There is a question and corresponding downvoting of my answer, so I decided to ask this question.
There is my answer on it:
"...The most theories of free fields are invariant under global gauge transformations. It can be interpreted as instantaneous "rotation" of all space-time relatively gauge space, which contradicts the causality (of gauge transformations) in relativity case, because the speed of interactions is limited by $c$ according to it. So we introduce local gauge transformations, where "rotation" depends on space-time point. The gauge states is indistinguishable in a case of free field. But when our global gauge invariance field interacts with other fields, degeneracy is lifted, and this leads to instant changes in interaction. Thus we satisfy causality principle for gauge transformations.
Also I can add some example. Let's have global gauge isotopic invariance of QFT. So we identify proton and neutron as the state with some value of isotopic spin, and then if we choose that what to call a proton at one space-time point, we also must choose what to analogically call a proton at other points, so it little contradict the principle of local field theory. If the nucleon interacts with EM field, we may instantly change it state by using global gauge transformations, which also instantly changes the interaction between nucleon and EM field in all points of the space-time. It leads to non-local theory and contradict the causality.
Corresponding arguments were used by Yang and Mills in 1954...."
Where did I make the mistake (I really do not see the error)?