On legibility, and the cost of making things clear.
Every interface that explains itself is also an interface that decides what can be explained. The cost is rarely paid by its author — and that is the part worth writing down.
Legibility is the property of a thing that allows it to be read from the outside. A tax form is legible. A forest is not. Between these two poles lies most of what we build — and most of what we argue about.
Definitions
The word is James Scott’s, but the problem is older. When a state wants to tax a forest, it first has to turn the forest into something it can count. Rows, grids, species lists, board-feet. The forest has not changed. Our ability to act on it has.
Interface design inherits the problem wholesale. Every dashboard, every form, every filter is a small act of legibility — a decision that some features of the world will be representable and others will not. The decision is often invisible, and almost always consequential.
The question I want to sit with is not whether to make things legible. You have to. The question is: what does the act of making-legible remove, and who bears that removal?
The cost
In the cases I find most interesting, the cost is paid by a party who was never consulted in the design. A scheduling tool resolves ambiguity by forcing a single choice — and the person whose situation did not fit the choice now has to carry the mismatch on their own time. The interface got clearer; the world did not.
A good interface is one that hides the right things. A bad one is one that hides things its users would have wanted to know. — field note, 2024
There is a version of this story that ends in quiet nihilism — the one where every simplification is a kind of violence and the only ethical interface is no interface. I do not believe that. I do believe that we ought to keep track of what we are deciding not to see, and to whom we are charging the difference.
A working rule
The rule I try to hold: before shipping an interface, write down the three things it makes harder. Not theoretical things — specific things, for specific people. If the list is empty, the interface is probably not doing enough work. If the list is long and the people on it were not consulted, the interface is probably doing the wrong work.
Applied in earnest, the rule is unglamorous. It slows things down. It sometimes kills projects that should have been killed earlier. It occasionally saves a feature by making the team honest about who it is for. I would not call it a method — more a habit that keeps the design honest when the calendar stops doing it.
Coda
If I have a bias in this work it is toward the interfaces that admit their own shape. Tools that say, here is what I can see; here is what I cannot. Not because honesty is a virtue in itself — though it is — but because it leaves room for the user to meet the tool with the parts of themselves the tool could not model. That room is where the interesting work happens.
The rest is just plumbing.