Ask the question you don’t know the answer to
You state the fact and they dig in. Nine conversations ask instead. It works on one condition the accounts insist on: you have to be able to lose.
Built from people who kept being right and kept losing — thirty conversations, 400 accounts, 2013 to 2025 — and from the nine conversations that asked instead of telling, the three that attached the condition that you must be able to lose, the three that named what it becomes without that, and the eight contraindications that say when not to open your mouth at all.
You are right. You have the fact. You say it, and the other person — partner, parent, colleague — digs in, and now you are both defending a position instead of looking at the thing. The people in these conversations, 400 of them across thirty conversations, argue about that exact moment, and nine separate conversations do something else with it.
They ask. Not ‘why do you think that?’, which two conversations say is heard as an accusation, but ‘how did you get to that?’; or they say the last three words of what the person said back as a question and wait. A person led to the gap in their own reasoning by their own answers will look at it; a person told the fact will defend the wall. One account asks about the criteria — what makes a good version of this — because criteria are shared and cases are not.
Ask the question you don’t know the answer to — and only that one, because if you already know, it is not a question. Three conversations attach the condition and will not let it go: this works only if you are genuinely willing to be the one who was wrong. Three more name what it becomes otherwise — a way of tiring someone out with questions — and two corrections say the method was never a trick for winning but a way for two people to find where they actually part. The one exchange that splits three to three is exactly this, and its decider is your tone, your intent and whether you say their position back in good faith.
The fences are the other half of the page, and there are eight. Not when they are angry or upset — name the feeling first and ask nothing yet. Not about feelings at all, because a question cannot make a feeling wrong. Not with a partner about trust or safety, where it lands as coldness. Not on someone who will not engage, and not, one caution rated high says plainly, on someone cleverer than you. Three cautions say the reaction can be a punch or mockery; one caution says with a trace of condescension it does not work a little, it does not work at all.
Two readers this page is not for. If the person in front of you is upset right now, this is the wrong page for today, and ‘I can see you’re upset’ is the whole of what to say. And if you are asking in order to win, the accounts’ correction is aimed at you: that is the misuse, and it is felt as one.
The people here could not give you a script for someone who refuses to answer, and this page will not invent one; the contraindication there is to stop. Their wordings are here — how, not why; say it back; what would a good one look like — and this page’s own line is marked as its own: before you ask, know what answer would change your mind.
The far side, as this page pictures it, is a kitchen-table argument that ended with one of you saying ‘I hadn’t thought of that’ — and, some of the time, that one being you.
Common questions
Why would a question work where the fact didn’t?
Because the fact arrives as an attack on the person and a question arrives as an invitation to think, and nine conversations say the question works where the fact does not. Nine separate conversations — the widest agreement in these conversations — say that asking questions which lead someone to notice the gap in their own reasoning changes minds where stating the fact or contradicting them does not, because a conclusion you reached yourself is one you own; nine accounts in the prevalence notes recognise the method by its ancient name, and one correction from the history is that his ‘Apology’ was a defence, not a sorry — he defended himself aggressively. Two conversations use a smaller version from negotiation: repeat their last three or four words back to them with a question in your tone, and they will explain themselves further — which is often where the reconsidering starts. Two conversations say the word matters: ‘why’ is heard as an accusation in a good share of conversations, and ‘how’ is not — ‘how did you get to that?’ opens what ‘why do you think that?’ closes. Single accounts add the phrasings that worked for them: ask about the criteria rather than the case — what makes a good president, rather than arguing about the current one — because criteria are shared and cases are not; admit confusion to invite an explanation — ‘I’m not sure I follow, can you help me out?’; start from something you both agree on before you go near the disagreement; at work, ask about the consequence rather than the blame — ‘how should we handle the two deadlines this creates?’; and do not try to finish it in one conversation, because one account says the change happens later, when the person has had time and feels respected. One account who does this for a living says lawyers ask open questions, not leading ones, because leading questions presuppose a position and make you look bad, where open ones show real interest. And one more from a single account, for the reader who will not change this person’s mind however they ask: the people listening in are also deciding, and how you ask changes what they think of you.
What is the condition? It sounds like a trick.
It is a trick the moment you cannot lose, and the people here are firmer on this than on the method. Three separate conversations say the questions only work if you are genuinely open to being persuaded yourself — if the answers could move you; three more name what the method becomes otherwise: a way of exhausting someone with questions until they give up, which has a name online and is felt for what it is. Two corrections in these conversations correct the very framing of the advice: it is a method for mutual understanding and humility, they say, a back-and-forth to find common ground or the foundational disagreement — and it requires you to be willing to be the one who learns. The one exchange in these conversations that splits evenly, three accounts to three, is exactly this: three say it works well and is a standard professional tool, three say it comes across as condescending or manipulative and people just get annoyed — and the accounts’ own decider is your tone, your intent, and whether you actually restate their position back to them in good faith. The cautions rated high are the tests: if there is a trace of condescension in you, it fails completely; if you enter the conversation certain you are right, let them ‘discover’ your truth and ignore any answer that does not fit, they will feel it and you will look like someone with something to hide; if you cannot keep your own feelings out of your tone, it fails; and it needs an open mind on the other side too, which one caution says is rarer than you would like. One account says the practical tell: acknowledge the points that are good, out loud, as they make them, because a questioner who never concedes anything is not asking. One caution says not to make questions the whole meal — a question at the right moment is a tool, and a conversation that is nothing but questions is an interrogation. This page’s own version of the condition, marked as its own: before you ask, know what answer would change your mind. If nothing would, do not ask; say what you think and let them argue back.
When should I not do this at all?
When they are upset, when it is about feelings, and when they are simply not going to play — the people here have eight contraindications and this page carries every one. Not when the other person is angry or emotionally charged: three contraindications say questions then read as dismissive or as a cold tactic, and one account’s order is to name the feeling first — ‘I can see you are upset’ — and only then, later, ask anything. Not when the disagreement is about trust or safety in a relationship: a contraindication says it lands as cold or manipulative when the other person is upset about trust or safety, and a caution rated high says in personal relationships it reads as condescending, and one exchange resolves on whether the partner wants a logical resolution or wants to be heard — and several accounts say they wanted to be heard. Not on feelings at all: two contraindications and one account say facts do not validate feelings and a question cannot make a feeling wrong. Not on someone who reacts to any question with aggression, or whose reasoning is not going to follow the steps — a contraindication says it escalates, and two cautions say people have been punched for it or mocked. Not on someone who will walk away or hang up the moment they detect a technique — a caution says save your time and energy. Not on someone cleverer than you, one caution rated high says bluntly, because they will see the shape of it and name it. And not in a fight about fundamentals — what counts as evidence, who has to prove what — because a contraindication says the method has nothing to grip. Three conversations name where it fails even done well: on someone acting in bad faith, and on someone so invested in a belief that admitting error costs too much — one account calls it sunk cost, another says a deeply held identity — and two more say that once a person recognises the method being used on them, it is over, the way a sales technique is over once you see it. One caution adds family and sensitive subjects as where it comes across as condescending, because they already know you do not believe their premise. One account’s honest alternative for the topic where neither of you actually knows enough: say so, and decline to argue it.
Isn’t it better to just say what I think?
Sometimes, and the people here argue it — including about what the method is even for. One exchange, one account each way, puts questioning against laying out your position plainly: the questioner says it lets the other person arrive at the conclusion and avoids the defensive reaction, the plain-speaker says questioning is pretentious and wastes everyone’s time when you could just make your case; the accounts’ decider is whether the other person is open at all, because the question only helps someone who is willing to walk. A second exchange, two accounts to two, is the same argument in a different room — it worked for its inventor and it is taught in law school, against it is condescending and can hurt — and the decider is the relationship. The bigger argument runs through six exchanges and this page carries it as the accounts left it: is this a way to persuade, or a way to find out what is true? One side says it is a persuasion tool, a way to guide someone collaboratively to the conclusion; the other says that is the misuse — the method is for discovering where two people’s premises part, and using it to steer someone to an answer you have already chosen is manipulation wearing a philosopher’s coat. Two conversations note that its classical form was a teaching method, not a debating one, and these conversations score the truth-seeking side as the better evidenced in the one exchange it scores at all. This page does not settle the definition, and does not need to: the condition in the second question already does the work, because a questioner who could lose is seeking truth whatever they call it, and a questioner who could not is persuading whatever they call it. One account gives the version that survives all of this: sometimes the right move is not to change a mind at all but to listen until you understand what the person is actually saying, and then decide whether it is worth the effort of trying. One conversation says the technique is a formal skill in counselling and in negotiation, which is not this page.
What do I actually say, then?
The accounts gave more principles than lines, and the gaps these conversations name are exactly that — the scripts — so this page passes on the wordings the accounts did give and marks its own. From the accounts: ‘how’ rather than ‘why’ (two conversations); mirroring — repeat their last few words back as a question and wait (two conversations); ‘what would a good one look like?’ about the criteria rather than the case; saying, plainly, that you are not sure what happened and asking them to help you out; ‘I can see you’re upset’ before anything, when they are; and at work, ‘how do you want to handle what this does to the other deadline?’ rather than ‘this is a bad decision’. Two conversations warn about the question that is not a question: anything that presupposes the bad thing — ‘have you stopped doing X?’ — is a trap and is heard as one, and two cautions say phrasing that implies you did not listen, or carries a sneer, ends the conversation. One account says match the register: people who work with systems want the problem, people who manage people want to be heard first. What the accounts could not give you, and this page will not invent: a script for the person who refuses to answer any question at all — the contraindication there is to stop — and a way to tell, in the moment, a fair question from a loaded one when you are the one being asked. This page’s own line, marked as its own, for the second: if the question has an answer that would embarrass you either way, it is not a question, and you may say so. And the sibling pages: for what to do when you think you have been misunderstood, the page on asking what they meant; for the silence after the question, the page on the pause.
What people worked out
Shorter, plainer notes on the same ground — each with the number of people behind it.
Full tip: https://findangel.org/tips/ask-the-question-you-dont-know-the-answer-to · FindAngel.org — free, always.