I think the talk could have been accepted, honestly - it's well researched and I think it's good to listen to critical voices in the community.
I agree that the GC shouldn't be super high-priority (reference counting makes a lot more sense, especially when performant C-compatible GC'd languages like D and Go exist) but the ".." thing for example is objectively a bug. About procedures using type classes being compiled to specialized generic procedures, can this ever be a problem in practice?
