On cruel mistakes in the calculation of multi-loop superstring amplitudes, the ambiguity of the modular integral and the integration over the module space

Widely spread cruel misconceptions and mistakes in the calculation of multi-loop superstring amplitudes are exposed. Correct calculations are given. It is shown that the cardinal mistake in the gauge fixing procedure presents ab ovo in the Verlinde papers. The mistake was reproduced in following proposals including the recent papers. The modular symmetry of the multi-loop superstring amplitudes is clarified, an incorrectness of previous conjectures being shown. It is shown that the Berezin-type integral versus boson and fermion moduli is doubt under non-split transformations mixing fermion integration variables to the boson integration ones. In particular, due to singularities in moduli of the given spin structure, the integral can be finite or divergent dependently on the integration variables employed. Hence, unlike naive expectations, the multi-loop superstring amplitude is ambiguous. Nevertheless, the ambiguity is totally resolved by the requirement to preserve local symmetries of the superstring amplitude. In the Verlinde world-sheet description it includes, among other thing, the requirement that the amplitude is independent of the gravitino field locations. In action the resolution of the ambiguity in the Verlinde scheme is achieved by going to the supercovariant gauge. As it has been argued earlier, the resulted arbitrary-loop amplitudes are finite.