As another example of how the author failed to take into account basic standard practices, the author asserts that there is no standard for HTTP headers that specify API versions, but this baseless assertion requires ignoring the fact that media type versioning does exist and has been used for ages. The author even realizes that a solution based on the Accept header, which he prefers to invent a Accept-Version header, would fit his requirements. Ok, so why doesn't he just follow what has been standard for quite a few years and simply state the document version in the media type, which already is supported by any content-based routing scheme and is cached quite nicely?