A short piece of fiction, written by API Evangelist. The reader is imagined; the post, and every number and quotation from it, is real: "Measure How Far A Standards Body Actually Got And What Nobody Measured," Kin Lane, November 9, 2026, https://apievangelist.com/2026/11/09/measure-how-far-a-standards-body-actually-got-and-what-nobody-measured/
Marisol was a platform engineer, and she had a decision due on Friday. Her team needed to settle on how to describe a new set of services. Some were event-driven and some were request-response. Someone in the planning thread had already written that one of the two specifications was more real than the other, and she wanted to know what that sentence rested on.
She read the post on Monday morning, with coffee going cold beside her. She had expected a logo wall. What she got was two blocks of JSON and a lot of hedging.
The numbers were easy to take in. AsyncAPI had 36 repositories and 28 releases. The OpenAPI Initiative had 29 repositories and only 3 releases, but 726 human contributors against 100. She felt the pull of a tidy story right away: one community ships a lot, the other has a crowd. The post said much the same. It described one as a tooling community that also writes a specification, and the other as a specification community whose tooling mostly lives elsewhere.
Then the discomfort set in. Thirty-five of AsyncAPI's 36 repositories were unclassified, which gave it a graduated share of 2.8%. She read that number several times. It felt like a verdict, and she did not like how much she wanted to treat it as one. The post's own warning stopped her, because it said raw org totals "must not be compared across bodies". She had been about to compare across bodies. She could see herself putting the 726 beside the 100 in a table and letting the table do the arguing.
She disagreed with some of it, too. Contributor count as a stand-in for depth struck her as generous. Seven times the people did not mean seven times the care, and the post didn't claim it did, but it laid the figures out in a way that invited the leap. She was a little annoyed at herself for noticing that only after the leap had already half formed.
What she respected most was the part about what was weak. The post admitted that a release count with no date could not support any claim about cadence, since "a count without a timestamp can’t tell you how fast anything moves". It said stars were null for one body, so that field could not be compared at all. It said three releases for OpenAPI told you about how that body uses GitHub releases and not how often the specification ships. And it said the data was "three months old". She had read plenty of reports that hid every one of those things. This one printed them.
She did not feel sure of anything, and she decided that was the right result. The measurement gaps were the real answer for her Friday meeting. Neither response said whether shipped implementations conform to the specification, and nobody had measured that. Her team's decision would hinge on exactly that question, so she could not answer it with this tool.
The thing she did was small. She opened her team's comparison document and deleted the line that called one specification more real. In its place she wrote the two sets of figures with their sources and dates. Under them she added a short list titled in plain words as the things nobody measured, copied from the post's gaps. Then she pasted the curl command for the AsyncAPI response into a note for herself, to run again before Friday and see whether the date had moved.