I spend all my time making engineers more alike. This book made me nervous about that.

If you've followed what I write here, you know what I do for a living: I reduce variance. Standards, competency matrices, boundaries, a mentorship that trains young engineers toward a shared idea of "good." Almost all of it is machinery for making my people more consistent, more aligned, more the same. In regulated fintech, that's not a flaw. Consistency is how you stay reliable when money is moving. I'd defend every bit of it.

Then I read Matthew Syed's Rebel Ideas, and it named the shadow of the thing I'm so good at.

His argument isn't the one you expect. It isn't really about demographic diversity as a matter of fairness. It's a performance argument, and a sharper one: cognitive diversity, a genuine range of thinking styles and frames of reference, is what makes a group intelligent. A team of brilliant people who all think alike isn't a strong team. It's a set of overlapping blind spots. Syed's phrase for it stuck with me: homogeneous groups reinforce each other's assumptions through mirroring, and certainty becomes inversely correlated with accuracy. They get more confident and more wrong at the same time, and because everyone agrees, it feels like alignment. It feels like a team on the right track. That comfort is the danger.

The line that unsettled me most: a group of wise individuals can become an unwise board. The problem isn't any one person. It emerges from the sameness of the whole.

And I had to sit with the fact that I run a variance-reduction machine. Every standard I write makes my engineers a little more alike in how they approach a problem. That's the point, and it works. But taken too far, the exact discipline that makes us reliable could quietly make us collectively blind, a room full of capable people all looking at the same side of the elephant, mistaking their agreement for correctness.

So the question the book left me with isn't "should I lower my standards." It's harder than that: how do you build consistency without breeding conformity? How do you keep a team aligned on how to do the work, while protecting the disagreement that stops them from being confidently wrong together?

I don't have a clean answer, but I've come to protect a few things on purpose. I keep the hard debates alive rather than smoothing them over, when engineers argue over an approach, that friction is often cognitive diversity doing its job, and the worst outcome would be a quiet room. I try not to hire only in my own image, or reward only the people who think the way I do; the engineer who reliably disagrees well is worth more than another version of me. And I watch out for the warm, agreeable meeting where everyone nods, because I've learned that's not always the sound of a team being right. Sometimes it's the sound of a team that has stopped being able to see.

Standards keep us safe. Dissent keeps us honest. The work is holding both, and never letting the machine that makes us consistent quietly turn us into clones.