Anderson asked important question about GLUE.And yes, I am fully aware of Anderson's RVB theory. I've even chatted with him about it. And for your information, Anderson was also one of those who predicted that superconductivity cannot go beyond 25K before the discovery of the cuprates, a fact that Robert Laughlin never failed to bring up.
You're citing him as if he's a religious prophet. Zz.
If scientists for a 25 years searched for GLUE and found nothing about origin of HTSC, may be it is worth for NSF to fund the NONGLUE direction a little bit?
Let me give example without GLUE. Quantum mechanics for many electron system (solid and atom for example) try solve Schroedinger equation for one electron in the averaged field of other electrons and AFTER SOLVING put electrons on calculated quantum levels.
This method IS very successful.
Let us we have initially two ORBITALS for ONE electron with the collinear wavevectors in opposite direction near Fermi surface:
psi1(p1), psi2(p2), |p1-p2|=2*pF (exactly!!!)
There may be ORDER PAPAMETER interaction (matrix element, connecting those two orbitals) of some origin (Bragg reflections, phonon mode, CDW, SDW,...). Elementary quantum mechanics says there must be repellion of levels and there must be a gap at Fermi level.
For lower level at Fermi level unnormalized wave function:
If we switch on magnetic field, find optimal PSI in magnetic field and calculate current of optimal PSI we get microscopic London equation for one superconducting electron with doubled coefficient before vector potential:
Doubled becouse eA/m for both initial orbitals have the same value (p1 an p2 have opposite direction).
So we have gap, we have half flux quantization. And we don't need glue.
I don't want to debate, what is true in HTSC. I only give example, that it shouldn't to ignore nonglue (nonpairing) variants from the beginning.