The 20-minute version
It's Tuesday evening and you're in the car on the way to your in-laws, and you've said the second exit is quicker. Your partner says the third. You explain why the second is quicker, and it's a good explanation, and about halfway through it you remember the roadworks that have been on the second exit since March.
You know, at that moment, that the third is quicker. You've known it for about a second.
What you say next is that the roadworks are nearly finished. You have no idea whether they're nearly finished. Then that it depends on the time of day, and by now you're at the roundabout. You take the second exit and sit in the roadworks, and neither of you says anything for a mile. You arrive four minutes later than you would have and about forty minutes further apart.
Or it's the meeting, where you said the new supplier would be cheaper, someone has brought figures that say otherwise, and you hear yourself saying the figures don't include the onboarding cost. You don't know whether they do. Or it's the group chat, where you said the film came out in 2009, somebody posts a screenshot saying 2011, and you type "pretty sure it was released earlier in some countries". It wasn't.
That's the habit. You said something, somebody pushed back, somewhere in the pushing you saw they had a point, and the next sentence out of your mouth was a defence anyway. I'll call the moment you privately stop believing your own line the turn. What you do after it is either hold, slide quietly round to the other view without saying so, or climb down out loud.
I'm going to argue that holding makes sense from the inside, because backing down feels like losing standing and the feeling arrives faster than any reasoning about it. I'll also argue that the standing you're protecting is mostly not what other people are measuring. There's research on how people judge someone who admits they were wrong, and it points the other way from the feeling. And there's a short sentence that goes with the turn, which gets easier every time you say it.
This isn't about arguing less. Some positions are worth holding under pressure, and there's room here for the honest middle, where you're not wrong yet but you're no longer sure.
Two cautions. If the person pushing you to concede is someone who pushes until you give in, about everything, and then holds what you conceded against you, this isn't about them, and there's more on that near the end. And if you're holding on harder than you used to, snapping at people and unable to let small things go, and it's come with feeling low or irritable most of the day for a fortnight or more, the NHS says to see a GP. Irritability and intolerance of others are on its list of symptoms of depression.
One thing before you read on. For a week, notice the turn without changing anything. When you feel yourself stop believing something you're saying, make a mark on your phone or the back of your hand, and note what you did in the next thirty seconds. Held, climbed down, or slid.
Forty minutes further apart
Nobody holds a line they've stopped believing because they've decided that's the honest course. They hold because at the turn, with the other person waiting, holding is the only move that costs nothing right now.
It seems to keep your standing, and it spares you the discomfort of "you're right", which lands somewhere in the chest and feels for a moment like being smaller. It denies the other person the gloat you can already see forming. It feels like integrity. And it has momentum, because you're already talking, and the next defence is easier to produce than the first silence.
Start with weakness, since the other beliefs sit on top of it.
In 2015 Adam Fetterman and Kai Sassenberg surveyed 281 published scientists, people whose standing depends more than most on being right, about what happens to a researcher's reputation when a finding of theirs fails to replicate. That's about as public a way of being wrong as an academic can manage. The scientists overestimated the damage. And when the researcher admitted the finding was probably wrong, the damage was smaller than when they didn't. Refusing was what hurt.
That's one scenario study of one profession, so hold it lightly. It fits a wider picture, though. Susan Fiske, Amy Cuddy and Peter Glick, reviewing the research in 2007, describe two dimensions on which people size each other up almost at once. Warmth is whether you mean well, and competence is whether you can do what you intend. Now run a climb-down through both. On competence it costs a little, but the other person already knew you were wrong, and saying so didn't change how much they knew it. On warmth it gains a lot, because someone who can say "you're right" isn't going to fight you for the last word. Holding gains nothing on competence, since the wrongness is already visible, and costs you on warmth. I'm extending their framework here, so treat the accounting as mine. The direction is the point.
Then there's the audience. Look at who's really there. In the car it's one person, who's forgotten which exit you argued for by the time the kettle's on, because they're thinking about their mother. In the group chat, seven of the nine haven't opened it. Anna Bruk, Sabine Scholl and Herbert Bless ran seven studies in 2018 on what they called the beautiful mess effect. People imagined themselves or others showing something they'd rather hide, admitting a mistake among them, and judged their own showing much more harshly than other people's. You see your own admission up close, with all the wobble and heat. Someone else's you see from a distance, where it just looks like honesty.
Try it the other way round. Think of the last time someone climbed down in front of you. "Actually, I've looked again and you're right about the dates." You probably thought slightly more of them and forgot it within a day. If you can't think of an example, that's because climb-downs don't register as events. They register as the argument ending.
"They'd never admit it, so why should I?" The grievance is often fair. But their holding is the same habit as yours, and tells you nothing about whether they think they're right. Karina Schumann and Carol Dweck found in 2014 that people who accepted responsibility more readily were the ones who saw personality as something that changes, and their reading is that admitting fault is less threatening if you think you can grow past it. The people who hold hardest may simply be the ones for whom being wrong feels most like a verdict on who they are. And your climb-down is the only move in the room that's yours to make. Either they climb down too, which is more common than you'd think, because you've taken the loss out of theirs, or they don't, and you've still kept the evening.
You said it with some force, maybe twice, and climbing down now feels like wasting all that. Barry Staw described this in 1976 as escalating commitment. Having chosen a course that's going badly, people put more in, especially when they were personally responsible for it so far, because more feels like the only way to make the first choice come right. Your earlier sentences are the investment. The turn is when it turns out to be failing. The roadworks being nearly finished is you putting more in.
What you've already said is gone. If you've taken a cab to the wrong address, the fare's spent whether you get out or sit there with the meter running. The climb-down stops the meter.
Sometimes the hold arrives dressed as humility. You're not wrong, you're just not explaining it well, so you try again, and each version sits a little further from anything you believe. Kunda, reviewing the research on motivated reasoning in 1990, described how people can reach the conclusion they want, but only if they can build what looks like a reasonable justification for it. The third version of your explanation is a harder search for one, and the strain is the sign. Leonid Rozenblit and Frank Keil showed in 2002 that people believe they understand how things work, zips and flush toilets, in far more detail than they do, and the belief collapses when they write out the steps. Philip Fernbach and colleagues found in 2013 that asking people to explain in detail how a policy would work made them less sure of it and more moderate, while asking them to list reasons did nothing.
So try the test. Stop explaining it to them and explain it to yourself, silently, step by step. The actual route, the roadworks, the minutes. If it holds up, offer the better version. If it falls apart somewhere around the roadworks, that was the turn, and it came before the third version.
Ten minutes after the roadworks
Somewhere around the fourth defence, the argument stops being about the roundabout.
You can feel it change. It's now about whether you get bullied out of a view, or whether they always have to be right, or whether you're allowed an opinion in this house. The roundabout is long gone, and you can't climb down from a principle.
Here's how it got there. In 1975 Ross, Lepper and Hubbard gave people false feedback about how well they'd done at a task, then told them, thoroughly, that the feedback had been made up. Their beliefs about their own ability persisted anyway. Beliefs outlive their reasons. Your line was built on the roadworks not existing, the roadworks exist, and the line should have fallen over. Instead it went looking for a new foundation and found the principle. That second argument has nothing in it that could be checked and no natural end. It started at exactly the moment the first one was lost, and it can't survive one plain sentence. "I was wrong about the roundabout" ends the first argument, and the second, with nothing left to be about, ends with it.
Consistency feels like a virtue, and it is one. But there are two things you could be consistent about. One is your conclusions. The other is your method, believing things for reasons and updating when the reasons change. At the turn they come apart and you have to choose. Holding keeps the conclusion. It's consistent the way a stopped clock is.
The fear is looking like a weathervane. Mark Leary and colleagues studied what they called intellectual humility, how far someone recognises their beliefs might be wrong, and found that people high on it were less inclined to see a politician who'd changed position as flip-flopping. The people who'll judge your change of mind harshly are the ones who can't imagine being wrong themselves. Everyone else sees someone who's learned something, especially if the reason is said out loud. "I've just remembered the roadworks, the third is quicker." Nobody hears a weathervane in that.
Then the floodgate. Admit the roundabout and the supplier is next, then the film, then everything. Peter Kim and colleagues found in 2004 that when someone was accused of a competence lapse, getting something wrong, an apology repaired trust better than a denial. Being wrong about a roundabout is a small competence matter. Holding turns it into a question of integrity, because the other person stops wondering whether you knew about the roadworks and starts wondering whether you'll keep saying it when you obviously know. I'm carrying their workplace finding into the car, so that's my reading. Still, think of someone you know who never admits the small ones, and how much weight you give their big ones.
Strip the reasons away and often what's left is a feeling. The climb-down feels humiliating, with heat in the face and a drop in the stomach. Karina Schumann, reviewing why people don't apologise, names the threat to how you see yourself as one of the barriers, and no argument reaches a feeling directly. Looking at it does. How long does it last, the real one, on the occasions you've climbed down? My observation is about a minute. Heat while you say the words, a beat of silence, then "no worries" or a change of subject. Holding doesn't avoid the feeling. It spreads a lower version of it under every defence and on through the silent mile, the car home, and two in the morning when the roadworks come back to you.
The slide feels kinder. You stop defending, never say you were wrong, and a few days later you're taking the third exit as if you always had. Donald Ferrin and colleagues compared apologising, denying and saying nothing much after being accused of a lapse, and found reticence the weakest of the three in two laboratory studies. From outside, the slide looks like someone who either hasn't noticed they were wrong or has and won't say. It also leaves the line alive. Six months later it's "you were the one who insisted" against "I never insisted", and you're arguing about what was said, with no roadworks to check.
You were partly right, probably. The second exit is quicker when there are no roadworks. That's the salvage, and it's holding in a smaller room. Watch for "but". "I was wrong about the roadworks, but normally it's quicker." Everything after the "but" is the part you meant. If the right part really matters, as the headline price might in the meeting, give it its own sentence after the climb-down. "I was wrong about the total. The headline price is still lower, if that matters for next year."
If you're afraid that sounding unsure means nobody listens, there's an odd finding. Uma Karmarkar and Zakary Tormala found in 2010 that speakers seen as non-experts persuaded more when they sounded certain, and speakers seen as experts persuaded more when they expressed some uncertainty. It was consumer research about product reviews, and I'm carrying it further than it was tested. In most of the rooms where you hold, though, it's your area or your family, and people already credit you with knowing things. Certainty is spent each time you use it. Hold on the roundabout, and your certainty about the supplier is worth less next week.
Sometimes the turn isn't a switch to the other view. You've just stopped being sure. Then there's a third sentence, "I'm not sure any more", and it's simply accurate. Tenelle Porter and Karina Schumann found in 2018 that people higher in intellectual humility were more open to hearing the other side's reasons in a disagreement. "I'm not sure any more" is what that sounds like in a car. It turns the argument into a conversation about the route, and the line has been set down, ready to pick up again if the roadworks turn out to be finished.
Online feels worse, because there's a record with your name on it. In four experiments in 2022 with 679 people, Adam Fetterman and colleagues showed people an argument on a social media wall in which one user was shown to be wrong and then admitted it, refused, or didn't reply. The one who admitted it was rated higher on reputation, and people were more willing to interact with them. Everyone can scroll up and see what "pretty sure it was released earlier in some countries" is. They can see "ha, 2011, I was wrong" too. Say it in the same thread, under the correction, and don't delete the original. Deleting is the slide in its online form, and the screenshot already exists.
That leaves the gloat. You can see the smile forming, the "told you" at the in-laws. There's no study for this one, so here's a plain answer. The gloat feeds on a contest. If you held for ten minutes and conceded in the roadworks, it has a lot to work with. If you say the words twenty seconds in, "the third, you're right, I forgot the roadworks", there was no contest, and "told you" is a victory lap round an empty track. Then say nothing. Most people who hear themselves say "told you" into a silence stop.
The step down
Holding feels free. It isn't.
Every hold is an argument that ran longer than it needed to, followed by a worse hour. The dinner where you're both being careful. The hour after the group chat spent checking. Then there's the second argument about whether you always have to be right, which some of this year's rows grew out of, three years ago. There's the effort of producing defences for something you don't believe while managing your face. And the quietest cost, I think, is the largest. People stop telling you when you're wrong. Spite has nothing to do with it. If correcting you costs twenty minutes and a bad evening, they'll do it less, then stop, and you'll be wrong about more things for longer. The person across from you saw your face when you remembered the roadworks. A junior colleague who watched you hold on the figures will bring fewer, and the ones they bring will come softened with "I might have this wrong, but". And a child who sees you climb down cleanly learns that being wrong gets fixed with a sentence. A child who watches you hold learns it's a catastrophe to be defended, and you'll meet that again at seventeen.
The method has three parts.
First, notice the turn. Most people miss it because it's fast and the next defence covers it. If you made fewer marks this week than you expected, it's because the turn went by unmarked. For most people there's a small physical drop, half a second where the ground under the line goes, and then a reach for a weaker argument than the one you were using. "The roadworks are nearly finished" is weaker than "the second is quicker", and you can feel the step down as you take it. When the argument you're about to make is worse than the one you just made, the turn has happened. Other signs to watch for.
- the subject moving to their tone, "there's no need to be like that", because tone is the only ground left
- your volume rising
- a sudden interest in a detail that doesn't matter, like which month the roadworks started
- "well, anyway"
- the pull to check your phone for something you already know
For the first week or two, don't try to say anything. When you catch it, say to yourself, silently, "that was it", and then do whatever you'd have done. Catching it is a separate skill from acting on it. Turns that arrive in the car home or three days later still count.
Second, the five words. "I was wrong about that." "I was wrong" says the thing. "About that" fences it, so it's about the roundabout and leaves you out of it. There's no "sorry", because being wrong about a roundabout isn't a fault. If there's real harm to apologise for, that's a different sentence and it comes after.
Then say nothing. Let the silence sit for a beat and let the other person fill it. They will. "No worries." "So the third, then." If you fill it yourself, you'll fill it with the salvage or a joke, and the ending gets missed. Leave out "but", "to be fair", "in my defence" and "although". Each is a hinge, and whatever is on the far side undoes the near side. Roy Lewicki and colleagues found in 2016 that acknowledging responsibility was one of the parts of an apology that mattered most, and the five words are that part alone, sized for a roundabout. Practise them when nothing's at stake. "I was wrong about the bins, it's Thursday." At the turn, with your face hot, you'll reach for whatever's nearest, so make it these.
Third, for bigger turns, the ones argued over days or in front of people, use three sentences. What you thought, what changed, what you think now. "I said the new supplier would be cheaper. I've seen the onboarding figures and they take it over the old one. I think we should stay where we are." The middle sentence is what separates changing your mind from changing sides. If you haven't got a reason yet, say "I'm not sure any more" until you have. Say it as soon as you have the reason, and in front of the same people who heard the line, because a quiet word with one of them afterwards is the slide with a witness. If someone brought you the figures, thank them once, plainly, which is the cheapest way to make sure they bring the next thing. Then let the meeting move on. A speech about how much you value being challenged turns the climb-down into a performance.
Here's the decision. When I notice the turn, I say so in the same conversation. If the turn comes late, the next conversation. The default is now, because now costs a minute and later costs an evening.
Peter Gollwitzer and Paschal Sheeran reviewed ninety-four studies of plans written as "when this happens, I'll do that", and found people who plan that way follow through much more often than people who only intend to. So write three.
- When I catch myself reaching for a weaker argument than the one I just made, I'll stop and say "I was wrong about that."
- When someone shows me a fact in a meeting that changes my view, I'll say what I thought, what changed and what I think now, before the meeting moves on.
- When I've climbed down and the other person says "told you", I'll say nothing and let it pass.
This part is my expectation rather than a finding. In the first week you'll notice the turn about half the time and act on it less than that, and the minute of heat will be a full minute. By the second week it'll be shorter, because the beat of silence keeps getting filled by the other person and the ceiling stays where it is. Somewhere in the first month someone will climb down to you, unprompted, about something small.
Four minutes early
A few situations need more than the rule.
- At work, with a boss, the fear is losing credibility when you can least afford it. Your boss has almost certainly seen you were wrong already. What they're waiting to see is what you do about it. With someone junior, the case is stronger still, because they're learning from you what happens when they bring a correction.
- In the group chat, after the climb-down, put the phone down. If someone piles on, there's nothing to add. Silence in a group chat reads as calm.
- With a partner, holding costs most and the words do most, because this is the person who'll be there next Tuesday. A partner who hears "I was wrong about that" cleanly a few times will start saying it back.
- With a child, say it plainly. "I was wrong, the bus does go at ten past." Then the beat, then move on. The sentence was the lesson, so don't add a lecture on admitting mistakes.
- If the same person gloats every time, and it's making you hold to avoid it, say so once, at a calm moment. "When I say I was wrong, I'd rather you didn't do the told-you thing." Most people can meet that.
When you weren't wrong, none of this applies. The turn happens inside you, and being pushed is something else. If someone is insisting, getting louder, and you've checked and the line's still standing, the five words would be a lie. Say "I've heard you and I still think the second," and let them hold if they want to. If there's someone in your life who pushes on everything until you concede, and then keeps a record of what you conceded, the honesty asked for here isn't owed to them in that form. Conceding to end pressure is a different thing from climbing down at the turn, and if that's what's happening, talk to someone outside the argument.
And if small corrections make you furious, you can't let anything go, and it's come alongside feeling low or hopeless most of the day for a fortnight or more, see a GP. The arguing is the part you've noticed. It may not be the part that needs attention first.
You'll hold again. Some Tuesday three weeks in, the turn will come, you'll notice it, and you'll say the roadworks are nearly finished anyway, knowing as you say it. Afterwards there's a particular flatness, worse than the old kind, because now you know what you did.
It doesn't undo three weeks. Phillippa Lally and colleagues, following 96 volunteers forming a new daily habit, found it became automatic over weeks and months, and that missing a day didn't set it back much. A hold is a missed day. I'm stretching their finding to this, but the shape fits. You noticed, and you skipped the second part once.
The climb-down is still available, and late is fine. "About the roundabout on Tuesday. I remembered the roadworks halfway through and kept going anyway. You were right." That does more than the five words would have on the day, because it admits the holding too, and the other person has almost certainly never heard anyone admit that. Then ask which belief did the holding that day. If it was the audience, do the next few climb-downs in front of people. If it was the gloat, practise the silence. Don't replay the evening. There'll be another turn by Thursday, and that's the one you can still do something about.
It's a Tuesday a couple of months on, and you're in the car on the way to your in-laws, and you've said the second exit is quicker. Your partner says the third. You start to explain, and halfway through you remember the roadworks, and the ground goes under the sentence, and you feel the reach for something weaker.
You stop. "Actually, the roadworks. I was wrong about that, the third's quicker."
There's a beat. Your partner says "I only know because of last week," you take the third exit, and you're at the in-laws four minutes early and talking about something else.
The next time you feel the ground go under something you're saying, don't reach for the weaker argument. Say the five words, then say nothing, and see what fills the silence.