An indie music app, by design.

Why I built a music app with no algorithm, no feed, and no follower counts — and what "small on purpose" actually means in software.

If you've been on the internet long enough, you've watched a specific story play out: someone builds a small, lovable thing for a specific group of people. The thing works. It grows. The growth attracts investors. Investors want returns. Returns require scale. Scale requires the algorithm. The algorithm requires the feed. The feed requires the engagement loop. Five years later the lovable thing has become indistinguishable from the rest of the internet — addictive, anxious, a little bit hostile.

This story is so consistent that we've stopped questioning it. The default architecture of a social product in 2026 is: a feed, a follow graph, a recommendation engine, a notification budget calibrated to maximize daily-active-users. Every meaningful design decision flows from those four primitives. We've built almost no apps that don't share that genome.

Vibes is an attempt to build one. It does the smallest possible thing music sharing can be — pick a song, pick a friend, send — and then refuses to grow into the rest of the playbook. No algorithm. No feed. No follower counts. No "people you may know." No "explore tab." No quote-share. No public counter on anything.

That sounds like a constraint. It's not. Refusing the feed is the entire product.

What got broken when sharing went social.

Here's what I think happened. Up until maybe 2008, sharing a song with a friend was a one-to-one act. You burned a CD. You handed someone a mix tape. You made a Limewire playlist on a shared computer. The audience was specific, the act was deliberate, and the medium implied effort.

Then social products figured out how to insert themselves between you and your friends. The mechanic was clever: give everyone a feed, let everyone post to it, and let the platform decide which of those posts surface to which other people. The result was a one-to-many channel pretending to be a one-to-one one. You'd "share" a song to a public profile, and somewhere out there an algorithm would decide whether your real friends or four hundred strangers got to see it first.

Refusing the feed is the entire product.

You can see the consequences in everyone's listening habits. Spotify Wrapped reveals what you played alone, in private, with nobody in the loop. The "currently listening" surfaces in Apple Music are buried because nobody really uses them. The closest most people get to friend-curated music is a Spotify Blend playlist that the algorithm crossfaded for them.

None of these are bad products. They're just products that solved different problems than the one I cared about. The problem I cared about was the act of sending a song to one specific friend, on purpose, and having it feel like an act.

What "small on purpose" looks like in code.

It mostly looks like saying no. The list of features I considered and didn't ship is significantly longer than the list of features in the actual app. A few examples:

• An ambient listening feed Cards floating past showing what your friends were just playing. Beautiful in mockup. Would have been the most-used surface in the app. Would have killed the entire premise. Why I almost built it.
• Public profiles with follower counts The path of least resistance. Every social product has them; people would have understood instantly. The reason I didn't is that follower counts turn friendship into a metric, and the moment you can compare numbers with strangers you've stopped having friends.
• An "explore" tab Probably the single biggest growth lever a social music app could have. I cut it because the algorithm-curated browse surface is structurally hostile to "I'd rather hear what my five real friends are sending me." If people want broad discovery, Spotify and Apple Music exist and are extraordinary at it.
• Notifications outside the day The app sends a single push when a friend sends you a vibe. It does not push at 11pm. It does not push at 7am unless you opt in. It does not batch up "your friends are listening" digests. The default state of the app is silent.
• Streaks "You and Maya are on a 7-day streak — don't lose it!" An entire generation of products has trained us to feel anxiety about not opening an app. We will not be among them. Streaks live only inside the weekly recap, end-of-week, gentle.

None of these refusals are technically hard. The hard part is staying refused. Every one of them got proposed by a beta tester who genuinely meant well. Every one of them sounded reasonable. Every one of them would have, on its own, made the product slightly larger and slightly more like the rest of the internet.

The economics of staying small.

The other reason every social product ends up at "feed plus algorithm" is that the financial model demands it. Advertising-funded products need engaged eyeballs in a defined surface — that's literally the unit you sell to advertisers. A small inbox with a per-day push and a weekly recap is not a sellable surface. It can't be. That's what makes it good.

So Vibes won't take advertising. Ever. The post-launch plan is an optional subscription tier — lossless audio, the annual recap, group vibes — and a free tier that has access to the entire core experience without any of those add-ons. The free tier is the product. The subscription is for people who want to support the project and unlock the nicer-to-have surfaces. If that funds the project, great. If not, I'll keep it small. Vibes is allowed to be small. That's not a backup plan; that's the plan.

The trick is that this only works because the product is single-purpose. A small inbox doesn't need a $15M Series A. A small inbox doesn't need a recommendation team or a growth org or 24-hour content moderation. Most of the cost structure of a modern social network exists to make the feed-plus-algorithm machine run. Refuse the machine, and the costs collapse with it.

The thing that's left.

What remains, when you take all of that out, is a small inbox. Songs come in from your specific friends. You play them. You react with one tap. You can send one back. Once a week you get a short, beautiful recap of what you sent and what you got. That's the whole product.

It's the first social-shaped app I've used in a decade where I open it, do the thing, and close it again — without scrolling, without ambient anxiety, without a notification dot somewhere in my peripheral vision asking for follow-up. The closest analog is the iPod. You opened it, picked the song, and put it away.

It will never be the largest app in your life. That's the point. There's no algorithm to feed, so it doesn't need to feed on you.

The shape of the next decade.

I don't think Vibes is going to be the only app of its kind. I think we're at the beginning of a wave of small-on-purpose products — apps that explicitly refuse the engagement playbook and survive by being indispensable to a small group rather than tolerable to a large one. Are.na is one. Posthog, in a developer-tools context, is another. Music has been weirdly slow to get one. I think we're close.

The pattern is the same across all of them. Refuse the feed. Refuse the algorithm. Refuse the growth-at-all-costs financing. Build the smallest version of the product that matters to a specific person on a specific day. Charge a fair price, or charge nothing, but never depend on advertising — advertising is what bends every social product into the same shape over time.

If you're building one of these, send me a note. I'd like to know who else is in the room.

If you want to read the longer argument — the rules and refusals laid out in plain English — the manifesto is the home for that. If you want to install the app and use it, that's over here. Send your first song to one person. See what happens.

— Scott · Brooklyn · May 2026