Recently, there has been a question by Voldemort concerning different resonance structures of $\ce{NCO-}$, requesting an explanation why one resonance structure would be more preferred than another.

There has been a little discussion on that between ron and Martin - マーチン that took my interest:

ron: "[...] structure #1 ($\ce{N#C-O-}$) is more stable (preferred) than structure #2 ($\ce{^{-}N=C=O}$)."

Martin - マーチン: When talking about resonance structures, I think stable is the wrong word. I agree with your analysis in terms of contribution. Therefore #1 should have the highest contribution of all.

Now while I agree with ron that structure #1 is the most preferred resonance structure (referring to electronegativities), I regret with him concerning his attitude towards the "stability" as Martin does.

But I mainly do this, because I used NBO6 with the Natural Resonance Theory (NRT) keyword on the Natural Bond Orbitals (NBO) produced of an MP2/6-311+G(2d) optimized wavefunction to find the contributions of all relevant resonance structures - $66~\%~(\ce{N#C-O-})$, $30~\%~(\ce{^{-}N=C=O})$ and $3~\%~(\ce{^{2-}N-C#O+})$ - and not because I have a superior understanding on all underlying theories.

For me, the resonance stuff lookes a lot like MCSCF treatment of some VB wavefunction, which means that the complete molecule is described best from it's contributing resonance structures.

Now on a more complete comparison, what are the main similarities and differences between both theories and as they don't exist but are rather part of an explanation, can resonance structures indeed be described as "stable"?