Zig has two core features which are fundamentally at odds: Result Location Semantics and Parameter Reference Optimization. Many community members would like improvements to both of these features, but they are locked against each other. Improvements to one usually cause problems when they interact with the other. Even in their current state, the language has a big footgun when the two features overlap. This leaves us in an unfortunate local maximum, which is difficult to break out of. This talk will explore the tradeoffs involved in tweaking these language features, to spur discussions about solutions.