There is a member in your community who answers every question within four minutes. They are warm, they are keen, and they are the most active person in the room. You have thanked them publicly at least twice.
They are also wrong about a third of the time. Not maliciously. Not even carelessly. They have read a lot, they have done the thing once, and they deliver every answer with the same unwavering certainty whether it is rock solid or completely invented. Last Tuesday they told a new member something about pricing that would cost her real money if she followed it. You saw it. You did not know what to say, so you said nothing, and the thread got fourteen likes.
This is one of the genuinely hard problems in community building, because the obvious fixes are all bad. Public correction humiliates a person who is trying to help. Private silence lets bad information calcify into house doctrine. And banning your most enthusiastic contributor over accuracy is a wildly disproportionate response to someone who is, in their own mind, being generous.
So let us solve it properly. The good news is that this is a norms problem, not a person problem, and norms are the thing you actually control.
Understand why they do it, because it changes the fix
Confident wrong answers usually come from one of three places, and each one responds to a different nudge.
Some people are pattern completing. They have seen a similar question before, they retrieve the shape of the answer, and their brain fills the gaps without flagging which parts are memory and which parts are guess. This is universal human behavior and it has nothing to do with arrogance.
Some people are answering to belong. Being the helpful one is their role in the room, and roles are load bearing for how people feel about themselves. If answering fast is how they earn their place, they will keep answering fast, and slowing down feels like losing status.
And some people genuinely do not know the edge of their knowledge, which is the hardest version because it is invisible from the inside.
Notice that none of these is bad faith. That matters, because the tone you take will be read as a verdict on their character, and if you get the tone wrong you will lose a person who was trying to be good to your community.
Correct the information, in public, immediately
Here is the rule that took me too long to learn. Correct in public, coach in private. Both. Not one or the other.
The public correction is not optional, and this is where most founders go wrong out of kindness. If bad advice sits uncorrected in your community, three things happen. The person who asked acts on it. Everyone reading learns the wrong thing. And the room learns that accuracy is not really the standard here, which is how a community slowly becomes a place where experienced people stop bothering to participate.
But there is an enormous difference between correcting information and correcting a person. Correct the information and leave the person alone.
The move is to add rather than to negate. "Adding to this: the part about annual billing works differently in practice, and here is what I have seen. If you do X you will run into Y, so most people end up doing Z." No "actually." No "that is not right." No tagging them. You simply put the better answer in the room, with your reasoning, at the same level of specificity.
If the bad advice is actively risky, money, legal, health, safety, be more direct and be quick about it. "Careful with this one, please do not act on it before checking with a professional, here is why it varies." Directness is a kindness when the stakes are real. Do not soften a warning into ambiguity to protect one person's feelings at the expense of another person's outcome.
And thank them in the same breath if you can do it without sounding sarcastic. "Great question to raise, and it is more slippery than it looks." That single clause tells the room that being partially wrong in public is survivable, which is a norm you desperately want.
Then have the private conversation, once
A week or so later, not the same day, send them a message. Not a warning. A promotion, in a sense.
"You are the most generous answerer in the community and I want to talk to you about that, because I think you could be even more useful. Sometimes the fastest answer is a confident one, and in a room where people are about to spend money or make a big decision, confident-but-approximate can cause real problems. Would you be up for a small experiment? When you are sure, answer like you always do. When you are extrapolating, say so, something like 'I have not done this exact thing but here is what I would try.' That one sentence would make your answers more trustworthy, not less, and it would set a tone for everyone else."
Look at what that message does. It leads with genuine praise that is not a sandwich. It names the harm without naming a fault. It gives one specific, tiny behavior change. And it frames the change as a gain in status rather than a loss, which is essential, because status is why they are doing this in the first place.
Most people take this well. Some are briefly embarrassed and then relieved, because operating at maximum confidence is exhausting and nobody had ever given them permission to hedge.
Make hedging high status for everyone
Here is the real fix, and it is not about that one member at all. In most communities, hedging is low status. Saying "I am not sure" feels like admitting you do not belong. So everyone performs certainty, and the loudest certainty wins.
You can invert that, and you do it by example rather than by rule.
Say "I do not know" in public, on purpose, at least once a week. Then say what you would do to find out. This is the single most powerful cultural lever a founder has, and it costs nothing but a small hit to your ego, which frankly could use the exercise.
Use explicit confidence markers in your own answers and watch them spread. "I am certain about this." "I am fairly sure, but check it." "This is a guess, treat it as one." Within a month or two you will see members using the same shapes, because people copy the founder's register far more than they follow the founder's rules.
Praise good corrections publicly. When a member gently corrects another member well, say so. "This is exactly how I hope we do this here." You are training the room in the behavior you want, using the only reward that reliably works in communities: visible approval.
And celebrate someone changing their mind. When a member says "I was wrong about this, here is what I learned," make a small fuss. Once you have made mind changing high status, half of this problem evaporates on its own.
The structural bits that quietly help
A few small things reduce the volume of confident wrongness without any interpersonal work at all.
Write down the three or four answers you are tired of seeing mangled, in your own words, in a document, and link it. Most bad advice in communities fills a vacuum where a canonical answer should be.
Add a line to your guidelines about serious topics. Something plain: in this community, money, legal, tax, health, and safety questions get answered with sources or with "check with a professional," never with confident guesses. Naming the categories does most of the work.
Ask questions in a form that invites experience rather than opinion. If your prompts say "has anyone actually done this" rather than "what do you think," you will get better answers from the same people.
And if you have moderators or helpers, tell them explicitly that adding a better answer is part of the job, not an act of aggression. Most helpers stay silent because they are not sure they are allowed to contradict a popular member.
When it does not improve
Occasionally you will get someone who cannot stop, or who reacts badly to the private conversation and doubles down. At that point it stops being an accuracy problem and becomes a behavior problem, and you handle it the way you handle any other one: a clear, written, private request, then a consequence you have actually stated, applied consistently.
For the specific case of someone giving risky advice repeatedly after being asked not to, you are allowed to be firm. A community where a member is dispensing confident financial or medical guidance against your stated norms is carrying a risk that belongs to you, not to them. Ask once more, plainly, in writing. Then act.
That is rare. In my experience the overwhelming majority of these people are lovely and simply need one sentence of permission to be uncertain in public.
Do this this week
Find the most recent piece of confidently wrong advice in your community and add the better answer underneath it today, in the add-not-negate shape, without tagging anyone. Notice how unremarkable it feels to everyone but you.
Then say "I do not know" somewhere visible this week, followed by how you would find out. Model the thing before you ask for it.
And if you have a fast answerer, send them the private message in about seven days. Lead with the praise, ask for one sentence of hedging, frame it as making them more trusted. You will very likely keep your most generous member and gain a more accurate room, which is a much better outcome than either the callout or the silence.