Cool neutral gas provides the raw material for all star formation in the Universe, and yet, from a survey of the hosts of high redshift radio galaxies and quasars, we find a complete dearth of atomic (Hi 21-cm) and molecular (OH, CO, HCO+ & HCN) absorption at redshifts z ≳ 3 (Curran et al. 2008). Upon a thorough analysis of the optical photometry, we find that all of our targets have ionising (λ ≤ 912 Å) ultra-violet continuum luminosities of LUV ≳ 1023 W Hz−1. We therefore attribute this deficit to the traditional optical selection of targets biasing surveys towards the most ultra-violet luminous objects, where the intense radiation excites the neutral gas to the point where it cannot engage in star formation (Curran & Whiting 2010). However, this hypothesis does not explain why there is a critical luminosity, rather than a continuum where the detections gradually become fewer and fewer as the harshness of the radiation increases. We show that by placing a quasar within a galaxy of gas there is always a finite ultra-violet luminosity above which all of the gas is ionised. This demonstrates that these galaxies are probably devoid of star-forming material rather than this being at abundances below the sensitivity limits of current radio telescopes.