What is Cunningham's law?
Engineering culture and software practice
Cunningham's law is the half-serious claim that the quickest way to get a correct answer online is not to ask a question but to post a wrong one, because people who would ignore a request will rush to correct a mistake. It is named after Ward Cunningham, inventor of the first wiki, though the wording and the label came from a former colleague, Steven McGeady, and Cunningham has disowned the popular version.
What this means
The idea behind Cunningham's law is a small, slightly cynical observation about human nature. A question posted online often sits unanswered, because answering it takes effort and nobody feels obliged. A confident but wrong statement, by contrast, tends to draw a swift and detailed correction, because pointing out an error is quick, satisfying and a chance to look knowledgeable. The law dresses that observation up as a rule: to learn the truth, state a falsehood and wait.
It carries a name because the pattern is genuinely useful and genuinely awkward at the same time. It names a tactic people already used without realising, and it puts a label on a real feature of online groups: assertions get engagement, questions get silence. The French have long had a version of the same thought, roughly preach the false to know the true, so the insight is older than the internet.
The label matters mostly as a warning as well as a trick. Used gently, it explains why a rough draft attracts more help than a blank page. Used cynically, it is a licence to waste other people's time with deliberate misinformation. The interesting part is the line between the two.
Why it matters
For any organisation that depends on getting good information out of its own people, the law explains a frustrating everyday fact: silence is not the same as agreement, and a request for feedback often produces nothing while a flawed proposal produces plenty. If your review culture relies on asking politely for comments, you will frequently get none. This is not because people do not care; it is because reacting to something concrete and imperfect is far easier than composing thoughts from a standing start.
That has a direct bearing on internal wikis, knowledge bases and shared documents. These improve fastest when there is something in them worth correcting. A page that is empty or vague invites nobody. A page that states something confidently, even if it is slightly wrong, invites the person who knows better to fix it. Many well-run internal knowledge stores quietly rely on this dynamic, whether or not anyone has named it.
The effect has re-emerged in AI-assisted work. People who would never sit down to write an answer from scratch will happily correct a confident wrong answer produced by an assistant. A draft generated by a model, then circulated for correction, can pull expertise out of a team that a blank request never would. The same asymmetry that governs forums governs the office.
There is a cost, though, and leaders should weigh it. Deliberately being wrong to provoke a correction is a form of manipulation, however mild. It works on colleagues' reflexes rather than their goodwill, and if it becomes an obvious habit it erodes trust. It also fails badly in one particular case, discussed below: when the wrong answer is not corrected at all.
How it works
Where the term came from
Ward Cunningham built the first wiki, WikiWikiWeb, in 1995, and is also the person who coined the idea of technical debt. He designed wikis specifically so that many people could correct and extend a shared text, and in interviews he has described watching pages move toward stability as readers fix what earlier authors got wrong. That interest in how a group repairs a document is the honest core of the law.
The naming, however, was not Cunningham's. His former colleague Steven McGeady attached Cunningham's name to the adage. McGeady posted it on 28 May 2010 as reader comment 119 on the Schott's Vocab weekend competition at the New York Times, phrasing it as the best way to get the right answer being to post the wrong one, and crediting advice Cunningham had given him in the early 1980s at Tektronix with reference to what became Usenet.
Cunningham has pushed back on the version everyone now quotes. He has said he never suggested asking questions by posting wrong answers, and has called the popular line a misquote that, with some irony, disproves itself by spreading across the internet exactly as the law would predict. The honest position is that the observation about wikis is genuinely his, the catchy formulation and the label are McGeady's, and the two men do not fully agree about what was said. Much of this origin trail is documented only on Wikimedia-hosted pages, which are not registered here; the attribution is corroborated in Cunningham's own recorded interviews about wiki design.
Why correcting beats composing
The mechanism is an asymmetry of effort and motive. Writing an answer from nothing requires you to structure a whole explanation, and it offers little reward beyond feeling helpful. Correcting a specific error requires only that you spot the gap between what was said and what is true, and it offers the quiet satisfaction of being right in public. Experimental psychology on how people learn from mistakes, such as work by Metcalfe and Xu on learning from one's own errors and those of others, sits behind the wider point that errors are cognitively sticky and attention-grabbing in a way that open questions are not.
There is also a social dimension. Being wrong in public is cheap for some people, expensive for others. A confident senior figure, or an anonymous account, can float a bad answer and lose nothing. A junior employee whose name is attached, in a culture that punishes mistakes, pays a real reputational price. So the tactic is not equally available to everyone, and a leader who relies on it should notice who in the room can afford to be wrong and who cannot.
How it shows up in real work
The most common appearance is the difference between a blank document and a rough one. Send a team an empty template asking for input and you will wait. Send them a clumsy first draft that gets several things wrong and you will get a flurry of corrections, some of them the exact material you needed. The same is true of straw-man proposals in meetings: a concrete, flawed option moves a discussion that an open question stalls.
Internal knowledge stores show the pattern over time. A wiki page nobody edits is often one that says nothing anyone disagrees with, or nothing at all. Pages improve when they contain a claim specific enough to be wrong. In AI work, the reflex reappears: a model's confident draft answer is a magnet for correction, which is one reason circulating a generated draft can be more productive than asking colleagues to start from scratch.
How experienced teams handle it
Sensible teams use the honest version and avoid the dishonest one. They circulate genuine drafts, clearly marked as drafts, and invite edits rather than praise. They seed a new wiki with real but provisional content so there is something to react to. They frame proposals as straw men that everyone is expected to knock down. Crucially, they do this openly, so nobody feels tricked. The moment the wrongness is manufactured purely to bait a reaction, the tactic shades into bad faith and stops being defensible, especially in any public forum where other people's time and attention are the cost.
Examples
Consider a professional services firm whose partners keep asking associates to review a new client-onboarding note and getting no replies. One associate rewrites it as a confident, deliberately over-simplified draft with a couple of obvious oversights in the compliance steps. Within a day, three colleagues have annotated it in detail, correcting the oversights and adding the nuance the polite request never produced. The draft was rough on purpose, and it was labelled as a draft, so nobody felt manipulated.
Picture a charity running a small internal wiki for its volunteer coordinators. For months the grants page stays empty, because writing it feels like a large task nobody owns. A coordinator finally pastes in a short, partly inaccurate summary of the application process. Almost immediately, two long-serving volunteers correct the deadlines and the eligibility rules they knew were wrong. The page becomes accurate not despite the initial errors but because of them.
Imagine a software team of eight adopting an AI assistant to draft its runbooks. Asked to write a database-recovery procedure from memory, the engineers produce nothing for a fortnight. The team lead instead has the assistant generate a plausible but flawed procedure and posts it in the channel. The on-call engineers, who cannot bear to leave a wrong recovery step unchallenged, correct it within an hour, and the corrected version becomes the record.
Common misunderstandings
People assume Cunningham's law was stated by Ward Cunningham as a law. It was not. Cunningham designed wikis and observed how groups correct each other, but the adage and his name were joined by Steven McGeady around 2010, and Cunningham has publicly rejected the popular wording. Treat it as folklore about group behaviour, not as a signed theorem.
Some read the law as a general endorsement of posting misinformation. That is a misreading. The defensible uses are drafts, straw-man proposals and seeded internal pages, where everyone understands the content is provisional. Deliberately misleading a public forum to extract an answer is a different act, and it imposes real costs on people who did not agree to be used.
There is a tendency to confuse this with the XY problem. The XY problem is about the person asking: they ask about their attempted fix rather than the real problem, so the framing is wrong at the source. Cunningham's law is not about the asker's framing at all; rather, it is about how a crowd responds, and specifically about provoking a correction. One is a fault in the question; the other is a tactic aimed at the answerers.
Many people believe the tactic always works. It does not. In quiet forums, small teams or niche topics, a wrong answer can simply sit there uncorrected, and then it becomes the accepted record, which is worse than silence. The law depends on there being motivated experts who will notice and care.
Finally, some treat it as harmless because it is funny. The humour is real, but the honesty cost is real too. Using engineered wrongness on colleagues works on their reflexes rather than their trust, and repeated use is quickly spotted and resented.
Risks and boundaries
The largest boundary is the uncorrected error. The law assumes a correction will arrive; when it does not, a confident falsehood becomes the default answer, and in a wiki or knowledge base it can persist for years. In safety, legal or financial content this is dangerous, and the tactic should never be used where an uncorrected mistake could cause harm.
The evidence for the law as a strong general rule is thin. It is an aphorism, not a finding, and the one widely reported informal test of it produced mixed results, with the author noting that being deliberately wrong attracted attention as much as accuracy. The underlying psychology of error and correction is well studied, but the claim that wrongness is reliably the fastest route to truth is folklore, and should be presented as such.
There is also a live ethical debate. Provoking corrections from willing colleagues on a clearly marked draft is broadly accepted. Seeding a public forum with fabricated wrong answers is widely regarded as trolling, and many communities treat it as a breach of good faith. The line sits at consent and disclosure: do the people responding understand what they are responding to.
What to do next
Stop asking for feedback on blank pages. If you want input, circulate a concrete draft, mark it plainly as a draft, and invite edits and corrections rather than approval. You will get far more useful reaction from something imperfect than from an open request.
Seed your internal knowledge base with provisional content and say so. A page that states something specific, flagged as a first attempt, will attract the corrections that make it accurate. Empty and vague pages stay empty and vague.
Use the effect honestly in AI-assisted work. Let an assistant produce a first draft, then put it in front of the people who know the subject and ask them to correct it. This turns the reluctance to write into a willingness to fix.
Watch who can afford to be wrong. If only senior people ever float the draft that gets corrected, you are teaching juniors that mistakes are unsafe. Make it explicit that draft errors are expected and welcome, so the correction reflex is available to everyone, not just those with rank.
Never use engineered wrongness where an uncorrected error could cause harm, and never deploy it against a public forum that has not consented to it. Keep the tactic to drafts and straw men among people who understand the game.
FAQs
Did Ward Cunningham actually invent Cunningham's law?
No. He invented the first wiki and observed how groups correct shared text, but the adage and his name were joined by Steven McGeady around 2010. Cunningham has said the popular wording is a misquote and that he never advised posting wrong answers.
Where did the phrase first appear?
In a reader comment McGeady posted on 28 May 2010 on the Schott's Vocab weekend competition at the New York Times, recalling advice Cunningham had given him in the early 1980s at Tektronix.
Why do wrong answers get corrected when questions get ignored?
Correcting a specific error is quick and rewarding, while composing a full answer is effortful and thankless. Errors also grab attention in a way open questions do not, which is one reason the tactic works at all.
Is it ethical to use the tactic deliberately?
It depends on disclosure. Circulating a clearly marked rough draft to willing colleagues is fine. Deliberately posting misinformation in a public forum to bait an answer is widely treated as bad faith and wastes other people's time.
How does this relate to the XY problem?
They are different. The XY problem is a fault in how someone frames a question, asking about their attempted fix instead of the real problem. Cunningham's law is about how a group responds, and about provoking a correction on purpose.
What happens if nobody corrects the wrong answer?
Then the tactic backfires. A confident, uncorrected falsehood becomes the accepted record, which is worse than an unanswered question. This is the law's main failure mode.
Can this help with AI tools?
Yes. People will readily correct a confident wrong answer from an assistant that they would never have written themselves. Generating a draft and asking a team to fix it can surface expertise that a blank request cannot.
Is there any hard evidence the law is true?
Not as a strong general rule. It is an aphorism rather than a research finding, and the best-known informal test of it was inconclusive. The psychology of correcting errors is well studied; the slogan itself is folklore.