The cringiest thing you can say in 2026 is that you have taste.
Everyone is saying it.
“Taste is the new moat.” “Taste is my edge.” “Taste is the one thing AI can’t take.” The taste tweet has become its own genre—posted by founders, VCs, and designers, retweeted with the kind of conviction usually reserved for political endorsements.
They’re not wrong—taste matters. In a world where any model can produce competent output on demand, the person making the call about which output is actually good—the person with judgment—is suddenly the bottleneck. The thing in shortest supply.
They’re just wrong about how to prove they have it.
Real taste is one of the few things that doesn’t survive being claimed out loud. The moment you tweet “I have taste,” you’ve made the least taste-revealing move available to you. You’ve gone to the menu and pointed at the picture of dessert.
Your moat is not tasteful.
So if announcing it doesn’t work, what does?
Quick definitions, because the word is getting muddy: taste isn’t preference (everyone has those) and it isn’t expertise (you can know everything about wine and still order terribly). It’s the consistent pattern in your judgment calls—what you keep, what you cut, what you’d defend at 2am with no audience.
Paul Graham once described his cofounder as having x-ray vision for character—able to read a person on sight. Taste is that, pointed at work instead of people. It’s also why he insists taste isn’t just personal preference: keep making things and your old taste doesn’t look different, it looks worse. You can’t fake the years.
So taste is mostly invisible to the person who has it—and nearly impossible to verify in someone you just met.
So how do we ever get a real look at it?
It’s 1975. Brian Eno has left Roxy Music, made a run of solo records, and just taken his first outside production credit. Eventually he’ll shape Bowie, Talking Heads, U2, Coldplay—the kind of career where his ear is what every studio in London is paying for.
And he’s stuck.
Sessions seize up. The clock is running, the room is waiting on a decision, and Eno can feel his own judgment getting squishy under the pressure. The thing he’s there for—in the moments it matters most—just evaporates.
So he and his friend Peter Schmidt, a painter, do something strange. They write little instructions on cards.
They call them Oblique Strategies. When the room locks up, Eno pulls a card. The card forces a constraint. The constraint forces a choice. The choice reveals what he actually thinks.
David Bowie used the deck across the Berlin trilogy. Recording “Heroes,” he and Eno each drew a card and agreed not to compare them until the track was finished. Eno’s said to make everything as similar as possible. Bowie’s said to emphasize differences. “Sense of Doubt” is what came out.
One of the most famously taste-driven people of the 20th century didn’t trust his taste in the abstract. He only trusted it in motion—under pressure, against a constraint, when he had to commit.
Taste isn’t something you have. It’s something that happens to you when the situation forces it out.
It’s not what you say you’d pick from infinite options. It’s what you actually pick when you only get one shot, the budget is tight, the deadline is real, and someone is going to remember the call. It’s a verdict, not a menu.
Which reframes the original question entirely. We aren’t really asking, do you have taste? We’re asking, can we set up the conditions to find out?
And there’s a deeper problem the “taste is my moat” crowd hasn’t gotten to yet.
A lot of the current attempts to quantify things about people—personality tests, behavioral analytics, the entire HR-tech assessment industry—fail at the most important step. They reduce a person to a four-letter code or a percentile bucket, and hand it back like that’s an answer.
They don’t measure, they compress—they squeeze the person right out of the personality.
If you actually want to measure taste, you have to do two things at once—capture something rigorous enough to be trusted, and give it back in a form that feels like recognition, not a verdict from a stranger.
If someone reads their result and doesn’t think “yeah, that’s me” for every sentence, the measurement failed. Doesn’t matter how many data points it ran on. That’s the bar. Honor the person, or don’t bother.
So back to the tweets.
They’re funny, but they’re also a little sad. They’re trying to claim something that, by definition, has to be shown.
Eno didn’t write essays about his taste. He built situations that pulled the verdict out of him. That’s the move. Not louder declarations—better moments. Small, structured spots where the call is yours, and the result hands you something you actually recognize.
You can’t tweet your way to taste. You can only choose your way there, one constraint at a time.