It isn't a new stance either. For years people have been posting their ideas for Go generics to golang-nuts, and the questions inevitably lead to asking how they'd implement it, how it would interact with interfaces, and so on. Most of the time the proposer hadn't thought it all the way through in those regards. The point of publishing these proposals is to help speed up that process: now the first response can be to point at these and ask for a similar level of thoroughness.