← notes from the desk

On product

We almost built a social network

For about three weeks in the spring of 2022 we had a product roadmap, a designer, and a very expensive interest in a private feed. Then we went to dinner.

by Kavya Iyer · February 18, 2026 · 5 min

01

Members did not want a fifth feed. They wanted fewer, better rooms.

02

A Wednesday letter and one chat per city outperformed the app we nearly built.

03

When the room already works, software’s job is to get out of the way.

And a few reasons, in hindsight, we're relieved we didn't.

For about three weeks in the spring of 2022 we had a product roadmap, a designer, and a very expensive interest in building a private feed. There were mockups. There was a name we're now grateful nobody remembers. There was, briefly, a plan.

What killed it was a dinner in Hauz Khas. We were around a table with fourteen members, and one of them — a founder who'd sold twice — said something like: I already have four feeds I don't check. Please don't make me a fifth. The rest of the table agreed a little too quickly.

The product that never shipped, and the sketches that killed it.
The product that never shipped, and the sketches that killed it.
The instinct to build a product is nearly always the wrong response to a room that already works.

So we killed it. What we shipped instead was a Wednesday letter, a group chat per city, and a much longer list of dinners. The letter has an open rate that would embarrass most product teams. The chats are quiet, on purpose. The dinners are booked six months out.

Sometimes the network is the product, and the software's job is just to get out of its way.

Filed by the desk. Quibbles to hello@baaaz.club.

We print four of these a year, on paper, and post them.
Ask about the letter.

Members get it in the post. Everyone else reads it here, a week later.