hmm... maybe he means: "Why is it not correct to use a simple product of occupied orbitals as the many-electron wavefunction trial ground state when calculating the energy expectation value?"
In that case, one might say that the hartree product does not satisfy the Pauli exclusion principle and so that makes it "bad". A better choice might be a slater determinent of orbitals rather than just a product. This does indeed lead to "corrections" to the ground state energy eigenvalue known as "exchange terms".