The example of implementor adoption of prefixed APIs is a distraction (and not directly relevant) to the setup of the polyfill risk point made in this section: a polyfill expecting a "fantasy" standard version of their prefixed utility which behavior is unknown/expected to be the same as the polyfill. I think Example 3 and the build-up to it could be dropped without impacting the points made later.
triblondon
commented
7 years ago
From #8