Skip to content

Public Publication

A public publication surface makes information reachable by anyone, with none of the machinery a signed-in surface needs: no login, no account, and no personal data collected or shown. It publishes the same underlying information an authenticated surface draws on, read-only, to an audience that is never asked to identify itself.

What makes it distinct

Nothing on a public surface depends on who is asking. There is no sign-in step to pass, no session to hold, and no per-person view to compute — every visitor is served the same read-only material. Because no caller is identified, nothing personal is collected or displayed; what is published is limited to information that is public by design.

What kinds of material are published

A public surface federates more than one kind of material into a single place: announcements, news, products and services, a regelcatalogus of rules and the services they implement, and a procesbibliotheek of process definitions. Each kind keeps its own shape, but all of it is reachable from one search.

Material that is not finished is published with that fact attached rather than withheld. A process definition carries its lifecycle status as a visible label wherever it appears — in a listing, in a search result, and on its own page — so a reader can tell an example or a work in progress from a settled one without needing to open it.

Searching and browsing across it

A single search spans every kind of published material at once, built from one federated index so that a listing and a search result can never drift apart — an item that appears in a list is resolved through the same index a search query matches against. A search can be narrowed by facets such as the kind of item, its source, or its intended audience, and each item has its own stable, directly linkable detail page.

Prerendered, then revalidated

Each published page is built ahead of time with its content already embedded, so the first frame a visitor sees carries real material instead of a loading placeholder that grows in and shifts the page underneath them.

That embedded copy is a snapshot taken when the site was last built, and the information it describes moves independently of deploys — so it is treated as a starting point rather than as the answer. A page paints the embedded copy immediately and asks the platform for the current version underneath it. The fetched answer replaces what is on screen only when it actually differs: an identical response changes nothing and causes no re-render, which is what preserves the absence of layout shift that prerendering exists for. When the request fails, whatever is already on screen stays.

The behaviour exists because the alternative was tried. Trusting the embedded copy whenever it was present published a catalogue frozen at the last build: two services that had already been retired were still listed, and two counters stood at figures the authenticated working environment, reading the same API, disagreed with. Every prerendered content page revalidates now, not only the one where that was noticed.

Served separately from the authenticated surface

A public surface is its own deployment, reachable at its own address, distinct from the signed-in working environment it draws its information from. It carries its own security headers restricting where its scripts and content may load from, and it reaches the platform only through the subset of endpoints deliberately left open to unauthenticated callers — see API Design — Public versus authenticated surface. It also accepts a small number of writes on that same public surface — such as an attachment or a message — subject to the stricter rate limit and proof-of-work check described in Security & Compliance.

Accessibility

A public government site commits to WCAG 2.1 level AA conformance. In practice that includes a skip link to the main content visible on keyboard focus, a visible focus indicator on every interactive element, a label on every form field even where it is hidden visually, a minimum text contrast of 4.5:1, and a landmark structure — header, navigation, main content, footer — with a breadcrumb trail. Known gaps against that target, and how to report a new one, are published on the site itself.

Concepts, divided by the side they sit on

A published rule names the concepts its service works with, and those concepts are divided rather than listed flat: the values the rules need in order to decide, and the values the rules determine. The division comes from the knowledge graph — it was always there and was simply being dropped on the way out — so a reader can see what a service asks for separately from what it answers, without opening the decision model.

A concept the graph leaves undirected is still shown, in a group of its own rather than filtered away.

Open data

Everything published is also available as open data: every item is machine-readable through the same anonymous API the site itself calls, with no key and no account required, and is free to reuse as public government information without copyright restriction.

That commitment constrains how the published shape may change. When concepts gained their direction, the divided list was added alongside the flat one rather than replacing it, because outside consumers read the flat array and a published field is not ours alone to reshape.