Delocalised molecular orbitals are not superior

A recent issue of the Journal of Chemical Education has several excellent articles responding to a recent controversial article by Alexander Grushow suggesting that hybrid atomic orbitals have no real physical and chemical basis and so should not be included in the undergraduate chemistry curriculum.
There are good articles rejecting this argument. Two of the articles are written by pairs of my favourite quantum chemists Landis and Weinhold, and Hiberty and Shaik.
They are worth reading because they highlight some key issues in quantum chemistry and chemical bonding.

At the heart of the matter is whether one favours a delocalised picture (with canonical molecular orbitals) or a localised picture (hybrid atomic orbitals and valence bonds).
Neither picture is incorrect. They are complementary ways of looking at the same chemical reality. [See for example this post discussing the old MO-VB rivalry].

Both sets of authors emphasize this by stressing that a single Slater determinant is invariant under a unitary transformation. Such a transformation takes one from one molecular orbital representation to antoher. e.g., from canonical molecular orbitals to hybrid atomic orbitals. Both give the same total electron density.

This debate also highlights that chemical bonding is a quantum many-body effect and must be viewed as such. For example, the one electron orbitals are not the key quantity but the many-body wavefunction. Also, photoelectron spectroscopy does not measure orbital energies but the difference in energy between the ground state and a cation.

I thank Seth Olsen for bringing these articles to my attention.