Buttons you can find without looking

My oven has a touch panel. Have you ever tried to use one with flour or raw chicken on your fingers? You tap, nothing happens, you tap harder, you wipe your hand on your shirt, you tap again, and now the timer is set to something you didn't ask for. It's an oven. It used to have knobs. Knobs worked. Why is there a touchscreen?

That's the small version of a problem I think will shape the next stretch of technology. We spent years putting every control on a sheet of glass. Now we're finding out which ones should never have gone there.

The car is the worst place to get this wrong

I've nearly had accidents in my own car because I was looking at the screen to change something simple. I drive a Ford C-Max and a Nissan, and both are on the better end. The C-Max is the better of the two. I still catch myself looking down.

Now picture the cars that put the climate, the heat and the windows on the screen. A bee flies in. Smoke from a truck ahead starts pouring through the vents. You need the air off and the window down right now, and the control is two menus deep on a screen that changes depending on what you touched last.

This isn't just my opinion:

In 2022, the Swedish magazine Vi Bilägare timed drivers doing six simple tasks at 110 km/h: seat heat, temperature, defrost, radio, trip reset and dimming the dash. A 2005 Volvo with buttons took about 10 seconds. The touchscreen cars averaged 24.7 seconds. The slowest took 45, which at that speed is about 1.4 kilometers of road driven while looking at a screen. (Vi Bilägare)
Euro NCAP's 2026 safety tests give cars better scores when the indicators, hazards, wipers, horn and emergency call are physical controls. (Automotive World)
Volkswagen's design chief, on putting the buttons back: "We will never, ever make this mistake anymore." (AutoGuide)
After the fatal 2017 collision of the USS John S. McCain, investigators found the destroyer's touchscreen throttles took away the feel sailors relied on. It wasn't the main cause, but it contributed. The Navy announced it would put physical throttles back on its destroyers, because the sailors preferred them. (USNI News)

When a navy, a safety agency and a carmaker all say the same thing, the buttons were right.

What diving taught me about controls

I'm a rescue diver, and underwater you often can't see much. So I had a simple rule for my gear: if I couldn't work it with my eyes closed, it wasn't useful to me.

The second test was gloves. In 3 mm gloves your fingers are clumsy. If I couldn't pick up a dime or a quarter wearing them, I couldn't trust myself to work small controls either. So the gear that earned a place had big, distinct buttons, and every piece lived in the same spot every dive. I set it up the same way each time, with a procedure and a checklist, so my hands knew where things were before I needed them.

That's the whole idea. Controls you can feel, in places that don't move, that you've learned with your hands instead of your eyes.

Don Norman explained this decades ago

If you design anything, read The Design of Everyday Things. The idea that stuck with me most is the simplest one: a door with a handle. Where's the handle? If you can't tell whether to push or pull, the door failed, not you.

Norman gives names to what a good physical control does:

Affordance: what the object lets you do. A knob can be turned. A switch can be flipped. Flat glass only affords one thing, a touch, so every control on it feels exactly the same to your hand.
Signifiers: the cues that tell you where and how. On glass, the cues are visual only, so you have to look.
Mapping: the layout of the controls matches the thing they control, like stove knobs lined up with their burners. A touchscreen reshuffles its layout every time the menu changes.
Feedback: the click, the notch, the resistance. You know it worked without checking. Glass gives you a beep at best.
Constraints: a switch that only goes two ways can't be set wrong.
Knowledge in the world, not in your head: a button stays where it is, so your hand learns it. A menu makes you remember where things went.

So here's my problem with touchscreens in one sentence: they take the knowledge out of the world, put it in your head, and then make you look away from the road to find it.

One more thing a physical control has: no lag. A knob doesn't boot up, freeze, or wait for an update. It's connected to the thing it controls.

It's not just screens. It's things with nothing to touch at all

I've taught people to use virtual and mixed reality. With HoloLens, most people couldn't figure out the interface, because there was nothing to touch. You pinched at the air and hoped. Sometimes we bound the menu to your wrist just to give people one place they could reliably find. It was still confusing. There's a reason Oculus shipped controllers you hold in your hands. People need something real to grab.

It's the same at work. At Mattress Firm, sales associates carry their tools on tablets. You need a free hand. And people don't like it when you walk up holding something to take notes right as you meet them. It's awkward and puts them off. The best design for that moment is the one that gets out of the way, so the person can look the customer in the eye.

Stop using your phone as a remote for the room

Here's the version of this that bugs me most. More and more things that are right in front of you only work through an app. The light that needs your phone. The appliance with no buttons. The thermostat a guest can't change.

That's backwards. The thing is already in the room. Making me pull out my phone, unlock it, find the app and wait for it to connect is longer and worse than a switch on the wall. My phone shouldn't be a stand-in for an interface the thing should have had.

What I picture is the opposite. Screens where they're needed, sometimes in your pocket. But the technology that lives in a place, like your car, your kitchen or your office, should work right there, with your hands, without your phone.

Where screens do belong

I'm not against screens. Screens are great at showing you things. Maps, photos, messages, settings you change once a year while parked.

My rule is simple. If I can do it with my eyes closed, it gets a physical control. If I need to see something to do it, a screen can help. Touch controls only make sense when I need to touch the thing and see it at the same time, like moving a map.

And a screen that just shows what's going on, like the temperature or the battery, is a great screen. Let it report. Keep the controls physical.

A checklist before you replace a button with a screen

Can someone do it with their eyes closed? Then give it a physical control.
Is it urgent or about safety? Physical, and always in the same place.
Do they need to see something to do it? Then a screen can help.
Do they need to touch it and see it at the same time? That's the one case for touch controls.
Could a click, a sound or resistance tell them it worked? Then don't make them look.
Is the screen only reporting status? Good. Let it show, and keep the controls off it.
Will it be used with wet, dirty, gloved or busy hands? Test it that way.
Are you sending people to their phone to control something that's right in front of them? Don't.

Design has to leave the screen

Most of us who do UX and UI work only design for screens now. That's where our tools live, and it shows. Everything starts to look like an app, even the oven.

We need to get back out into the world. Sit in the car. Stand at the stove with messy hands. Watch someone at work with their hands full. Some of the best interface decisions aren't on a screen at all. They're a knob in the right place, with a click you can feel.