I wonder if the concise method body proposal should not be extended to add parameters types/return type inference in case of overriding methods.
In LVTI, we drew a bright line between inference in implementations and inference in declarations, for a good reason. Clever observers commented that "but you could safely use `var` to infer the return type of a private method, without inviting API instability", but we declined on the basis that the incremental complexity of the rules (having a coupling between accessibility and inference) was not justified by the incremental expressiveness. The current rules restrict inference to local variables, which is nice and clean and orthogonal. I would make the same argument for not introducing a similar coupling here.
