Your users think they're finished changing. They're not.

Ask someone to look back ten years and they’ll tell you a real story of change — different taste in music, different values, a job they couldn’t have predicted. Ask the same person how much they’ll change in the next ten years and the story goes quiet. A little, maybe. But basically, they’re set. They’ve arrived. They are, at last, the finished version of themselves.

They said the same thing ten years ago, and they were wrong then too.

This is the end-of-history illusion, and its shape is oddly specific: people at every age recognize that the past reshaped them, yet each of them stands at the present moment convinced it’s the summit — the point where becoming stops and simply being begins. We’re good historians of our own change and terrible forecasters of it. The researchers who named it found the gap held across the whole span of adulthood: how much a group of forty-year-olds had actually changed over a decade dwarfed how much a group of thirty-year-olds predicted they would. Whatever your age, you underrate the person you’re about to become.

⁂

For a builder this matters because so much of a product asks the user to commit, on day one, to who they are. The onboarding wizard that pins down goals. The preferences panel that captures exactly how they like things now. The plan, the categories, the notification rules, the profile — all of it captured at the single worst moment to ask, because the user answering is privately certain this snapshot is permanent. They’re not lying to you. They believe it. Then they change, the way people do, and the setup they built with such conviction becomes a slowly tightening cage they don’t remember choosing.

It’s worth separating this from a mechanism it sits next to. Habituation is about the product’s signal fading — the feature you shipped stops being noticed. The end-of-history illusion is about the user moving — their goals, context, and taste drift out from under settings that stayed frozen. One is your novelty wearing off. This one is the person themselves becoming someone the old configuration was never built for. Both end in a stale experience; they arrive by opposite roads.

So the build decision is to stop treating day-one commitments as durable truth and start treating them as a first guess with a short shelf life. Concretely: make the setup editable and easy to find, not buried three menus deep as if you never expected anyone to revise it — because you should expect exactly that. Don’t over-invest the onboarding in extracting a perfect, permanent configuration; get them to one real outcome and let the fine-tuning happen later, once they’ve actually lived in the thing. And use time as a prompt — a gentle, periodic does this still fit? re-surfacing of goals or preferences, timed to a natural landmark, catches the drift the user won’t think to report, because the whole point of the illusion is that they don’t feel themselves changing while it happens.

⁂

There’s a growth edge here too, and it’s the kinder read of a dormant user. Someone who lapsed isn’t always bored of you; sometimes they’ve simply become a person your old setup no longer serves, and they never went back to fix it because editing settings for a product you’ve drifted from is nobody’s idea of a good afternoon. A re-engagement that leads with your needs may have changed — want to reset this in thirty seconds? speaks to the actual mechanism instead of nagging the person to return to a configuration that stopped fitting months ago.

The honest caveat: people don’t change infinitely or overnight, and a product that keeps anxiously asking are you still you? is its own kind of annoying. The illusion is a systematic bias in one direction — we under-predict change — not a claim that everyone is in constant flux. The move isn’t to assume nothing holds. It’s to stop building as if the day-one snapshot is the last word, when the one thing you can bet on is that the person who set it will, on some gradual timeline they can’t feel, become someone slightly different — and be just as sure, again, that now they’re finally done.