She couldn’t work out why it had unsettled her. It hadn’t happened to her, and it hadn’t gone particularly badly for anyone. She brought it to our conversation, still turning it over.

A board meeting, earlier that week. A colleague presenting the case to exit a market they’d held for a decade. The analysis had been assembled in two days – work that would once have taken three weeks – and he’d said so, with some pride. The first question he handled easily.

Then the chair asked the second one. What would have to be true for this to be wrong?

“And he just, reached,” she said. “You could see it. He was holding the deck by its edges, and in that moment it slipped through his hands.”

So we sat with it. Not with him, with the slipping. What had been missed? She started listing, slowly. The drafting and redrafting. Waiting on numbers. Arguing with the data on a Tuesday night. The version you abandon on the fourth day because it can’t survive your own questions.

And then she went quiet, and I watched the realisation land: the position he was defending had never been his.

“That’s why it unsettled me,” she said. “I’ve got two decks like that in my bag right now.”

The word I keep coming back to is thin.

Thin judgement looks like the real thing. A decision gets made, a position is held, and it passes every test except one: it can’t take the weight two questions deep.

It’s tempting to explain this as a problem of time. But it isn’t. You can take three weeks over an analysis and still only skim it – the time itself never did the work. What the three weeks offered was contact. You can’t get through them without handling the numbers, forming a view, watching it fail. The judgement that came out was yours because you had made it.

What’s changed is what the work asks of us. A position can arrive complete, competently reasoned, wanting nothing but agreement. We can still get inside it, look for what it missed, refuse it until it ‘survives’ us. But taking a finished thing apart is not the same as having been in the material while it formed. One tidies the edges. The other is where a view comes from. And when we agree without either, we are adopting a view we believe we authored.

There’s another route to thin that I see, and it has nothing to do with what AI writes. It is the volume. More arrives than we have room to take in, so we read the summary and act on it. The decision is still ours in name, but someone else has already decided what was worth keeping – and what wasn’t.

Judgement used to be a by-product. You couldn’t complete the work without producing it, so nobody had to think about it. That’s not the case anymore. The work can now be done well without any judgement being formed, which makes forming it something separate, and something that has to be chosen.

Mostly it comes down to knowing whether the view is authored or adopted, and there is a way to find out. Ask the chair’s question. What would have to be true for this to be wrong?

An authored view has somewhere to go. It knows its own weak points, because you found them. An adopted view has nothing behind it – it’s thin – and you will feel it slip through your hands.