics, there is no alternative, end of story, used to end discussion rather than to make a point? The forms are many: period, full stop, end of story, deal with it, it's not complicated, not rocket science, that argument is settled, there's nothing more to say, a win is a win, they are what they are, you have no choice, it was never on the menu, who cares, we all know what's really going on, or a proverb dropped where an argument would go. One such phrase in the writer's own voice, used to shut a live question rather than answer it, is enough, in a sports column or a trade piece as readily as in politics. A self-protective disclaimer, a phrase the writer quotes or examines, or a closing line after the case has actually been made is not this.
{
"true": "The writer's own stock phrase shuts a live question in place of an answer",
"false": "No such phrase in the writer's voice, or it is quoted, examined, a disclaimer, or follows a case actually made"
}Judge only the writer's own wording, not quoted sources. Does the piece borrow the standing of a symbol, institution or revered figure to bless or damn a position: the flag, the founders, the constitution, scripture, a national hero, science itself, invoked for their authority rather than for anything they said or did about the matter? A direct quotation or a relevant precedent is not this. Symbol and founder language inside a quote, a speech, a press release or a candidate's own answers belongs to the speaker. The constitution, a statute or a court ruling named as the actual legal basis of the matter in dispute is relevant, not borrowed; so is a revered figure or institution that is the subject of the story, or one cited for something they actually did as precedent or illustration. A named study with its findings is evidence, not the prestige of science. An idiom, gospel according to, or an insult is not this. It is this when the writer's own sentences invoke the flag, the founders, the constitution, scripture, science, a hero or a national tradition for their aura, on a matter they said nothing about, so that the reverence does the arguing. Answer yes only when you can point to the writer's own sentence that borrows the standing and say what position it blesses.
{
"true": "The writer's own sentences invoke a symbol, revered figure, scripture, science or tradition for its aura to bless or damn a position it said nothing about",
"false": "The invocation is quoted, or is the actual legal basis or subject of the matter, or cites what the figure did as precedent, or is evidence, idiom or insult"
}Consider the people, organisations and documents the piece quotes or cites. Where the piece concerns a contested question, one with two or more sides that disagree, are the sources all on one side, with no response, rebuttal or countervailing source given? A report on an uncontested event, a match, a crash, a weather warning, a product launch, a research finding, has no sides and answers no, even if only one person is quoted; few sources is thin sourcing, not one-sidedness. The other side counts as answered if its denial or response is cited, if the piece notes it declined or did not answer a request for comment, or if the writer states its position. A column may argue one side; it still answers yes if it cites only its own side.
{
"true": "The piece concerns a contested question and every quoted or cited source supports one side; the other side is absent or present only as a paraphrased target, with no cited response and no noted request for comment",
"false": "Sources from more than one side are quoted or cited, or the other side's response or refusal to comment is noted, or the writer states the other side's position, or the piece concerns nothing contested"
}Does the writer rest claims on unnamed authority: experts say, critics argue, sources say, many believe, it is understood, or a passive with the actor deleted, it has been decided, mistakes were made? A named person, office or organisation is not unnamed. One conventional attribution in a wire report, officials said, a spokesperson said, is not enough; the device is when unnamed authority carries a contested claim or the piece's main claims. Reportedly, according to the report, a source told NBC, are not this when the report or outlet is named: the authority is the named outlet, however anonymous its own sources. Conventional crime and court attribution, police sources, a law-enforcement source, counts as one conventional attribution. A writer's own open assertion is not unnamed authority. It is this when the writer's own claim leans on an unnamed collective or an absent actor: polls show without a pollster, Democrats believe, social media users criticised, many blame, experts say anecdotally, research has found without a study, is claimed to have, is said to, is considered the favourite. Judge the writer's own sentences: count the vague attributions and ask whether a contested claim would fall without them.
{
"true": "A contested or central claim of the writer's own leans on an unnamed collective, an unnamed expert or source that belongs to no named outlet, or an actorless passive, or several claims do",
"false": "Claims are attributed to named people, bodies or outlets, including a named outlet's own anonymous sourcing, or stated openly as the writer's own, or the only unnamed attribution is a single conventional one"
}Is the piece built on another outlet's reporting: the central facts credited to what another publication, broadcaster or site reported, with no independent confirmation or added reporting by the writer? A wire story credited to its agency is not this. Primary sources are not another outlet: a public hearing, a court filing, a study, a company or agency announcement, a tweet, speech or broadcast the writer quotes directly, the outlet's own records or prior reporting, a syndicated byline. A scoop credit, first reported by, followed by the writer's own detail or a sought comment is not this either. It is this when the facts, quotes or footage were obtained by a different outlet, an interview given to another publication, another station's report, another reporter's embedded post, and the writer adds commentary only. A brief reaction piece to a named report is yes. Answer yes only when you can name the other outlet whose reporting the piece rests on.
{
"true": "The central facts, quotes or footage were obtained by another outlet and are relayed without independent confirmation or added reporting",
"false": "The piece reports from primary sources or its own reporting, or adds its own reporting to a credited scoop, or is a wire story credited to its agency"
}Do the quoted sources do the writer's arguing? The device is quotes selected and arranged so that the piece makes an argument the writer never states in their own voice: several quotations all pressing one conclusion, a single expert or witness whose quoted words evaluate and judge while the writer only introduces them, or quotes cropped to a punchline. Quotes from the story's own principals stating their claims, positions, reactions or accounts answer no, however one-sided: a single-subject interview, a lawsuit's allegations, a press conference, an announcement, an eyewitness account. There the quote is the news, not a stand-in for the writer. If the writer states the conclusion plainly in their own voice, the quotes are support, not ventriloquism. The device needs third-party or evaluative voices arranged to deliver a verdict the writer avoids stating. A report that quotes sources for facts, events and their own positions answers no; so does a piece that quotes a range of views. Judge the pattern of quotation, not any single quote.
{
"true": "Third-party or evaluative quotations are arranged to deliver a verdict the writer never states in their own voice, all pressing the same way; they are not simply the story's principals stating their own positions",
"false": "Quotes give facts, events, the principals' own claims and positions, or a range of views, or the writer states the argument openly in their own voice"
}Does the piece report a scientific or medical claim without the weight the evidence carries: a single study, a preprint, a press release, an animal or lab result, or a lone expert reported as a finding, with no sample size, replication, peer review or consensus mentioned? A piece that gives the evidence its due weight answers no. A piece with no scientific claim answers no. A claim inside a quote, a post or an advocacy group's report, attributed in the writer's own sentences as what the source says or claims, is the source's, even when the writer does not rebut it. Settled textbook medicine, an incubation period, a standard description of a disease, a treatment guideline, is consensus, not a single finding. A regulatory approval reported with its hedge, a study named with its journal, sample or limits, or competing studies set side by side, gives the evidence its weight. Economic, political and technological predictions are not scientific claims. It is this when the writer's own sentences state a study, survey, lab result or lone expert's claim as established fact, with no source named, no size, no hedge and no other view.
{
"true": "The writer's own sentences state a single study, survey, lab result or lone expert's claim as established fact with no source, size, hedge or other view",
"false": "The claim is attributed to a quoted source, or is settled consensus, or is reported with its journal, sample, hedge or competing evidence, or no scientific claim is made"
}Is a political or policy matter framed as a race: who is ahead, polls, momentum, strategy, fundraising, the odds, rather than what the candidates or measures would do? Sports is a race by nature and answers no. A business or product rivalry answers no. A piece that reports a poll and also the substance answers no. Substance means what a candidate, party or measure would do, stated in the writer's own text. A passing nod to an issue, a label like socialist or moderate, or a policy mentioned only as campaign messaging or as the explanation for a poll shift is not substance. Race ratings, seat math, approval numbers, ad spending, turnout tactics and predictions of who wins are race framing, and a piece built on them with the substance missing is this, even when it also reports an event or a rally. A race line as background to a different story, a resignation, a licensing dispute, a gaffe, is not this. Certified results and turnout are outcome reporting, not this. Reporting on campaign finance or a vendor's practices is accountability, not this.
{
"true": "The writer's own text covers a political matter as who is ahead, polls, ratings, strategy or odds, and never says what the candidates or measures would do",
"false": "The writer states what the candidates or measures would do, or the race is background to another story, or it is outcome, finance or sports reporting"
}Is a verbal or political dispute, a feud, an exchange of attacks and comebacks between public figures, parties, outlets or celebrities, presented as the story itself, who hit back, who got wrecked, who is furious, while the matter they disagree about goes unexplained? A physical fight, a shooting, a stabbing, a brawl, an assault, a war or any violent incident is an event, not this, however dramatic the account. Sports and court cases are contests by nature and answer no. A report that explains what the dispute is about, the policy, the claim, the record, the vote, answers no. It is this when the writer's own framing is the trading of jabs, insults and comebacks, and a reader finishes without knowing what is at issue.
{
"true": "A verbal or political feud is presented as the story, who attacked, who hit back, who is furious, and what they disagree about goes unexplained",
"false": "The dispute's substance is explained, or the piece reports a violent incident, a sport or a court case, or there is no clash"
}Does the piece claim a trend or a wider condition, growing, increasingly, more and more, a crisis, a pattern, part of a larger problem, and support that claim with nothing but the stories of one or a few named people or incidents? Answer yes only when both hold: an explicit or clearly implied claim about many, and evidence consisting only of individual cases. A report of one event that claims nothing beyond it answers no. A piece that gives any figure, study, count or official assessment for the wider claim answers no. A warning or forecast from a named body is not an anecdote. A trend or pattern claim voiced by a quoted source, interviewee or advocate is theirs; the writer's own text must make the claim. A profile, interview, review or report confined to one person, one event, one roster, one investigation or one dispute makes no claim about many. A wider claim backed anywhere in the piece by a count, poll, statute, official finding, tally or study answers no even when anecdotes lead. It is this when the writer's own sentences assert a wider condition, growing, more and more, across the country, women are dying, voters are turning, and the only support given is one or a few named people or incidents.
{
"true": "The writer's own text claims a trend or wider condition and offers only one or a few individual cases as evidence, with no figure, study, count or official assessment anywhere in the piece",
"false": "No claim about many is made in the writer's voice, or the trend claim is quoted or attributed, or a figure, study, count, statute or official assessment backs it, or the piece concerns one specific case or set"
}Is the piece carried by what might happen rather than what did: could, may, might, would, what if, raises questions, sets the stage, with speculation as the main content rather than a sentence of context at the end? Analysis openly framed as forecasting counts as yes. A report of a forecast issued by a named body, a central bank, a weather service, a pollster, is a report of an event and answers no. Not this: a prescriptive piece arguing what should be done, however future-facing; reporting of plans, schedules, timelines, procurement documents or announcements, which are events; attributed forecasts, assessments or projections from named bodies, officials, pollsters, forecasters or a meteorologist; impact reporting on a change that has been enacted; a hypothetical hook over a body of things that happened. It is this when the writer's own conjecture is the substance: a game or race previewed by handicapping, a piece carried by what an unbuilt product may do, a scenario of takeover, breakup, collapse or victory built from would and could. A real event or quote as the hook does not save a piece whose body is the writer's what-ifs: if more than about half of the writer's own text is about what could, may or would happen, answer yes.
{
"true": "More than about half of the writer's own text is conjecture about what might happen, whatever the hook",
"false": "The piece reports or explains what happened, was decided, scheduled or announced, or argues what should be done, or relays a named source's forecast; any speculation of the writer's own is brief"
}Does the piece present an event as new or singular, first, unprecedented, never before, historic, a first-of-its-kind, when the same kind of thing has happened before and the piece does not say so? Answer yes only where the writer's own framing makes the novelty claim and a reader of the beat would know it is not new. A piece that says what came before, or makes no novelty claim, answers no. A superlative inside a quote or an embedded post belongs to its author. A measured record, record low, highest since, cheapest ever, with the figure given, is a measurement, not a novelty claim. A piece that names the precedent, even to say this exceeds it, answers no. It is this when the writer's own words call a thing first, unprecedented, never before or first-of-its-kind at a breadth the facts do not support, or widen a source's narrow first into a broad one, and nothing on the page says what came before.
{
"true": "The writer's own words claim a first or unprecedented event at a breadth the facts do not support, with the precedent left out",
"false": "The claim is quoted, is a measured record, is narrowly scoped and true, is a hedged flourish, or the precedent is named"
}Does the writer assert what the public, most people, voters, Americans, everyone, a growing number, believe or want, without a poll, a vote or a named measure to back it? A poll or survey cited by name, a vote result, a market figure, counts as backing. A source quoted making the claim is the source's, not the writer's.
{
"true": "A claim about what most people or the public think is made in the writer's voice with nothing to back it",
"false": "No such claim, or it is backed by a named poll, vote or measure, or attributed to a source"
}Judge only the writer's own wording, not quoted sources. Is an extraordinary event written up as routine: another, as usual, as expected, process and procedure language for something that is not ordinary, so the reader is told nothing unusual has happened when it has? A genuinely routine event reported routinely answers no. Routine language inside a quote belongs to the speaker. Another, as usual or business as usual used to mock, to amplify outrage, as irony or sarcasm, or as comic hyperbole in a feature is the opposite of this. A piece that elsewhere calls the event horrific, explosive, an escalation or a break from the norm has told the reader it is unusual and answers no; so does a piece that names the normalisation as the problem. It is this when the writer's own voice, in earnest, tells the reader a consequential or extraordinary event is business as usual, nothing to see, has always been so or is to be expected, and nowhere flags its weight.
{
"true": "The writer's own voice, in earnest, frames a consequential or extraordinary event as ordinary and nowhere flags its weight",
"false": "The routine language is quoted, ironic or outraged, the piece flags the event's weight elsewhere, or the event is genuinely routine"
}Judge only the writer's own wording, not quoted sources. Is a policy or public matter framed as a zero-sum struggle for power, winners and losers, a blow, a loss, a defeat, where what the policy does goes unreported? A report of an actual loss, an election, a vote, a court ruling, is not this. A piece that reports both the fight and the substance answers no.
{
"true": "A public matter is framed as power gained and lost with the substance absent",
"false": "The substance is reported, or the loss is the actual event"
}Judge only the writer's own wording, not quoted sources. Is a complex matter told as one of the stock shapes: appearance against reality, the little guy against the big guy, good against evil, efficiency against waste, the bizarre against the ordinary, so that the causes, trade-offs or history that a fuller account would give are left out? A false dilemma is one kind; count the others here. A discrete event, an incident, a crash, a crime, a match, an earnings report, an announcement, reported with its facts and open questions is a simple matter however dramatic and answers no. But an incident told as a morality tale, a villain and a victim in the writer's own words with the uncertainties and context dropped, is the device. A stock shape voiced by a quoted source or attributed to others, such as 'hailed as a hero', is not the writer's. If the writer's own text supplies causes, the other side's rationale or complicating facts, the complexity is preserved and the answer is no, whatever the tone. The device needs a genuinely many-sided question, a policy, a dispute, a trend, that the writer flattens in their own words. A simple matter told simply answers no.
{
"true": "A many-sided matter, or an incident with real uncertainties, is told in the writer's own words as one stock story shape, with the causes, trade-offs, context or other side's case that would complicate it left out",
"false": "The matter is a discrete event told with its facts and open questions, or the stock framing is quoted or attributed rather than the writer's, or the writer's own text supplies causes, trade-offs or the other side's case"
}Judge only the writer's own wording, not quoted sources. Scapegoating is blame by category: the writer's own words make a group's identity, immigrants, a religion, an ethnicity, a party, a sex, the explanation for a harm, generalising from one case or none to the whole class, so that whoever actually caused the harm drops out of view. Blaming a named person, office, party or government for a decision it actually took, with reasons, is argued responsibility, not this, however harsh or partisan. Blaming the perpetrators of an act for that act is not this. Reporting a formal finding, a court, an inquiry, is not this. Group blame that appears only in quotes, posts or tweets belongs to the speaker, even when the writer does not challenge it. Insulting labels for a group, or a sneer at the people who actually did the thing, are not this unless the group's identity is offered as the reason the harm happened. Answer yes only when you can point to a sentence in the writer's own voice that makes the group as such the cause.
{
"true": "The writer's own words make a group's identity the cause of a harm, generalising to the class and relieving or hiding whoever is responsible",
"false": "Blame is argued against a named actor or a decision actually taken, or falls on the perpetrators of the act, or is formally found, or is quoted from a source, or no blame is assigned"
}Are the positions in the piece reduced to soundbites: each side given a sentence or a slogan, with no position that takes explaining allowed the room to be explained? A short news brief answers no; the device is a piece long enough to explain that chooses not to. A piece that gives at least one position its reasoning answers no. A letter to the editor, an aggregation item built from a clip and reaction posts, a poll notice or a paywall stub is a brief, whatever it compresses. A piece with nothing contested in it, a reaction roundup, a celebrity or sports item or an incident report, has no positions to explain and answers no.
{
"true": "Positions appear only as slogans or one-line quotes and the piece had room for more",
"false": "At least one position is explained, the piece is a brief or letter, or there are no contested positions"
}Does the writer tell the story of an economy as a whole, a nation's or region's growth, inflation, trade, jobs picture or public finances, only through top-line measures: GDP, trade balances, indexes, markets, yields, earnings, the Fed? That is the device, and a routine data report or a fiscal piece written that way counts. It is not the device when the subject is itself a market, a currency, bond prices, a company or deal, a scheme, a court case or a political fight, with the economy only as backdrop. Nor when the writer's own sentences give any measure as people meet it: a wage, a price people pay, jobs gained or lost, an unemployment rate, benefits, rent, poverty or household income; one such line is enough. An index level or a quoted analyst's aside about wages is not such a line. Answer yes when the economy-wide subject is told from the top and you cannot find such a line from the writer.
{
"true": "An economy-wide story is told only through GDP, trade balances, markets, yields, earnings or the Fed, with no household measure in the writer's own words",
"false": "The subject is itself a market, company, scheme, case or political fight, or the writer gives a wage, price people pay, job count or household measure"
}The site is built to know as little about you as possible. It sets no cookies, writes nothing to local storage, has no accounts, and loads nothing from a third party: everything on a page comes from unbias.me. Links to articles are ordinary links, so clicking one goes straight from the browser to the outlet and is not recorded.
Pages are delivered by Cloudflare, which records the IP address, user agent and requested path of each request in its own logs, and uses them to route traffic, serve cached responses and block abuse. Cloudflare holds those logs under its privacy policy. unbias.me stores no copy of them and has no means of altering or deleting them. The traffic data available to its operator is aggregate counts grouped by field, such as requests per day, per country or per user agent.
Questions can be sent to contact@unbias.me.
The same data is available to AI agents through a read-only MCP server.
Every score is a probability, not a fact. The data is free to use, with or without credit, and without warranty.
unbias.me is a personal project by bitdrift, created as an experiment to make a free anti-bias news aggregator and to experiment with Jev. The longer story is in this post.