The Door That Lies: What UX Actually Is

A glass door. A vertical handle on your side. You take hold of it, pull — nothing. You pull harder, glance around, then push, and the door opens easily. Somebody behind you saw the whole thing.

For a second you felt stupid. Note that feeling: it is the raw material of this entire sphere. Because the mistake wasn't yours. A handle is an invitation to pull; you accepted the invitation the door itself made. Someone hung a pull-handle on a push-door, and the person who paid for it is you.

The Norman door

Don Norman described this in 1988 in The Design of Everyday Things, and doors like it have been called Norman doors ever since. The book's central claim is uncomfortable and entirely reversible in its consequences: when a person can't work something out, the usual conclusion is that they are careless or "not technical." The honest conclusion is that the thing failed to explain itself.

This matters more than it sounds. As long as the failure belongs to the user, there is nothing to fix — you can only train them, warn them, add a sticker saying PUSH. That sticker is a confession: the object couldn't say it on its own, so we bolted an apology to it. Count the apology stickers you meet in a day; every one marks a design decision that went wrong.

The moment the failure belongs to the object, it becomes a problem someone can solve. That reassignment of blame — from the person to the thing — is what the whole profession is built on.

If people keep getting it wrong, it isn't a people problem.

Design is decisions, not pictures

The word "design" in Russian and English alike drags a lot of decoration behind it: colours, fonts, rounded corners. That association is the single most expensive misunderstanding in this field.

Norman coined the term user experience at Apple in the early 1990s precisely because the existing words were too small. "Interface" meant the screen. "Usability" meant not tripping over anything. He needed a word for everything a person goes through with a product — including the box it came in, the moment it fails, and the phone call afterwards.

So the useful definition is: design is the set of decisions about what happens, and the picture is what those decisions look like. Which of the two things a user needs comes first. What a person must have to hand before they can start. What the system does when the network dies mid-payment. Which of the two words on the button is true. Whether this feature exists at all.

Three terms get confused constantly, so it's worth pinning them down:

• UX — the decisions: what the product does, in what order, on which path, and what happens when it breaks.
• UI — the surface those decisions arrive on: layout, controls, type, colour, spacing.
• Graphic design — the visual language: identity, imagery, style.

They aren't ranked. A brilliant flow rendered in an unreadable interface fails, and a beautiful interface over an incoherent flow fails more expensively, because people trust it further before it lets them down. But they are decided in an order, and the order matters: what happens, then how it looks. Reverse it and you get the familiar catastrophe of a gorgeous screen for a task nobody has.

The experience is longer than the screen

The word "experience" is doing real work in "user experience." It stretches beyond the pixels in both directions.

Someone hears about your product from a friend, arrives with an expectation already formed, meets a signup form, waits for an email that lands in spam, finally gets in, doesn't understand what to do first, tries a thing, breaks it, hunts for help, gives up, remembers the product a month later as "confusing." Exactly one moment of that story happened on a screen you designed.

There's a useful piece of psychology here. Daniel Kahneman and Barbara Fredrickson described the peak–end rule: people don't remember an experience as an average of its moments, but roughly by its most intense point and its ending. This has a blunt practical meaning. Twenty smooth screens and one humiliating error message do not average out to "fine" — they are remembered as the error message. And the moment you close a task is remembered out of all proportion to its length.

Which is why the parts nobody wants to design — the error states, the empty screens, the cancellation flow — carry more weight per pixel than the homepage everyone argues about.

The five questions

Strip away tools and job titles, and a designer is someone who answers five questions and can say why:

1. Who is this for? Not "users" — a specific person in a specific situation, with the situation named.
2. What are they trying to do? Their goal, in their words, not the feature you want to ship.
3. What is the path? The steps, in order, including where they enter and where they leave.
4. What happens when it goes wrong? Empty, slow, offline, wrong input, wrong person, half-finished.
5. How will we know it worked? Something observable — not "it looks better now."

You can hold every one of these in your head before you open any design tool. That's deliberate. Most bad interfaces are not badly drawn; they are the accurate rendering of a question nobody answered.

Why "I like it" isn't an argument

Design decisions get made in rooms with other people, and the currency in those rooms is reasons. "I like it" is not one, and neither is "I don't."

Taste is not useless — it's a fast, compressed judgement built out of thousands of examples, which is exactly what expertise looks like from the inside. But it can't be checked, transferred, or argued with, and it fails in a specific way. Masaaki Kurosu and Kaori Kashimura showed in 1995 that people rate more attractive interfaces as easier to use, even when they demonstrably aren't — the aesthetic–usability effect, since replicated many times. Beauty doesn't just please; it quietly disables your judgement of whether the thing works.

So the working move is to translate every instinct into a claim someone else could test: not "this feels cluttered," but "there are four competing calls to action and the primary one is the smallest." The instinct pointed at something real. Your job is to say what.

In practice

Collect Norman doors. For one week, note every object or screen that made you feel stupid — a tap, a lift panel, a checkout, a parking machine. Write one line: what did it promise, and what did it actually do? This is how an eye for design failure is trained, and it costs nothing.

Ask what was decided here. Open any product you use daily and find three decisions someone made deliberately: an order, a default, an omission. Defaults are the richest — almost nobody changes them, so whoever set them designed the behaviour of most users.

Rewrite your reactions. Every time you catch yourself thinking "ugly" or "nice," force the sentence into the form: this makes it hard to X because Y. If you can't finish it, you haven't found the problem yet — only noticed it.

Design one error state today. Take any form you've built or used and write what it says when the input is wrong, when the server is down, and when the thing succeeds. That's three screens most products never wrote.

Ask the five questions before opening a tool. Answer them in a text file. If question one or two comes out vague, no amount of layout work will save the screen.

Check yourself

Close the article and answer in your own words:

1. Why is "the user made a mistake" almost always the wrong conclusion?
2. What does a PUSH sticker on a door tell you about the door?
3. What is the practical difference between UX and UI, and why does the order matter?
4. What does the peak–end rule change about which screens deserve the most attention?
5. Why is the aesthetic–usability effect a problem for a designer specifically?
6. Which of the five questions do you most often skip, and what does skipping it cost?

In short

• A Norman door is an object that invites the wrong action and blames the person who accepted the invitation.
• Reassigning the failure from the person to the thing is the founding move of the profession: only then is there something to fix.
• Design is the set of decisions about what happens; the visual is how those decisions look. Decide in that order.
• Experience is longer than the screen — it starts before the first click and ends long after the last.
• Peak–end rule: people remember the worst moment and the ending, so error states and closings carry outsized weight.
• Five questions define the work: who, what goal, what path, what when it breaks, how we'll know.
• Taste is real but unusable in argument, and beauty distorts judgements of usability. Translate instinct into a checkable claim.