Stop winning debates at work

Roll a die, but call the result out loud first: "four!" The die lands on five. Now listen to what your brain serves you. "The table is crooked." "I actually meant five." "Off by one, that barely counts." "Someone bumped my arm." Anything, except the one honest sentence: "I was wrong."

That reflex is funny at a game table. In a meeting room, it quietly ruins the way we make decisions together. This article is about that failure. It starts in our heads, so that's where we'll start too.

Being right feels wonderful

Being right is a pleasure. We savor it so much that we even enjoy its bitter version, the "I told you so" that follows a disaster we predicted and would have preferred to avoid.

Being wrong should just be the same coin, flipped. Not quite. An error that stays private costs almost nothing. We notice, we adjust, we move on. What we dread is the moment it becomes visible to others, when the error stops being information and turns into a verdict about us. Keep that fear in mind, because it's the one doing the damage in every debate of this article.

Kathryn Schulz points out a trap underneath: being wrong feels exactly like being right, until the moment you find out.

Put the fear next to the trap, and being wrong stops being one thing. It splits into three situations, worth telling apart.

Sometimes we know, and pride keeps the confession down. Sometimes we sincerely believe we're right: there's a mistake somewhere in our reasoning, and from the inside, nothing feels wrong. And sometimes we couldn't have known: the information changed under our feet. Only the first one is a character flaw. The other two are the normal condition of anyone who thinks.

The machine that must be right

Evolution shaped our brains to chase what feels good, and being right feels good. So we are, in a very real sense, machines built to be right. The catch is that the machine takes the cheapest path to the feeling, and that path is almost never predicting correctly. It's twisting reality after the fact.

The twisting has names. Cognitive dissonance reduction makes us rewrite the story rather than sit with the discomfort. Confirmation bias picks which facts we notice. Motivated reasoning builds the justification we wanted, though never past what stays plausible: we can't tell ourselves just anything. And once we've defended a position publicly, we double down on it. Education adds its own layer on top of the wiring. School taught us early that being wrong is what the bad students do.

Reading a list like that one leaves a comfortable feeling of being cured. It's an illusion, and I'm proof. These biases keep catching me, and I know them by name. What naming them actually buys is smaller, and still precious: you can't watch for something you can't name. The rest of this article is about where to aim that watchfulness.

The three debates

Where do we chase being right the hardest? In debates. And debates come in three flavors.

Convince the other

The everyday kind. It's noon, you want a salad, your colleague wants a burger, and lunch can only go one way. So you plead your case, your colleague pleads back, and whoever runs out of arguments first eats what the other one picked. Small stakes, real duel.

Convince the crowd

The political kind. Two candidates share a stage, and no candidate in history has ever answered "you know what, good point". That's because the opponent was never the target. Every line is spoken over the adversary's head, at the audience holding the votes. It looks like a dialogue and works like two monologues aimed at the same room.

Convince no one

The internet kind. Two strangers trade insults under a post, each defending a flag nobody else can see. Neither will convince the other, no audience worth winning is watching, and both know it. The exchange goes on anyway, sometimes for days, because it stopped being about convincing long ago. It's about not leaving the field.

Three different targets, one common shape. These are debates you try to win.

Where playing dirty is almost rational

The sophists understood it in antiquity. An argument doesn't need to be true to win, it needs to convince. Arthur Schopenhauer even wrote the catalog, thirty-eight stratagems for prevailing while wrong.

There's a colder reason the dirt sticks. In a debate of opposition, being the only fair player is a losing strategy. Admit that your salad's dressing is mostly oil, while the other side concedes nothing, and you've just booked yourself a burger for lunch. So you keep the inconvenient facts in your pocket. Lying by omission, but lying. In a debate you must win, honesty is a handicap you choose.

The fourth debate

There is a fourth kind of debate, and I left it off the list on purpose. It's the one this article is about. The other three are debates of opposition. This one is a debate of cooperation, the kind a team has when it argues about how to solve a problem. Make no mistake, it's still a fight. What changes is the opponent.

In this debate, ideas stop being flags to defend. They are candidate solutions, and the whole point is to stress-test them until the best one is left standing, even if the best one isn't yours. A serious theory of reasoning points the same way. Hugo Mercier and Dan Sperber argue that human reasoning evolved for argumentation, biased alone and remarkably effective in groups, precisely when people genuinely exchange arguments instead of performing.

The short circuit

The trouble is that we walk into the fourth debate carrying the reflexes of the other three. Omissions, sophisms, the whole toolkit, deployed against our own team. Nobody would stand up and defend "let's pick a worse solution", and yet that's exactly what winning against your own side means.

Part of it is habit. But part of it has a mechanism: the audience. A workplace debate is rarely a one-on-one. Colleagues watch, and above all the boss watches, and the moment pleasing the audience matters more than solving the problem, you've silently switched to debate number two. You're not solving anymore. You're performing.

Hierarchy makes it worse. When the highest-paid person speaks first, every opinion after it is a career decision, which is why the classic counter-measure is exactly that person speaking last. And there is a skill worth training on its own: noticing the switch on yourself. The plea building, the inconvenient fact sliding into your pocket. Catch it early enough, and posture can be corrected mid-debate. The alternative is the quiet failure, a team that stops debating altogether and nods itself into groupthink.

One assumption runs through everything that follows. People in the room are allowed to disagree. Where speaking honestly gets punished, the first problem to solve isn't debate technique.

Structure the debate

How do you keep the fourth debate cooperative? First, aim everything at the problem. Roger Fisher and William Ury said it plainly: separate the people from the problem, and judge options on objective criteria. "I just prefer this one" is a taste. "This one answers the problem without the side effects" is a criterion everyone can rally to. Definitions come first for the same reason. Half of the fights I see are two people using one word for two things.

When agreement won't come, dig instead of pushing. The real disagreement usually lives upstream of the solutions, in what each person takes as given. Walking back to those axioms (the double crux technique gives it a protocol) either dissolves the argument or reveals the one fact you actually need to check.

Some disagreements aren't about reasoning at all. One side knows something the other doesn't: a constraint, a promise made to a customer, a failure nobody else lived through. The asymmetry wears the mask of a difference in judgment, and it dissolves with the simplest question of the whole method. What do you know that I don't? And when the missing piece can't be shared, saying that it exists is already progress. A decision that rests on private information is easier to accept when it stops pretending not to.

Two more rules keep the debate alive and honest.

First rule: the debate needs strong opinions.

"I'd go with A." "I'd go with B." "Sure, B then." Nothing happened there. No advantages compared, no reasons given, and for all we know, whoever proposed A simply didn't feel like arguing. The team walked into B unexamined. That's the false status quo, and it's worth being suspicious of any decision that arrives without resistance.

So hold positions firmly, even volunteer to defend the option nobody loves, or the room will never really look at it. Volunteer is the operative word. A standing devil's advocate, assigned to object by routine, produces objections nobody believes, not even their author.

One tempering note. Strong opinions are for questions that are actually open. Some answers are obvious, and staging a debate around them is theater. And some decisions arrive already made, by the company, by a regulation, by a contract. Those breed the worst kind of false debate, where a team spends its energy, sometimes a full slide deck, proving wrong a decision nobody in the room can reopen. When the constraint won't move, the real debate is how to work within it.

Second rule: the debate needs an ending, an explicit agreement on who decides and when. Without one, the debate replays every meeting until the most stubborn person wins by exhaustion, which is opposition with better manners. With one, losing gracefully has a name: disagree and commit.

It helps to write the decision down, even as one line. The next time the topic drifts back, "we settled this" plus a pointer closes it in a sentence. The lock has one legitimate key, though. A decision closes the debate as it stood, with the facts it had. Bring a genuinely new fact that touches the subject, and reopening isn't stubbornness. It's the method working.

Criticize kindly

Everything above assumes something uncomfortable: that flawed arguments get called out. A sophism in a cooperative debate works against the whole group, so letting it slide sabotages the solution. But pointing at it lands, almost always, as a personal attack. The skill of this whole game is doing one without the other.

Two habits carry most of the weight. Judge the argument, never the person, and say so explicitly, choosing the words with care, because some words offend even when aimed at the argument. And ask questions instead of asserting. A good question walks people deeper into their own reasoning, where they either hit the real disagreement or discover the weakness themselves, which is infinitely easier to accept than having it shown to them. It's the oldest trick in philosophy, and it works twice over: asking casts you as someone requesting their knowledge, and the reflex to help someone who asks runs deep. In a cooperative debate there's a third layer: the problem sits in the middle of the table, so walking away from the question isn't really an option.

Rapoport's rules formalize the same spirit: restate the other position so well they thank you for it, list what you agree with and what you learned, and only then criticize.

One last piece, and it's the hinge of the whole method. Strong opinions only work if errors are cheap. To hold a position firmly, you must accept that it can be wrong and fold it when it's shown to be. That is genuinely hard, and it trains. Two things help. First, a liberating fact: we are usually the only ones making a drama of our own mistakes. Everyone else barely notices, or is quietly glad we caught it. Second, someone has to go first. Point at one of your own past mistakes, laugh at it, and the room learns, better than from any speech, that this is a place where being wrong costs little. Teams that industrialize the idea call it a blameless postmortem.

Debate to solve

Next time a discussion heats up at work, check the target. If your arguments are aimed at a person, you're winning something nobody needed won. Aim them at the problem: define it, criticize ideas and not their authors, ask more than you assert, defend your position hard, and drop it gladly the moment it breaks. The team that debates this way doesn't just decide better. It's a team where thinking out loud is safe, and that's where the best solutions were hiding all along.