Quick fixes stay forever

Sooner or later, every product reaches the same fork in the road.

One path leads to the thing you know is right. It's steeper: it takes longer, it needs more conversations, and someone will be unhappy along the way. The other path leads to something everyone can agree on. It works. It ships. Nobody argues.

Earlier this year I stood at that fork and took the easy path. The feature did what it needed to do. But it was long where it should have been short, plain where it should have been clear, and untested where it should have been tried. I knew it the whole time, and I let it ship anyway.

This issue is about that path, and the three lessons I took from walking it.

Path one: the easy way down

Water never decides to flood a basement. It just keeps finding the lowest point.

Teams work the same way. Each department follows its own slope. Engineering flows toward what is easiest to build. Business flows toward what it sees as valuable. Design stands in the middle trying to hold its line, and every disagreement costs time and energy.

When the pressure rises, design is usually the part that gives. In many companies, what users actually see is treated as decoration, not as part of the quality. So the details that would make it work well quietly drop away, to keep the ball rolling and to keep working as a team.

No one chooses mediocrity. It's simply where the water ends up when nobody builds a channel.

Lesson one: temporary is permanent

Every town has a temporary fence that has been standing for ten years.

Our quick fixes are the same. We call them trade-offs, which makes them sound deliberate and short-lived, as if we'll come back for the better option later. We rarely come back. Many trade-offs stay forever, and each one that stays lowers the bar a little.

Traders know this pattern well. One small loss looks harmless, but a strategy that loses a little every month still ends up deep in the red. Shipping on time is good. Shipping mediocre pieces on time just means they pile up faster, until the whole product is mediocre and nobody remembers deciding that.

Lesson two: basics first, then the extras

There's a fair case against everything I've just said.

The easy path got the work done, and the team stayed a team. This newsletter usually asks whether chasing high results is worth losing people. This time I chose the people. Isn't that the healthy choice? And isn't "something I'm proud of" just my personal taste?

Think of a hotel. Nobody forgives a dirty room because there was a chocolate on the pillow. The basics come first, and that part isn't taste. But once the basics are right, the small extras are what guests remember and tell their friends about. Many people in tech call those extras gimmicks. In the end, they're what people remember.

In banking there's one more twist. It's like a daily commute on a bad road: you can't choose another route, so you drive it every day and resent every pothole. Users who can't leave still pay the cost of mediocrity. They just pay it quietly.

Lesson three: build the road yourself

For years, designers drew the map and waited for someone else to build the road. Whether the road matched the map depended on time, budgets and other people's priorities.

AI changes that. With tools like Codex or Claude Code, a designer can build exactly what they have in mind, working exactly the way they intend. The only limits left are your thinking, your execution and your craft.

So the next time a compromise gets called a trade-off, don't just describe the better path. Build the first stretch of it. In a bank you won't take it to production yourself, and you shouldn't. But a working version in the room ends the "we can't build that" argument before it starts.

Back at the fork

The fork never shows up just once. It shows up every week, in small decisions that each look reasonable: one more field, one more step, one test we can skip this time.

On any given day, the easy path costs almost nothing. That's what makes it dangerous. By the time you notice the cost, it's spread across the whole product.

If I stood at that fork again, I'd stop making things I'm not proud of. Not every detail matters equally. But the ones I let go never came back.

Mediocrity is never one decision. It's the total of every easy one.