A good empty state makes a promise, not an explanation

There’s a small, precise theory of curiosity that has quietly held up for thirty years, and once you see it you can’t design a first screen the same way again. The economist George Loewenstein proposed it in the mid-1990s, and it reframes curiosity not as a personality trait or a mood but as a mechanical event: curiosity is what you feel when you become aware of a gap between what you know and what you want to know. Not a vague hunger for information — a specific, located absence. You learn there are exactly fifty states and you can name forty-three; the seven-shaped hole starts to itch. The gap behaves almost like a mild deprivation, a small discomfort the mind wants to close, which is why a half-heard piece of gossip or a cliffhanger can hijack your attention against your will.

The part builders should tattoo somewhere is the shape of the effect. Curiosity is weak at both ends and strong in the middle. If you know essentially nothing about something, there’s no gap you can even perceive — you don’t feel the absence of knowledge you can’t frame, so a blank void generates no pull at all, only mild unease or indifference. And if you already know essentially everything, there’s no gap left, so there’s nothing to chase. The itch lives in the narrow band between: you can see the outline of the thing you don’t yet have. Enough context to feel the edge of the hole, not so much that it’s filled. Curiosity is highest when a small, well-defined piece is visibly missing.

Now stand that next to the screen every new user meets: the empty state. The first project list with nothing in it. The dashboard before any data. The inbox at zero. This is the single most under-designed surface in most products, and it’s the exact place Loewenstein’s theory is telling you the game is won or lost — because an empty state is a decision about the size of the gap you hand a stranger.

Most teams get it wrong in one of two opposite directions, and both kill the pull. The first mistake is the void: a genuinely blank screen, maybe a grey icon and the word “No items.” That’s the know-nothing end of the curve. There’s no outline of what could go here, no shape of the missing thing, so there’s no gap the user can even perceive — just emptiness, which reads as dead rather than inviting. The second mistake is the overcorrection every thoughtful team makes: the wall of explanation. A tour, five tooltips, a paragraph describing every feature, a checklist of nine things to set up. This is the know-everything end — you’ve pre-filled the gap with a flood of answers to questions the user never got curious enough to ask. You closed the loop before it ever opened. Comprehensive, dutiful, and inert.

The build move sits in the middle of that curve, and it’s a promise, not an explanation. A good empty state shows the user the outline of the filled state and hands them one concrete way to open it. Not “No projects yet,” but a single ghosted example of what a finished project looks like here, plus one button that produces a real one. Not a nine-item setup checklist, but “Add your first transaction and this chart draws itself” — a claim about a specific, near, missing result they can now see the shape of. You are deliberately manufacturing the small gap: here is the thing you don’t have yet, here is exactly how close it is, here is the one move that closes it. That’s the sweet spot of the curve rendered as a screen.

There’s a discipline in this that’s easy to lose. Every instinct in a proud team is to use the empty state to explain the product — to finally tell the new user everything the product can do. Loewenstein’s curve says that’s precisely the move that flattens the pull. The empty state’s job isn’t to answer; it’s to make the user feel one specific, closable absence and want to close it. Explanation is what you earn the right to give after they’ve opened the loop, not the thing you lead with.

The test I’d hold any empty screen to: does a stranger, five seconds in, feel a specific thing is missing that they now want — or do they feel either lost or lectured? If it’s lost, you’ve handed them a void. If it’s lectured, you’ve filled the gap before they felt it. The screen you want makes them feel the exact shape of the hole, and hands them the one tool to fill it.