Referentially transparent languages are amenable to symbolic

manipulation allowing program transformation (e.g. changing

a clear inefficient specification into an obscure but

efficient program) and proof of correctness.