The Millsian shootout has arrived
Mills, Holverstott, Good, Hogle, Makwana, and Paulus have taken it
upon themselves to conduct the Millsian vs. Spartan shootout (
https://www.electro-tech-online.com/custompdfs/2008/10/modeling101208S.pdf ).
"In a new molecular modeling paper, the energies of exact classical
solutions of molecules generated by Millsian 1.0 and those from a
modern quantum mechanics-based program, Spartan's pre-computed
database use 3-21G and 6-31G* basis sets at the Hartree-Fock level
of theory, were compared to experimental values. The Millsian
results were consistently within an average relative deviation of
/0.1% of the experimental values. In contrast, the 3-and 6-
31G* results deviated over a wide range of relative error, typically
being >30-150% with a large percentage of catastrophic failures,
depending on functional group type and basis set."
A LOT of data has been compiled in this paper. The meat of the
shootout is on pages 67 - 81, and things look very good for Millsian.