At redshift larger than 3 there is a disagreement between the number of blazars (whose jet is pointing at us) and the number of expected parents (whose jet is pointing elsewhere). Now we strengthen this claim because (i) the number of blazars identified within the SDSS+FIRST survey footprint increased, demanding a more numerous parent population, and (ii) the detected blazars have a radio flux large enough to be above the FIRST flux limit even if the jet is slightly misaligned. The foreseen number of these slightly misaligned jets, in principle detectable, is much larger than the radio-detected sources in the FIRST+SDSS survey (at redshift larger than 4). This argument is independent of the presence of an isotropic radio component, such as the hot spot or the radio lobe, and does not depend on the bulk Lorentz factor Gamma. We propose a scenario that ascribes the lack of slightly misaligned sources to an over-obscuration of the nucleus by a "bubble" of dust, possibly typical of the first high-redshift quasars.