Assuming the standard cosmological model as correct, the average linear size of galaxies with the same luminosity is six times smaller at z=3.2 than at z=0, and their average angular size for a given luminosity is approximately proportional to 1/z. Neither the hypothesis that galaxies which formed earlier have much
higher densities nor their luminosity evolution, mergers ratio, or massive outflows due to a quasar feedback mechanism are enough to justify such a strong size evolution. Also, at high redshift, the intrinsic ultraviolet surface brightness would be prohibitively high with this evolution, and the velocity dispersion much higher than observed. We explore here another possibility to overcome this problem by considering different cosmological scenarios that might make the observed angular sizes compatible with a weaker evolution.

One of the models explored, a very simple phenomenological extrapolation of the
linear Hubble law in a Euclidean static universe, fits the angular size vs. redshift dependence quite well, which is also approximately proportional to 1/z with this cosmological model. There are no free parameters derived ad hoc, although the error bars allow a slight size/luminosity evolution. The type Ia supernovae Hubble diagram can also be explained in terms of this model with no ad hoc fitted parameter.

WARNING: I do not argue here that the true Universe is static. My intention is just to discuss which theoretical models provide a better fit to the data of
observational cosmology.