An "exotic" idea proposed by Viktor Ambartsumian was that new galaxies are formed through the ejection from older active galaxies. Galaxies beget galaxies, instead of the standard scenario in which galaxies stem from the evolution of the seeds derived from fluctuations in the initial density field. This idea is in some way contained in the speculative proposal that some or all QSOs might be objects ejected by nearby galaxies, and that their redshift is not cosmological(Arp, G./M. Burbidge and others). I will discuss some of the arguments for and against this scenario; in particular, I shall talk about the existence of real physical connections in apparently discordant QSO-galaxy redshift associations. On the one hand, there are many statistical correlations of high-redshift QSOs and nearby galaxies that cannot yet be explained in terms of gravitational lensing, biases, or selection effects; and some particular configurations have very low probabilities of being a projection of background objects. Our understanding of QSOs in general is also far from complete. On the other hand, some cases which were claimed to be anomalous in the past have found an explanation in standard terms. As an example, I will show some cases of our own research into this type: statistics of ULXs around nearby galaxies, and the Flesch & Hardcastle candidate QSOs catalog analysis. My own conclusion is neutral.