Dave Rupert just wrote a piece called People are not friction and I just had to write a short reaction blog post, because Dave names something I’ve been thinking about for a while now. His main argument: the AI marketing dream of a “frictionless” workflow – where you automate away every task you don’t enjoy, every job you don’t know how to do, and every person who slows you down – is not efficiency. And a dangerous place to end up.

He’s right. But I want to take it a step further, because I think there’s something hiding behind the word “friction” that we need to reclaim: friction isn’t the enemy of good work. Friction is where good work gets its shape.

Think about what actually happens when a designer and an engineer disagree about an interaction pattern. There’s a moment of tension – maybe even frustration. The engineer says it’ll be fragile. The designer says it’s essential for the experience. Neither is wrong, necessarily. But the conversation – if your process allows for it to happen – that back-and-forth where both sides have to articulate why they believe what they believe, is where the design becomes robust and both people gain experience. Not in the Figma file. Not in the pull request. In the friction between two people who care about different things and are forced to find a shared answer.

That conversation can be slow, yes. It can feel cumbersome, inconvenient, and inefficient. It’s the kind of conversation a well-prompted agent will never ask you to have, because it doesn’t count as productive. But removing it doesn’t make the work faster. It makes the work thinner. It creates continvoucly morged work.

The best creative work I’ve been part of always involved exactly this kind of productive resistance. A colleague who pushed back on an assumption I hadn’t questioned. A client who asked an awkward question that revealed a flaw in the concept. A developer who said “that won’t work the way you think it will” and was right. Every one of those moments felt like friction. Every one of them made the result better.

What the “frictionless” vision really sells is the removal of dependency on other people’s experience and judgment. And that’s the part that should make us uneasy (and also a bit angry, frankly). Because judgment is the one thing a large language model genuinely cannot provide. It can pattern-match. It can synthesise. It can produce something that looks like judgment if you squint, derived from the logical structures encoded in language. But as Dave points out, it doesn’t care if it’s right or wrong. It has no stakes. No skin in the game. No lived experience that makes it wince when a decision feels off. No introspection. No goals.

When someone on your team pushes back, that resistance carries weight precisely because they have context you don’t have. They’ve built things that broke. They’ve watched users struggle with the last version. They remember the architectural decision from six months ago that makes your clever idea a nightmare to implement. That friction isn’t making the work worse. That friction is the point.

There’s a concept in craft and material design that I keep coming back to (over and over again – human brains, you know …): working with the nature, the “grain” of a material rather than against it. The grain of collaborative human work is often messy and slow. It requires negotiation, translation, and sometimes genuine disagreement. If we try to sand all of that away in the name of speed, we’re not working with the real material anymore. We’re pretending it’s something it isn’t. Professional work always has friction.

So yes – people are not friction. But friction is essential to human work. Friction is what a sculptor feels when the chisel meets the stone. Friction is what turns a rough surface into something smooth and polished. The frictionless version of creative work isn’t faster creative work. It’s no creative work at all.

~