One of the primary goals of Chandra, XMM, and Spitzer was to unveil the population of obscured AGN at high redshift (z>2). These objects, known as Type-2 QSOs, are postulated to have the same bolometric output as unobscured Type-1 QSOs, but are attenuated by dust and gas. But Type-2 QSOs at z>1 with QSO-like luminosities are conspicuously missing, even from the largest multiwavelength surveys. This is particularly puzzling, given decades of work on Type-1 demographics which establish that the QSO luminosity density peaks at z~2.5. Either the Type-2s and Type-1s are distinct populations, requiring a major re-thinking of current models, or the luminous high-z Type-2s have somehow been missed. Here we present the spectra of ``extremely red quasars'', which have spectra from SDSS/BOSS but are photometrically selected from a new WISE 22um (W4) selection technique. We argue that this is a way forward in identifying and investigate the elusive high-z Type 2 QSO population.
high-$z$ Type-2 QSO population.