When an argument stops being about the work

What it says

Some opinions are doing a job that has nothing to do with being true. They say something about you — that you're careful, or realistic, or the one who thinks about the customer. Push better evidence at an opinion like that and it doesn't move. It gets defended better. Kahan and his colleagues found that the people best at thinking things through were the most dug in of all, because being good at an argument is also being good at defending your own side. That includes you. Especially if you're the one who turned up with the evidence.

Where it comes from

Dan Kahan and colleagues, working out why people argue about climate science, guns and vaccines. The identity being protected there is a side you belong to. An argument at work is usually smaller and more personal than that — not your side, just how you come across. So we're borrowing this. Nobody has tested it on a Tuesday afternoon row about a deadline, and we'd rather say that than dress it up.

What to do with it

It hands you a question rather than a trick. Ask what it would cost the other person to drop their position. If the answer is "being the one who's careful about this", then the argument was never about evidence and more of it won't end it. Now ask it the other way round, which is the harder half. What would it cost you?

When it's the wrong lens

When nobody's dug in yet. If the two of you simply haven't spoken, this turns a diary clash into a theory about human nature.

What it says

What keeps a working relationship working isn't the big conversation. It's small and constant: saying out loud that you want to keep working together, doing the bit you said you'd do, and now and then talking about how the two of you work rather than only about the work. Stafford and Canary found five of these habits. One of them is sharing the tasks — carrying your own share. So the research's name for keeping a relationship healthy includes doing your job.

Where it comes from

Research on how people talk to each other, and most of it is about couples. Say that out loud and it sounds like we're recommending date night with a colleague, so here's the honest version: the research on workmates exists and there's less of it. Waldron, in the early nineties, found people doing the same kind of small upkeep on the relationship with their manager. Not the same list — a shorter one, and about managing upwards. The shape holds; the detail does not transfer cleanly.

What to do with it

An argument costs you roughly what you didn't bank before it. Two people who only talk when something is wrong have nothing to spend when something goes wrong. So the useful move isn't in the argument at all — it's in the weeks when nothing is wrong. One conversation that isn't about a task, and finishing the small thing you said you'd finish. Dull. Also the whole of it.

When it's the wrong lens

In the middle of the argument. This one is for the month after. Reach for it on the day and you're the person offering a coffee to somebody who wants a decision.

What it says

Three different problems, kept apart because they behave differently. Overload is more real work than there are hours. Conflict is two people wanting opposite things from you, both of them being reasonable. Ambiguity is not knowing which of it is actually yours. Overload wears you out. Ambiguity makes you defensive, because when nobody has said where your job stops, every request is also a quiet question about whether you're any good at it.

Where it comes from

Kahn and colleagues set this out in 1964. Rizzo, House and Lirtzman built the survey in 1970 that most of the research since has run on. It's the most measured of the three cards here and the least about personality, which is an odd thing for us to be recommending. That's exactly why it's here.

What to do with it

All three come out as being annoyed with whoever asked most recently, which is why this gets mistaken for a problem between two people. So check the queue before you check the relationship. Can several people add to your work without seeing each other's requests? If they can, the queue is the problem, and the most careful conversation in the world won't fix it. Write the list down and take it to whoever can see all of it.

When it's the wrong lens

When the queue is clean and it still grates. Then it's about the two of you, and one of the other two cards is the better place to start.

The move

"You're right about the number — that's settled. The thing I want to sort out is how we found out about it on Thursday." Two sentences, in that order, out loud.

Why it works

What people actually dread in a correction is the fact and the judgement arriving as one package. Split them and somebody can agree about the number without agreeing anything about themselves.

How it fails

Doing it without meaning it — handing over the fact as a move, on the way to the real accusation. People hear that straight away, and it's worse than not bothering.

The move

Before you start, finish this: "What I want out of this is ____." An apology, a changed process, a decision, or just to be told you were heard. Any of those is a real answer.

Why it works

Said out loud at the start, it gives the other person something to agree to. "I'm not asking you to redo it, I'm asking that the next one comes to me first" is answerable in a way a general complaint isn't.

If you can't finish it

That's the finding, not a failure to prepare. A conversation with no ask turns into a conversation about who was more wronged, because that's the only question left in it.

The move

End by saying which parts are settled, which are deliberately not, and where the parked ones go. A line in a shared doc is enough. What matters is that both of you can see the same list.

Why it works

It stops the next conversation being about two things at once. An unparked item comes back in the next argument as evidence, and that version is harder than the original.

How it fails

Getting skipped, because by the time you reach it the hard part feels over. It's also the step that decides whether you have this conversation once or four times.