That's really a question about the language design of variadic generics, not the ABI. If you can think of a situation where uniquing variadic generics by type identity would be useful, it's worth bringing that up in the variadic generics design thread. Purely from an ABI perspective, it would be a different feature in the source language and could use a different representation at runtime, so the decisions we make about list-variadics wouldn't have to limit set-variadics.