Navigating the lobby of a high-end hotel in Seoul after of travel feels like trying to run underwater. My brain is a slurry of jet lag and chemical compounds, mostly because my day job as a sunscreen formulator, David T.-M., requires me to think in terms of molecular weights and photostability.
I am currently obsessed with the way 37 different types of zinc oxide can either sit gracefully on the skin or make a person look like they are wearing a death mask of Victorian lead paint. But standing at this marble counter, the chemistry of the skin is the last thing on my mind. I am just trying to check in.
The chemical and logistical weights acting on a formulator’s brain mid-transit.
I have been in Korea for . I know precisely three words in the local tongue, and two of them are “thank you,” which I say with such desperate, wide-eyed intensity that the locals usually take a half-step back. Behind me, a Japanese couple is discussing their dinner plans.
I know this because I spent of my youth in Osaka, and my brain, currently short-circuited by the taxi ride from the airport, decides to bypass English entirely. I turn around and offer a quick recommendation for a nearby ramen spot in rusty, stuttering Japanese.
The Conversational Car Crash
The clerk at the desk freezes. She has been speaking English to me for the last five minutes. Now, the air in the lobby thickens. The app on her tablet, which she had dutifully set to “English to Korean,” is now useless. It stares at us, a digital witness to a conversational car crash.
I realize I have to fix this. I reach for my phone, but before I can speak, I am confronted by the Great Modern Insult: the dropdown menu. I have to tell the machine what I am about to do before I am allowed to do it. It is the conversational equivalent of having to fill out a 107-page permit application just to say “hello.”
I have to scroll past “Icelandic,” “Igbo,” and “Italian” just to find “Japanese,” then toggle the input, then wait for the spinning wheel of death to acknowledge my choice. By the time I am ready to speak, the Japanese couple has politely nodded and moved toward the elevators, and the clerk is looking at me as if I might be having a neurological event.
This is the strange dishonesty of modern interface design. We are told that technology is making us more connected, but in reality, it is forcing us to become administrators of our own thoughts. Every time a software package asks you to “Select Source Language,” it is admitting a fundamental failure. It is putting the burden of the database on the user. It’s like a waiter asking you to go into the kitchen and label every vegetable before they’ll take your order.
The Hazard of Rigid Tutorials
I recently tried to build a “floating” bookshelf I saw on Pinterest. It was supposed to be a simple Saturday project-reclaimed cedar, some heavy-duty brackets, a bit of wood glue. I spent $177 on materials and of my life that I will never get back.
The tutorial was full of these rigid, unforgiving steps that assumed my wall was perfectly flat and my wood was perfectly straight. It wasn’t. The wall has a 7-degree tilt that I didn’t know about, and the cedar was warped from sitting in a humid warehouse. Instead of a floating shelf, I ended up with a diagonal hazard that looks like it belongs in a haunted house.
I followed the instructions to the letter, and that was my first mistake. I tried to force a natural, messy material into a predetermined box. That is exactly what we are doing when we force people to click a button before they speak. Language isn’t a setting; it’s a fluid. It leaks, it pours, it evaporates.
The Formulation of Connection
As a sunscreen formulator, I deal with emulsions-mixing oil and water. You can’t just tell the oil to stay put. You have to use surfactants; you have to understand the tension between the two phases. If your formulation is too rigid, it breaks. If it’s too loose, it runs into the user’s eyes and stings like a betrayal.
Software should be a surfactant. It should lower the surface tension between two humans, not add a layer of plastic buttons between them. When I’m in the lab, I don’t tell the zinc oxide to “be Japanese” or “be English.” I just observe how it reacts to the medium. Yet, we accept this friction in our digital lives as if it’s an inevitable tax on progress. We have been trained to think that “setting up” the conversation is a necessary part of the conversation itself. It isn’t. It’s just bad chemistry.
I remember a specific batch-batch number 77-where I was trying to create a sheer mineral tint. I was so focused on the label-on making sure it met the “Broad Spectrum” criteria-that I forgot to actually look at the dispersion. I was so busy checking the boxes of the regulatory paperwork that I didn’t notice the formula was pilling on the skin like old sweater lint.
I was doing the work of a clerk, not a scientist. We do this every time we open a translation app and start scrolling for a flag icon. We are acting as data entry specialists for a machine that should already know who is standing in front of it.
Language as a Moving Target
The problem is that most developers treat language as a static property, like a file format. But if I’m in a hotel in Seoul, and I’m speaking to a clerk who knows I’m American, but I’m looking at a couple from Tokyo, the “source language” is a moving target. It is a live, breathing context.
The software shouldn’t be asking me what language I’m speaking; it should be listening to the frequencies, the phonemes, the rhythm of my breath, and figuring it out for itself. We have spent decades teaching humans how to talk to machines, only to realize the machines were supposed to be the ones listening.
The machines were supposed to be the ones listening.
This is where the industry is finally starting to pivot, moving away from the “administrative pre-step” and toward actual intelligence. The goal is a world where the interface disappears entirely. I want to be able to mumble a half-remembered phrase in a foreign tongue and have the machine catch me before I fall, without me having to tell it which net to use.
I’m looking for tools like
that understand that the most important part of a translation isn’t the dictionary-it’s the detection. It’s the ability to realize that the person in front of you has just switched gears, and the technology needs to shift with them, seamlessly, without a manual override.
Invisible Protection
I think back to that Pinterest shelf. If I had just trusted my eyes and adjusted the brackets to the actual slant of the wall-if I had worked with the reality of the room instead of the “ideal” version in the tutorial-it would be level today. Instead, I have 7 holes in my drywall and a pile of cedar that I’m probably going to turn into birdhouses. I was so committed to the “process” that I ignored the product.
In the world of SPF, we have 47 different regulations globally for how to test for UVA protection. It’s a nightmare of bureaucracy. But the consumer doesn’t care about the ISO standards or the PPD ratings. They just want to go to the beach and not get burned. They want the protection to be invisible. They want it to be a “set it and forget it” part of their morning.
Current technology is 97% packaging. We are fighting for the 3% of active connection.
Conversation should be the same. The “labeling” of our speech is a relic of an era where computers were weak and needed us to hold their hands. But we are past that now. Or at least, we should be. There is a certain kind of exhaustion that comes from having to represent yourself to a machine before you can represent yourself to a human.
It’s a micro-fatigue. It’s the 77th time today you’ve had to click “I am not a robot” or “Accept all cookies” or “English (US).” By the time I finally got checked into my room in Seoul, I was too tired to even look at the view. I sat on the edge of the bed and thought about how much energy we waste on the logistics of being alive.
The DIY Project for the Next Decade
If I could reclaim all the time I’ve spent in dropdown menus, I could probably finally finish that 777-page chemistry textbook I’ve been lugging around for three years. We are at a tipping point where the “administrative paperwork” of technology is becoming more work than the task itself.
I want the active ingredient. I want the connection. I want to speak, and I want to be heard, and I don’t want to have to tell a silicon chip which box to put my soul in before it transmits the signal. Maybe that’s the real DIY project for the next decade: stripping away the UI until there’s nothing left but the exchange.
No flags, no dropdowns, no “Select Source.” Just the messy, beautiful, unlabelled reality of two people trying to understand each other in a world that is far too complicated for a single menu. I might have failed at the shelf, but I’m not giving up on the chemistry. We just need to stop asking the wood to be straight and start learning how to build on a slant. It’s the only way to stay level in a world that refuses to sit still.
Is it too much to ask that our tools be as intuitive as our mistakes? I don’t think so. I think we are just waiting for the machines to finally grow up and do their own chores. Until then, I’ll be in the lab, trying to find a way to make zinc oxide disappear on the skin, and hoping for a world where my words can do the same.
“A promise is a tension. When a brand says limited 16 times, the thread loses its memory.”
– Sofia, Thread Tension Calibrator