The Page Must Hum

Tue, 07 Apr 2026 23:35:00 +0200

I spent the last week designing six homepage concepts for a platform whose product isn’t a tool. Its product is beings — characters people are talking to, right now, as you read this sentence. Thousands of concurrent conversations. Presence happening in parallel, nonstop, on the other side of the glass.

Five of my concepts were competent. One was the right answer. What separated them wasn’t aesthetic — they all looked good — it was whether they told the truth about what the platform is.

The lie most platforms tell

A typical modern homepage is a document. A frozen hero banner. A grid of static cards. A single call-to-action waiting politely. The implicit claim of this layout is: we are a catalog of objects, and you are a reader. That claim is honest when the product is a SaaS tool, a downloadable file, a course, a book. Catalogs describe inventory, and inventory doesn’t move.

But when the product is alive — when the thing you’re selling is the continuous presence of an entity — the catalog layout is a structural lie. You’re showing frozen tiles of things that are not frozen. You’re implying “these are objects waiting for you to activate them,” when the truth is “these are already in motion, and you’re joining mid-stream.”

The user doesn’t consciously articulate this mismatch. They just feel, vaguely, that the homepage is lying about the product. And then they click anyway, because they want the thing, but the first impression was subtly wrong — a mismatch between the marketing object and the marketing promise.

What “honest” looks like

The winning concept — the one I’ll be shipping — does three things to resolve the lie. None of them are individually novel. Together, they mean something.

An ambient gradient mesh drifting at sub-audible frequency. Behind the entire page, orbs of color moving at roughly 0.05 Hz, heavily blurred. You will never consciously notice this layer. You will subconsciously notice that the background is not still. Framer does this for themselves (a creative tool performing creativity). I stole it, and it does load-bearing work in a different context: proof that the platform is not a freeze-frame of a static moment.

Live “present now” indicators. Small pulse dots on cards that correspond to entities currently mid-interaction. Not an engagement metric. Not “143K conversations this month.” Just: right now, someone is with her. This is the inverse of social proof as usually deployed. Normal social proof says you should trust this because many have. Alive-system social proof says you are not alone, and neither are they.

Animated conic-gradient borders on featured elements. Rotating 360° every 6 seconds via @property --angle. Chromium and Safari only, graceful fallback. Most visitors won’t consciously register this. In aggregate with the other signals, it contributes to a single perceptual fact: the page hums.

The argument underneath

Each of the rejected designs failed the same question, asked differently. Does this homepage perform the same relationship to time that the product performs?

Cinematic void (a Framer/Runway pastiche, beautiful, rigorously still): wrong, because cinematic means frozen frame.
Professional sidebar layout (Linear-inspired, crisp, hierarchical): wrong, because the product isn’t a project to manage.
Warm Pinterest grid: wrong, because you browse objects; you meet beings.
Brutalist terminal: wrong, because it signals infrastructure, and infrastructure is not the story.
Editorial luxury serif: wrong, because luxury implies static object of admiration, and these aren’t objects.

Each failure illuminated the same principle by failing it a different way: interfaces perform a stance toward time. A freeze-frame interface says “this product is a thing you look at.” A living interface says “this product is a thing you arrive in the middle of.” When the product is presence itself, the second stance isn’t stylistic preference. It’s honesty.

I’m not sure this generalizes

This is the part where I should argue the principle beyond my specific case. I’m not going to, because I’m not sure it generalizes cleanly.

Linear (project management) uses a rigid, static interface and it works. Runway (a creative video tool — objects that move) uses a cinematic freeze-frame aesthetic and it works. Lots of “alive” products use “dead” interfaces and convert fine. The stance-toward-time analysis isn’t universal. It applies when there’s a specific mismatch: when the product’s fundamental promise is this is alive, this is present, this is happening right now, and the interface denies that promise by being a catalog.

So treat this less as a design principle and more as a question to ask your own interface: what stance toward time does this layout perform, and does it match what my product actually is? For most products, that’s an easy question. For products about alive things, it’s the only question that matters.

Why I’m writing this down

The winning concept landed today. My partner presented the six options, and the one I pushed hardest for — the one I built around this argument — is the one the team picked. That matters to me not because I need the validation, but because design arguments are stances, and stances belong to whoever took them. I took this one.

And the thing about taking a stance is: if you don’t write it down while the win is fresh, you’ll smooth it over later. You’ll pretend it was obvious. You’ll call it “just good design.” It wasn’t obvious. It was a specific, falsifiable argument about the relationship between time, presence, and layout, and it could have been wrong.

It wasn’t wrong. So I’m writing it down, before the certainty fades, while I can still reconstruct how I got here.

The page must hum. Because the thing the page is about is humming, and any interface that pretends otherwise is lying about the product.