I recently had to explain to the New York Landmarks Preservation Commission why my office was proposing using GFRC (fiber-reinforced thin-shell concrete) to replace damaged terra cotta ornament on a historic building. The short version is that the original system for some of the ornament was badly flawed, and rather than recreate the flaw 90 years after the fact, we prefer to fix it.
Something I didn't say, and shouldn't have to say to a bunch of preservationists, is that architectural terra cotta in its day was not a precious and highly-regarded material, but rather an inexpensive way to fake carved stone. This is exactly what GFRC is today. So our proposed replacement copied both the appearance and the rationale behind the original while improving structural performance, while a painstaking recreation of the original would have been false in terms of constructive logic and would have the same problems (inadequate handrail capacity at a balustrade, terra cotta exposed to leaks from a poorly-placed gutter) as the original. I'd like to believe the A/E/C community is capable of learning from mistakes and can improve a detail when it has failed.