On this page
A useful system begins with a question that is specific enough to test. Not a promise about everything it could eventually become, but a clear account of what it should do, for whom, and under which conditions.
That sounds simple. In practice, it changes where the work starts. Instead of choosing an interface and filling it with features, we begin with the decisions people need to make and the evidence those decisions require.
Start with the work
An early prototype is a way to ask a better question. It makes assumptions visible: which information matters, which actions should be easy, and which distinctions must survive as a project grows.
We test the complete path, not only its most photogenic moment. A reader should be able to discover a story, understand its context, follow a reference, and return to where they started. Each transition is part of the experience.
The point of a prototype is not to make uncertainty disappear. It is to make the important uncertainties possible to investigate.
— Zortex · Research notes
Let evidence change the structure
Some patterns deserve to become shared rules. Others turn out to be consequences of a particular title, image, or viewport. The distinction matters: a shared rule should continue to work when the content changes.
We keep the counterexamples. A long title that breaks a layout, a navigation item that cannot be reached with a keyboard, or an image that loses its subject when cropped is not an inconvenience to hide. It is a requirement the next version has to meet.
Build for continuity
A system becomes useful when other people can extend it without reconstructing the original author's intentions. Content models, interaction states, and the relationships between elements need to be explicit enough to carry forward.
The result should be more than a convincing first page. It should remain coherent as articles get longer, collections get larger, and new kinds of work appear.
An image from the supplied reference archive, used here as demonstration artwork.
