Message encoding formats, interoperability and insanity

Has someone here encountered crazy message encoding decisions in their projects, examples include: * Outer Json with base64 encoded json as payload * Outer Json with base64 encoded protobuf as payload * Outer Avro with protobuf as payload

As horrible as this sounds, all these are pretty much in the same repo/project. Some of the proponents of these decisions are hard headed, irrational and impossible to reason with - especially to make the case of the users of this system who have to deal with more complexity.

Any advice?