Bến

What I was trying to figure out: whether you could map the parts of a city that don't want to be found. Sài Gòn is loud and getting louder, and the thing I love about it is disappearing into reviews — every quiet corner eventually gets a star rating and a queue. Bến was a small act of preservation: a map of the slow places, made by the people who use them, designed to resist becoming a list of trending spots.

Context

Bến means a landing, a quay — the place by the water where you stop. I wanted the word to carry the whole intent: somewhere you pause, not somewhere you consume. The enemy was the review economy. Once a place is on the usual maps with a 4.6 and three hundred photos, it stops being a quiet place; it becomes a destination, which is the opposite thing.

So the brief was almost contradictory: build a public map of places that survive because they're a little hidden, without that map becoming the thing that ruins them. Most of the design work was figuring out how to point at something gently enough that pointing didn't break it.

Approach

I started by deciding what Bến would not have. No ratings. No photos of food. No "popular nearby." No way to sort by busiest. Stripping those out wasn't minimalism for its own sake — each of those features is an engine that turns quiet into crowded, and I wanted none of them running.

What's left is a dark map of Sài Gòn scattered with warm pins, each one a place and a short note written by whoever added it. Not "great cà phê sữa đá," but "Hẻm 47, after the rain, before the shops open." The notes are the product. They carry the texture of someone actually being there, which a rating never can.

I gave the map a sense of time. Each place has a best hour, and pins warm slightly as that hour arrives — so the same map feels different at dawn and at dusk, and the city's rhythm is part of the interface. Technically it's MapLibre over PostGIS with time-of-day tags; experientially it's the difference between a directory and a thing that's a little alive.

Decisions

A feeling, not a score. Submitting a place asks what it's for — to read, to disappear, to watch the river — instead of how good it is. This is the decision the whole project rests on. Scores aggregate into rankings; feelings don't. You can't make a leaderboard out of "to disappear."

Soft discovery, on purpose. There's no trending view and no notifications when a place gets popular, because popularity is the failure mode. You find places by wandering the map, the way you'd find them in the city. The friction of having to look is doing protective work.

Vietnamese-first, fully accented. The notes, the place names, the interface — all in proper Vietnamese first, with diacritics treated as non-negotiable. This is a map of a specific city by the people who live in it, and the language had to feel like home, not like a translation.

Light moderation, human-paced. New places sit in a quiet queue I read by hand. It doesn't scale, and that's fine — a map of slow places can afford to grow slowly. The day it needs an algorithm to keep up is the day it's already failed.

Outcome

Bến is a small living map with a few hundred places, most of them in districts I walk. It's used by a quiet circle of people who treat it less like an app and more like a shared notebook — and that's exactly the relationship I hoped for. Nobody's "checking in." People add a place when they find one worth protecting, and read it when they want to be somewhere unbothered.

The truest outcome is that it changed how I move through my own city. I add to it constantly, and the act of writing the note — what is this place for — has made me pay closer attention to where I am. The tool taught its maker to look.

Lessons

I almost added photos a dozen times, because every instinct from years of building consumer apps screamed that a map needs images. I'm glad I didn't. The text-only constraint is the reason the places stay quiet; a wall of beautiful photos would have turned Bến into the exact discovery engine it exists to oppose. The hardest decisions were the features I had the skill to build and the discipline to refuse.

If I did it again I'd think harder about the failure case where Bến works — where a place becomes beloved through the map and then crowded because of it. I don't have a good answer yet. The honest position is that a tool like this carries a responsibility it can't fully control, and the best I can do is design it to point gently and hope that's enough.

Credits

Solo build. Map data from OpenStreetMap. The whole idea owes a debt to every friend who ever took me somewhere with the words "đừng nói cho ai biết."