A short piece of fiction, written by API Evangelist. The reader is imagined; the post, and every number and quotation from it, is real: "Europe's next generation of geodata APIs speaks OpenAPI," Kin Lane, November 11, 2026, https://openlint.org/blog/2026/11/11/europes-geodata-speaks-openapi/
Marit coordinates spatial data for a regional authority, and she read the post on her lunch break with a cold coffee and a half-finished map layer open in another window. Her team publishes through WMS and WFS. They have for years, and nobody has ever complained.
The numbers came first. About 29,600 reported High-Value Datasets, and about 19,100 of them distributed as OGC Web Map Service. She felt a small, unreasonable pride at that, as if her team's quiet habit had turned out to be the whole continent's habit. Then she reached the sentence that took the pride away: "OpenAPI isn’t in the top forty formats."
She read on with her arms crossed. The author sampled 893 endpoints and called each one. "Seven declared themselves as OpenAPI." Six of those seven were OGC API – Features services. Marit's first reaction was to argue. Seven out of 893 is not a trend. It is a rounding error with a blog post attached. Her own services would have been in that sample as plain WMS, answering correctly and describing nothing, and she did not see why that counted against them. A service that works is a service that works.
But the paragraph about the validators stayed with her. The post says the classic conformance tools tell you whether a service implements a standard. They don't tell you whether a public body's APIs are consistent with each other, or follow national design rules, or can be used without a phone call. She thought about the phone calls. There had been three this autumn, all from developers asking which layer name meant what, and she had answered each one from memory. Her team passed every conformance check they ran. None of those checks would have noticed that she was the documentation.
She did not enjoy that. It was a little like being told your house is up to code and also that nobody can find the front door.
She still wasn't persuaded that her team should migrate anything this year. The post doesn't claim they must, and she noticed that it doesn't. It describes where the data says things are heading and says what linting could do once agencies get there. She could live with that argument, even if she resented how reasonable it was.
What she could not shake was the closing request. The author wrote that they would like to hear "how you check what you publish." Marit realised she had no tidy answer. They ran the validator before each release and then trusted each other.
She closed the map layer without finishing it. Then she opened the project's Discussions page, started a new thread, and wrote three plain sentences about her team's current routine: which validator they run, what it catches, and the phone calls it doesn't prevent. She read it once, left out the part where she was annoyed, and posted it.
Then she put a reminder in her calendar for Thursday, in case the office hours turned out to be worth an hour.