Assume we are dealing with a Lagrangian $\mathcal{L}$ for matter field $\psi$ which has a global $G$-symmetry and it's possible to promote this global $G$-symmetry to a local symmetry after the usual Yang Mills recipe to local symmetry, ie replacing the usual derivatives by the covariant derivatives containing the gauge fields which garantee the local invariance.
An classical example: Consider free Lagrange function $\mathcal{L}:= (\partial_\mu \psi)(\partial^\mu \psi^*) + k^2|\psi|^2 $ with global $U(1)$-symmetry $$\psi \rightarrow e^{-i\theta}\psi$$ on matter field $\psi$ for constant $\theta$. Then we have a well known cooking recipe how to "promote" this global symmetry to a local one: $$\psi \rightarrow e^{-i\theta(x)}\psi $$ ie $\theta(x)$ is space dependent. The usual differential $\partial$ is going to be replaced by the covariant one $D:=\partial_\mu - iqA_\mu$ with respect the gauge bosons $ A_\mu$ making the resulting Lagrangian local gauge invariant.
Question: I'm looking for the original source where the historical motivation/background for posing this "promotion from global to local symmetry" is discussed. As @Andrew in his answer here explained, the original motivation turned out to be wrong, even if finally it initiated the construction of a correct theory. And my concern is just where I can look up the discussion on this original motivation? (even if as Andrew said that it finally turned out to be wrong, I would like from didactical point of view take a closer look at it as a "start of my journey through this topic".)
So far I understood Andrew's historical short recap in the linked discussion correctly, Yang and Mills wanted originally to find a local and Lorentz invariant theory. And then they drawed a "conclusion", that this theory cannot have global symmetry, because "an observer is only capable of transforming the fields locally". But exactly this phrase I not understand literally.
What does it mean that "an observer is only capable of transforming the fields locally?" A guess: To turn it in formal physical language, does this phrase say that "any observable couldn't be invariant with respect to global symmetries, only the local ones"?
