The first time you tie a bowline knot, your fingers become strangers to themselves. The rope rebels. Your hands cramp into shapes they've never made.
You watch someone else's fingers loop, thread and pull through the familiar pattern, and yours respond with the grace of frozen sausages.
But somewhere between the fifteenth and fiftieth attempt, something shifts.
The rope begins to speak and your fingers slowly learn its language. In that learning you find something modern interfaces have made us forget: struggling with a thing is one of the ways we come to see it.
We live in an age of frictionless design. Our tools learn our habits before we know we have them. Spotify learns our taste faster than we do, Gmail finishes our sentences, and our phones unlock by recognising the specific geography of our faces. Everything bends to us and anticipates us. We've built a world that never makes us foreigners to our own fingers, and that is exactly what we've lost.
Consider the photographer in a darkroom, hands in developer, feeling for the edge of the paper in absolute darkness.
She cannot see her work, so she imagines it through touch and times it through breath. Leave the print in too long and the shadows go to mud; pull it too early and the highlights blow out. She becomes the light meter. Her lungs become a clock.¹
Now think of Instagram's one-tap filters. Valencia, Mayfair, Ludwig. Each one takes ten thousand hours someone spent learning how light actually behaves and turns it into a button.
The manual transmission is another example. Everyone who learned on a stick shift carries the muscle memory of failure: the shudder of a stalling engine, the smell of a burning clutch, the specific humiliation of rolling backward on a hill while someone honks behind you.
They also carry something else, an understanding in the body of how power moves through metal and how momentum builds and breaks. They know the car as a physical system, not a service.
Today's CVT transmissions find the right gear ratio on their own, with no human input, no human error and no human understanding required. The ride is perfectly smooth, and the driver learns nothing from it.
You might say this is nostalgia talking. I don't think it is. There's a reason people who learn by doing tend to understand more than people who learn by watching. When we struggle with a tool, we don't just learn the task. We learn something about whoever made it, and about the material itself.
Medieval apprentices didn't simply pick up techniques from their masters. After seven years of watching hands shape clay, they had absorbed how to move with deliberation, how to read wood grain before cutting, how to wait for iron to reach the right colour. The clumsiness was the curriculum.²
Modern software tutorials teach us to be efficient operators instead. We learn the shortest path, not the terrain. We memorise keyboard shortcuts without ever wondering why a function lives where it does, what metaphors shaped the logic, or which human decided that "save" belonged under "file" and not "edit."
A chef teaching knife skills doesn't start you on a food processor. She hands you a dull knife and an onion.
You cry from the vapours and from the frustration, and from the small cuts that teach you where your fingers end and the blade begins. Slowly you learn the onion's grain, how much its layers resist, the particular sound it makes when you cut it right. The dull knife teaches you pressure, and the tears teach you the angle.
Now meal kits arrive with the vegetables already chopped into perfect portions, and AI recipe generators know your allergies but can't tell you why ginger wakes up carrots or why salt makes chocolate sing. It's all very efficient, and you finish the meal knowing nothing you didn't know before.
We're raising a generation on tools that never push back, and it shows up everywhere.
We give students AI tutors that adapt perfectly to their learning style, forgetting that wrestling with a teaching method that doesn't suit you builds a flexible mind. We auto-tune away bad pitch, forgetting that hearing your own bad pitch is how you train your ear. We GPS our way everywhere, forgetting that getting lost is how you learn to look around.
The word "empathy" comes from the German Einfühlung, "feeling into." It first described how people viewing a work of art project themselves into it, how we physically mirror what we see.
But you can't feel into something that never resists your touch. You can't empathise with a system that shapes itself perfectly around you. A tool that does everything for you leaves you a weaker user of it.
So, to designers, educators and anyone who builds the tools that shape people: put some friction back.
Build in moments where the tool asks the user to meet it halfway. Make your onboarding a little obtuse now and then. Let the interface speak its own logic sometimes instead of guessing yours. Give it a skill curve, the kind gamers know, where getting good means understanding how the system thinks, not just what it spits out.
Duolingo almost gets this right with its deliberately repetitive exercises, but it adjusts to your mistakes too quickly. A better app would sometimes insist on its own rhythm, slow you down when you want to rush, and make patience part of the lesson.
For AI systems, this could mean a copilot that occasionally declines to finish the task and walks you through the why instead.
"I could write this function for you," it might say, "but let's build it together so you understand what each line does."
Yes, it's slower. That's the point.
For physical products, it means resisting the urge to hide every bit of complexity. The tools people love most teach you through use. A Swiss Army knife shows you its logic while you fumble with it. A cast iron pan wears in its own patterns and needs a little ritual of care. It gets better with age and attention, not in spite of them.
The bowline still teaches what no quick-release carabiner can: that security comes from understanding the forces on the rope, not from trusting a mechanism. The rope knows things your fingers have to learn, and you only get fluent in a system by first being clumsy in it.
We've spent decades building tools that speak our language perfectly. Maybe it's time to build tools that teach us theirs, tools that make us students again and put us back in a beginner's mind.
Your fingers remember every knot they've ever learned to tie. What will they remember of tomorrow's tools?
¹ The darkroom timer's tick becomes a heartbeat. Ansel Adams talked about "visualising" the final print before he ever touched the paper, a mental picture built over thousands of failed attempts. Digital photography's instant feedback swaps that deep visualising for trial and error. We shoot a hundred versions instead of imagining one.
² Medieval guilds didn't require seven years of apprenticeship because the techniques took that long to learn. It took that long to become someone, through watching, failing and absorbing. The slowness was deliberate: it filtered for dedication and made sure knowledge passed on with its context, not just its content.
This is the first dispatch from Texture of Tomorrow. I write about what technology is doing to the texture of ordinary life. If this one stayed with you, send it to someone who'd argue with it.