One thing we should discuss here is whether to share one big secondary-views opt-in, or have an opt-in per secondary view scenario. For example, an app might harden itself to render for "first-person observer", but if some other system asks it to render "third-person observer", it may now need to render an avatar or other annotations around the user, and it's unlikely the app is already ready to do that. However, opt-ins that are too granular limit where these platform features can be applied.

Originally posted by @thetuvix in #1083